Home

Getting Started Manual - Applied Biostatistics II

1. i File Edit View Data Utilities Graph Analyze Advanced Quick Access Addons Window Help xX workspae 1 x A 29 SYSTAT Output HE USE imes New T ES a FE USE Ourworld Z 21 CSTAT GDP_ V File mantis S YSTAT SYSTAT13 SYSTAT_13 Data Ourworld syz Number of Variables 39 Number of Cases 57 SYSTAT Rectangular file mantis SYSTATISYSTAT13 SYSTAT_13 Data Ourworld syz Created data file Thu Nov 27 11 43 31 2008 containing variables COUNTRYS POP_1983 POP_1986 POP_1990 POP_2020 URBAN BIRTH_RT DEATH_82 DEATH_RT BABYMT82 BABYMORT GNP_82 GNP_86 GDP_CAP_ LOG_GDP EDUC_84 HEALTH84 HEALTH MIL_84 MIL GOVERN GOVS GNPS B_TO_D82 URBANS LIFEEXPM LITERACY GROUPS GDP LON MCDONALD Minimum 127 742 11 600 Maximum 19353 214 100 000 Arithmetic Mean 5424 386 73 563 3 i 5 Standard Deviation 5979 374 29 765 27 986 Qutpuk ss Examples A gt CLASSIC OFF pa gt USE Ourworld syz A gt CSTAT GDP_CAP LITERACY POP_1986 Done 24 Chapter 2 Data editor The Data editor displays your data in a row by column format SYSTAT c Program Files SYSTAT 13 data Newworld syz 3 Ea We El gt z H M mw Ar BE 3 H 2 ye d i qx gt 44 SYSTAT Output lt Startpage Untitied syo_ Newworldsyz9 gt gt SOS E USE HE USE POP_1983 POP_1986 POP_1990 POP_2020 URBAN BIRTH_82 BIRTH_RT DEATH_82 DEATH_RT
2. ee ee ee ee ee rr er eo err ee rr ree ee Log likelihood of Constant only Model Log likelihood of Full Model Chi square value dE p value R square Measu McFadden s Rho squared Cox and Snell R square Naglekerke s R square Evaluation Vector CONSTANT LDOSE res 1 000 VALUE 032 21 0 562 Oy 376 i Standard Error A ee A ee E 4 gt gt gt gt 95 95 Confidence Interval Lower Confidence Interval Lower Upper Upper 309 Quantile Table Probability 999 1995 490 STO 230 900 750 667 500 333 20 100 050 025 010 005 001 Oo0o0O0O0O0Oo0O000O00O0O0O0O0Oo0OoOo OORFRNNWA O1O O T a2 5 2 Ye des 5 O Case frequencies Variables 907 900 95 Upper 788 Bounds Lower s518 determined by value of variable COUNT The categorical values encountered during processing are RESPONSE 2 levels Dependent Variable Analysis is Weighted by Sum of Weights Input Records Records for Analysis Sample Split Category Count 0 RESPONSE 1 REFERENCE l I Levels 1 000 0 000 RESPONSE COUNT 25 000 9 9 15 000 10 000 l l Sa ee SS i l l Log Likelihood Iteration History Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood Information Crite
3. i I I no chicken no pasta yes chicken yes pasta CALCIUM 3 380 0 305 1 423 0039 af Mean Squares Ratio 1 1 807 1 432 1 39 298 31 136 1 7 908 6 266 18 T262 Standard Error N 0 397 9 000 LS Mean 859 l a FE i l 002 000 SYSTAT Basics interpreting main effects The main effect for DIET does not appear to be significant p value 0 247 but let us look at a scatterplot and see if that tells us anything more 94 Chapter 3 E From the menus choose Graph Scatterplot m Select CALCIUM as the Y variable and DIET as the grouping variable SYSTAT will automatically use the case number as the X variable Select Overlay multiple graphs into a single frame Click the Symbol and Label tab click Select symbol select a circle for the first symbol and a triangle for the second m Check Display case labels in the Case labels group and select FOOD as the case label variable m Click the Fill tab click Select fill in the Fill pattern group and select a solid fill for both the first and second fill patterns m Click OK 50 40 e pasta O pasta 30 A pasta O A pasta q 020 e chicken 10 ha obiakieltiaikpas ta DIET A pasta A pasta chicken no A chicken A chickes chicken e chicken A chiakehicken chicken A yes 0 5 10 15 20 25 Index of Case The scatterplot shows that all of the dinners with a square root value for CALCIUM over 4 are pasta dinners
4. eee 268 Mercury Levels in Freshwater Fish 268 GOES ee nio Aa ADA a eG aA 271 Bayesian Estimation of Gene Frequency Dil Manufacturing 23 o ch cue wes he ae oe AAA Re ee aa Se S 276 Quality CONTO 2 2 00 k 44 Baw eG wht SRM Se 276 Medical RESCATA A EO Se 278 Chicas oa Moye a ee ve Wand bare ve tie epee a 278 PSYCHOIOCY 2 net ARASH a Bat SSeS at a 291 Day Care Effects on Child Development 291 vili 9 Data Files Analysis of Fear Symptoms of U S Soldiers using Item Response Theory or eeraa i o DOCIOIO OY a a td ii ee World Population Characteristics 0 e E he he Bd Instructional Methods 000000048 TOXICOlO Ss paa Gide od POR Res BS ES Bw E Concentration of nicotine sulfate required to kill 50 of a group of common fruit flies Data REICICnCeS cu ip a aa AA ao Anthropology Data Sources 0048 Astronomy Data Source 00 000084 Biology Data Source Chemistry Data Sources Engineering Reference 09 dw a da a A e Environmental Science Sources Genetics Data Sources ee a Manufacturing Data Sources 0 0048 Medicine Data Sources 2 2 2 a ee Medical Research Data Reference Psychology Data Reference Sociology Data Reference Statistics Data
5. m Data Variable Properties Add Empty Rows Insert Variable s Delete Variable s Insert Case s Delete Case s Find Variable Go To First Selected Case in Column Previous Selected Case in Column Next Selected Case in Column Last Selected Case in Column and Invert Case Selection Graph Editing Classic mode Copy Graph Graph View Page View Text Tool Font Drawing Attributes Pointer Tool Draw Line Draw Polyline Draw Arrow 218 Chapter 7 Draw Rectangle Draw Circle Draw Ellipse Text Tool Pan Zoom In Zoom Out Zoom Selection Reset Graph Realign Frames Graph Tooltips Highlight Point Region Selection Lasso Selection Show Selection and Invert Case Selection Graph Editing DirectX mode Copy Graph Graph View Page View Format Painter Pointer Tool Pan Zoom In Zoom Out Reset Graph Realign Frames Graph Tooltips Highlight Point Region Selection Lasso Selection Show Selection and Invert Case Selection One or more of these buttons can be deleted and new ones can be added as described previously but the toolbars themselves cannot be deleted They can however be closed The Format Bar Data and Graph Editing toolbars can be closed by right clicking on the corresponding tabs and unchecking Show Toolbar repeat the same steps to display them again The Data Edit Bar can be closed by right clicking on the Data editor and unchecking Show Data Edit Bar repeat the same steps to display it again Othe
6. Edit Options Sort variable lists in dialogs by Fandom number generation File order Mersenne T wister algorithm Output Alphabetical order Wichmanr Hill algorithm Output Scheme Bubble Help Default command file format Display Bubble Help 2 Unicode Time delay SEC ANSI Command butter Graph File Locations Autocomplete commands Number of commands to keep ls f Collen soma ld Include commands submitted from Link data files to output file command prompt Save command log in output file _ files Commandspace and clipboard dialogs Perform substitutions specified by TOKEN commands Show Cancel dialog to terminate lengthy processing Prompt to save all documents while quitting application Sort variable lists in dialogs by You can sort source variable lists in dialog boxes by file order or alphabetical order For data files with a large number of variables it is often easier to find variables in source lists if the variables are sorted alphabetically If variables are grouped together in the file for a specific reason it may be easier to select related groups of variables if the variables are sorted in file order Random number generation SY STAT provides two algorithms for generating random numbers m Mersenne Twister This is believed to have a far longer period and far higher order of equidistribution than other random number generators It is the recommended option especially for Monte Carlo
7. urban city Furthermore for any command involving filenames such as USE and SAVE filenames and file paths containing spaces require quotation marks around them Braces If an option takes more than one value then the option values should be enclosed in braces For example CSTATISTICS urban babymort MEAN SEM MEDIAN ROWS row 1 row 2 row 3 Specifying matrices Some commands and options accept matrices as their arguments Enclose the elements in brackets and indicate the end of rows by semicolons except the last row Each row may be written on a separate line The following are two possibilities AMATREX tly AO Oo EL 0 do 7 El Ok O AMATRIX 1 0 O 1 1 1 OOF 1 0 oO 1 SYSTAT functions A typical SYSTAT function has the syntax FUN par1 par2 where parl par2 are the parameters of the function FUN When the number of parameters is more than one the parameters have to be separated by commas a space cannot be used as a delimiter The parameters are optional for many functions default values will be used in which case the function has to be written as FUN For instance ZRN will generate random numbers from the standard normal distribution Unit of measurement Certain commands and options related to graphs allow you to specify the unit of measurement The available units of measurement are inches centimeters and points that can be indicated using the keywords IN CM and
8. 1000000 0105858 0430849 0549881 1000000 0 030595 0 108538 0 103966 0 190436 1 000000 THERAPY Prior Therapy Status 0 011063 0 037053 0 043759 0 007643 10495157 1 000000 trtmntxtherapy 0 017339 0 041143 00714157 0 049936 0 507876 0 943499 1 000000 Done CGAP HTM WR NUM 190 Chapter 6 Each data file opened during a session creates a new tree folder in the Output Organizer Within each tree folder each procedure generates entries one for text results and one for every graph If there 1s no data file open the entry 1s created under the last tree folder Clicking an entry scrolls the Output editor to the corresponding output Double clicking on a graph entry opens the corresponding graph in the Graph tab When the Graph tab is active clicking a graph entry dynamically changes the graph that is displayed in the Graph tab Y ou can close folder icons by clicking the to the immediate left Clicking a opens the corresponding folder In case of the SYSTAT output tree you can also close open it by selecting Collapse Tree Expand Tree from the Edit menu However opening and closing folders in the Organizer does not affect the Output editor A second use of the Output Organizer is to reorganize the results in the Output editor Cutting copying or pasting in the Organizer yields parallel results in the Output editor For example clicking an icon in the Output Organizer selects that entry Clicking a folder
9. 2000 Design of experiments statistical principles of research design and analysis New York Duxbury Thomson Learning Laner S Morris P and Oldfild R C 1957 A random pattern screen Quarterly Journal of Experimental Psychology 9 105 108 Lange T R Royals H E and Connor L L 1993 Transactions of the American Fisheries Society Lawley D N and Maxwell A E 1971 Factor analysis as a statistical method Mad New York American Elsevier Publishing Company Lee J 1992 Relationships Between Properties of Pulp Fibre and Paper unpublished doctoral thesis University of Toronto Faculty of Forestry Lee P M 1989 Bayesian statistics An introduction London Edward Arnold p 179 Lindberg W Persson J A and Wold S 1983 Partial least squares method for spectrofluorimetric analysis of mixtures of humic acid and ligninsulfonate Analytical Chemistry 55 643 648 Long L H ed 1971 The world almanac New York Doubleday Longley J 1967 An appraisal of least squares program for the electronic computer from the point of view of the user manual Journal of American Statistical Association 62 819 841 Lubischew A A 1962 On the use of discriminant functions in taxonomy Biometrics 18 455 477 MacGregor G A Markandu N D Roulston J E and Jones J C 1979 Essential hypertension Effect of an oral inhibitor of angiotensin converting enzyme British Medical Journal
10. DIETS Mean Difference Lower Limit Upper Limit E df A a a 4 gt gt gt gt PROTEIN no 5 287 1 916 8 658 3 228 254385 i yes CALCIUM no 2 031 6 322 10 384 0 501 24 520 i yes Variable p Value A a a 4 PROTEIN i 0 003 CALCIUM i 0 621 Pooled Variance i 95 00 Confidence Interval Variable DIETS Mean Difference Lower Limit Upper Limit E df ts ec ee ets E ee ne ee ee ee ne er rr rr rr er rr er rere eee PROTEIN no 5 287 1 922 8653 ho A 26 000 i yes CALCIUM no 2 031 60 538 10 600 0 487 26 000 i yes Variable p Value ca a ce ee ame Me Cs 4 PROTEIN i 0 003 CALCIUM 0 630 Two sample t test Two sample t test 50 40 Zz 30 Hi 2 O 9 14 lt x oa O 20 10 DIET no X yes The t test procedure produces two density plots as Quick Graphs On the far left and right sides of the density plot for each test variable are box plots for each category of the grouping variable The box plot on the left side of each graph is for the DIETS no group and the box plot on the right side of each graph is for the DIET yes group 81 SYSTAT Basics The middle portion of each graph shows the actual distribution of data points with a normal curve for comparison The results in the box plots for PROTEIN are desirable The median horizontal line in each box is in the center of the box and the lengths of the boxes are similar A
11. Data Files HLEN2 Head length of the second son HBREAD2 Head breadth of the second son HEADDIMe Flury and Riedwyl 1988 These data are measurements of two hundred 20 year old male Swiss army personnel on the following characteristics MFB Minimal frontal breadth BAM Breadth of angulus mandibulae TFH True facial height LGAN Length from glabella to apex nasi LTN Length from tragion to nasion LTG Length from tragion to gnathion HEART DASL 2005 An experiment was conducted by students at The Ohio State University in the fall of 1993 to explore the relationship between a person s heart rate and the frequency at which that person stepped up and down on steps of various heights The response variable heart rate was measured in beats per minute There were two different step heights 5 75 inches coded as 0 and 11 5 inches coded as 1 There were three rates of stepping 14 steps min coded as 0 21 steps min coded as 1 and 28 steps min coded as 2 This resulted in six possible height frequency combinations Each subject performed the activity for three minutes Subjects were kept on pace by the beat of an electric metronome One experimenter counted the subject s pulse for 20 seconds before and after each trial The subject always rested between trials until her or his heart rate returned to close to the beginning rate Another experimenter kept track of the time spent stepping Each subject was always measured and timed by the sa
12. Name of Statistic lt statistic gt N NU Minimum MI Maximum MA Sum SU Median MD Arithmetic Mean ME Standard Deviation SD Variance VA Shapiro Wilk Statistic WS Shapiro Wilk p value WP Cleveland Percentile for percentile PTILE1 Following this naming convention the environment variable name for the 66th Cleveland percentile for the 3rd BY group for a variable VAR 32 would be SPTILE1 66 3 VAR 32 Example Computing Mean Using Environmental Variables Sometimes the data that we need to analyze is not available in a single file but scattered across different files say in different locations One approach to analyze such data is to append all the files and do the analysis In this example we illustrate an alternative approach whereby basic statistics are computed for the individual data files and the final statistic is computed using environmental variables We generate a random sample of size 200 000 from the normal distribution split 1t into two sub samples and compute the mean of the entire sample using the environment variables of the sub samples 156 Chapter 5 The input is RANDSAMP UNIVARIATE ZRN 5 1 SIZE 200000 NSAMP 1 RSEED 100 DSAVE rannormal SELECT CASE lt 100000 EXTRACT rannormall USE rannormal SELECT CASE gt 100000 EXTRACT rannormal2 USE rannormall ESTATISTICS Sl SUM N TEMP sum su s1 TEMP n nu s1 USE rannormal2 syz CSTATISTICS SLASUM ON TEMP sum Sum su
13. Other variables are AS B C A B and C MRCURYDMe Lange et al 1993 The data set consists of measurements of large mouth bass in 53 different Florida lakes to examine the factors that influence the level of mercury contamination Water samples were collected from which the pH level the amount of chlorophyll calcium and alkalinity were measured A sample of fish was taken from each lake for which the age of each fish and mercury concentration in the muscle tissue was measured older fish tend to have higher concentrations To make a fair comparison of the fish in different lakes the investigators used a regression estimate of the expected mercury concentration in a three year old fish as the standardized value for each lake Finally in 10 of the 53 lakes the age of the individual fish could not be determined and the average mercury concentration of the sampled fish was used The variables are 341 Data Files ID Lake ID LAKE Lake name ALKLNTY Measured alkalinity of the lake mg L as Calcium Carbonate PH Measured PH of the lake CALCIUM Measured Calcium of the lake mg l CHLORO Measured Chlorophyll of the lake mg l Average mercury concentration parts per million in the tissue of eee the fish sampled from the lake SAMPLES Number of fish sampled in the lake MIN Minimum mercury concentration in sampled fish from lake MAX Maximum mercury concentration in sampled fish from lake STDMERC Regression estimate of t
14. The data set shows the results of 10 students sitting 14 examination papers for a degree in Statistics Each result is a percentage The variables are TEST1 TEST14 SERUMe Crowder and Hand 1990 The data set consists of the antibiotic serum levels with two types of drugs applied to the same group of volunteers in two phases at different time points TIME 1 TIME2 TIMES TIME6 SICKDATE gt s The data file lists the diagnosed date of each patient s illness DIAGDATE and the date each died MORTDATE These dates are listed in day of the century format SIMUL1 and SIMUL2 These data contain three variables Y J and J Y is generated from N 0 1 5 350 Chapter 9 SLEEPDM Allison and Cicchetti 1976 This data set contains information from a study on the effects of physical and biological characteristics and sleep patterns influencing the danger of a mammal being eaten by predators The study includes data on the hours of dreaming and non dreaming sleep gestation age and body and brain weight for 62 mammals The variables are SPECIES BODY BRAIN SLO SLP DREAM SLP TOTAL SLEEP LIFE GESTATE PREDATION EXPOSURE DANGER Type of species Body weight of the mammal in kg Brain weight of the mammal in g Number of hours of nondreaming sleep Number of hours of dreaming sleep Number of hours of total sleep The life span in years The gestation age Index of predation as a quantitative variable Index of exposu
15. Working Capital as percentage of total assets Retained earnings as percentage of total assets Earning before interest and taxes as a percentage of total assets Sales of total assets in percentages Book value equity divided by book value of total liabilities 321 Data Files BARLEY Fisher 1935 The data are the yields of 10 varieties of barley in two years 1931 and 1932 at 6 sites in the Midwestern US The variables are Y1931 Y1932 VARIETY SITES BBD Myers amp Montgomery 2002 This data set contains observations on viscosity VISCOSITY at different level combinations of the three factors temperature TEMP agitation AGITATION and rate of addition RATE Each factor has 3 levels BIRTHS Walser 1969 The data set consists of information on the FREQUENCY of births in each MONTH labeled as 1 2 12 of a year in the University Hospital of Basel Switzerland BIRTHS 2 Conover 1999 These data were collected in a survey conducted in 7 hospitals of a certain city over a 12 month period divided into 4 seasons SEASON and the numbers of newborn babies BIRTHS in each season were obtained The variables are BIRTHS SEASON HOSPITALS BITS5e The file contains five item binary profiles fitting a two dimensional structure perfectly Variables in the SYSTAT data file are X 1 X 5 BLOCK Neter et al 2004 These data comprise a randomized block design Five blocks of judges BLOCK analyz
16. 2 1106 1109 364 Chapter 9 McFadden D 1979 Quantitative methods for analyzing travel behavior of individuals Some recent developments In D A Hensher and P R Stopher eds Behavioral Travel Modelling London Croom Helm Maltz M D 1984 Recidivism New York Academic Press Marascuilo L A and Levin J R 1983 Multivariate statistics in the social sciences Monterey Calif Brooks Cole Mels G and Koorts A S 1989 Casual Models for various job spects SAIPA 24 144 156 Mendenhall W Beaver R J and Beaver B M 2002 A brief introduction to probability and statistics Pacific Grove CA Duxbury p 424 Messina W S 1987 Statistical quality control for manufacturing managers New Y ork John Wiley amp Sons Metzler J and Shepard R N 1974 Transformational studies of the internal representation of three dimensional objects Hillsdale NJ Erlbaum Mickey R M Dunn O J and Clark V A 2004 Applied statistics Analysis of variance and regression New York John Wiley amp Sons Milliken G A and Johnson D E 1984 Analysis of messy data Vol 1 Designed Experiments New York Van Nostrand Reinhold Milliken G A and Johnson D E 1992 Analysis of messy data Designed experiments Vol I Chapman and Hall Montgomery D C Peck E A and Vining G G 2001 Introduction to linear regression analysis 3rd edition New York John Wiley amp Sons
17. 2 164 Log Likelihood Using Initial Parameter Estimates 270 982 STEP 1 Convergence Criterion 0 050 Stage 1 Estimate Ability with Item Parameter s Constant Log Likelihood Change LR 270 071 0 911 2 486 Greatest Change in Ability Estimate was for Case 80 Change from Old Estimate 0 134 Current Estimate n ZOO Stage 2 Estimate Item Parameter s with Ability Constant HOG Likelihood Change LR 269 662 0 409 1 505 Greatest Change in Difficulty Estimate was for Item BOWELS Change from Old Estimate 0 084 Current Estimate 0 MPSS 5 0 Current Value of Discrimination Index 1 206 STEP 2 Convergence Criterion 0 050 Stage 1 Estimate Ability with Item Parameter s Constant Log Likelihood Change LR 269 590 0 072 1 075 Greatest Change in Ability Estimate was for Case 87 Change from Old Estimate 0 006 300 Chapter 8 Current Estimate gt 2 011 Stage 2 Estimate Item Parameter s with Ability Constant Log Likelihood Change LR 269 549 0 041 1 042 Greatest Change in Difficulty Estimate was for Item BOWELS Change from Old Estimate 0 032 Current Estimate s Las Current Value of Discrimination Index 1 226 Latent Trait Model Item Plots POUNDING SINKING SHAKING PERCENT PERCENT PERCENT 4 2 0 2 0 2 4 4 2 0 ABILITY ABILITY ABILITY NAUSEOUS STIFF FAINT 10 a 5 E 5 i i i o o o th fh th a a a a 4 2 0 4 4 2 0 2 4 4 2 0 ABILITY ABILITY ABILITY VOMIT BOWELS URI
18. ANSFIELD gt Ansfield et al 1977 This study examines the effects RESPONSE of treatments TREATS on two patient groups CANCER those with cancer of the colon or rectum and those with breast cancer NUMBER gives the number of patients in each cancer treatment response group ANXIETY Data are from a National Longitudinal Survey of Young Men conducted in 1979 The data set has been extracted from data set NLS 320 Chapter 9 BANK The data set consists of the description of bank employees The variables are WEIGHT ID SALBEG SEX TIME AGE SALNOW EDLEVEL WORK JOBCAT MINORITY SEXRACE Employee code Beginning salary Sex of employee 0 Male 1 Female Job seniority in months Age of employee in years Current salary Educational level Work experience Employment category 1 Clerical 2 Office trainee 3 Security officer 4 College trainee 5 Exempt employee 6 MBA trainee 7 Technical Minority classification 0 White 1 Nonwhite Sex amp race classification 1 Black Females 2 White Females 3 Black Males 4 White Males BANKRUPTCY Simonoff 2003 The data were collected on 25 telecommunication firms that were declared bankrupt during the period May 2000 January 2002 and 25 telecommunication firms that were not declared bankrupt from December 2000 in their issued financial statements The potential predictors are based on five banking financial ratios WCTA RETA EBITTA STA BVEVL
19. ARIMA autoregressive integrated moving average ARL average run length ARMA autoregressive moving average ARS adaptive rejection sampling ASCII American Standard Code for Information Interchange ASE asymptotic standard error AVG average B BC Bray Curtis similarity measure BFGS Broyden Fletcher Goldfarb Shannon BHHH Berndt Hall Hall Housman BIC Bayesian information criterion BMP Windows bitmap BOOT bootstrap Expansions C C amp RT classification and regression trees CCF cross correlation function cdf CF cumulative distribution function CFA confirmatory factor analysis CGM Computer graphics metafile binary or clear text CI confidence interval COL col column CONV convergence COV covariance Cp process capability index Cpk Process capability index for off centered process CR confidence region CRN Cauchy random number CSV comma separated values CV coefficient of variation CVI cross validation index D DBE Dbase files dep dependent DEVI deviates observed values expected values df degrees of freedom DIM dimension DOS disc operating system DPMO defects per million opportunities DPU defects per unit DTA Stata files Acronyms DWASS Dwass Steel Chritchlow Fligner pairwise comparisons test DWLS distance weighted least squares E EM expectation maximization EMF Windows enhanced metafile EWMA exponential
20. BRAND 87 SYSTAT Basics The F ratio in the Analysis of Variance table at the beginning of the output indicates that there are one or more differences in average price among the seven brands F ratio 10 0415 p value lt 0 0005 Tukey Pairwise Mean Comparisons Let us use SYSTAT s advanced hypothesis testing capability to request Tukey s Pairwise Mean Comparison test From the menus choose Analyze Analysis of Variance Pairwise Comparisons m Specify BRAND under Groups and select Tukey under Tests Analyze Analysis of Variance Pairwise Comparisons Main Available effects Groups Error Terr BRANDS Add m BRANDS Tests Equal variances Tukey Duncan C Dunnett Bonferroni DP R E G4 g 2 Fisher s LSD Hochberg s GT Sidak Gabriel Scheffe Student Mewman Keuls Tukey s b Unequal variances Confidence 0 95 88 Chapter 3 m Click OK Post Hoc Test of COST Using least squares means Using model MSE of 0 100 with 21 df Tukey s Honestly Significant Difference Test BRANDS i BRANDS 3 Difference p Value 95 Confidence Interval Lower Upper gor he 0 190 0 984 0 975 USOS gor sw 0 423 0 590 ES 0 361 gor le 0 844 0 010 A 0 155 gor ww 0 860 0 009 1 549 0 171 gor st o 5 0 001 831 0 379 gor ty 1 440 0 000 2 166 0 714 he Sw 04233 0 968 1 072 0 605 he Te 0 654 0 115 1 404 0 096 he ww 0 670 0 100 1 420 0 080 hce st 0 915 0 016 1 700 S
21. Belsley Kuh and Welsch 1980 The data set is Boston housing prices used in Breiman et al 1984 The variables are CRIM ZN INDUS CHAS NOX RM AGE DIS RAD TAX PTRATIO B LSTAT MEDY BOXES Messina 1987 The ohms of electrical resistance in computer boxes are measured for five randomly selected boxes from each of 20 days of production Thus each SAMPLE contains five observations of resistance in OHMS for each of 20 days DAY BP Hand et al 1996 The data set gives the supine systolic and diastolic blood pressures mm Hg for 15 patients with moderate essential hypertension immediately before and two hours after administering the drug captopril The variables are Systolic blood pressure mm Hg with moderate essential hypertension before SREP EROR administering the drug captopril SYSBP AFTER Systolic blood pressure mm Hg with moderate essential hypertension 2 E hours after administering the drug captopril DIABP BEFORE Diastolic blood pressure mm Hg with moderate essential hypertension gt before administering the drug captopril Diastolic blood pressure mm Hg with moderate essential hypertension 2 DIABP_AFTER hours after administering the drug captopril BRODLIE Brodlie 1980 These data are X and Y coordinates taken from a figure in Brodlie s discussion of cubic spline interpolation BULBe Mendenhall et al 2002 A manufacturer of industrial light bulbs tries to control the variability in le
22. E jeankruptcy syz Bhlbarley syz JEbd syz E Jeiguorld syz E Jeirths2 SYZ E Jeirths syz BHES syz B Block syz B Blockecd syz Eoards yz Sy STAT Data spz syeu spd sys To add an instructional title to the dialog use the PROMPT option The specified prompt text appears in the title bar of the dialog Ensure that the length of the text is limited to that of the title bar Single Variable Tokens To substitute a single variable for a token specify one of the following TOKEN amp var TYPE VARIABLE TOKEN amp var TYPE CVARIABLE TOKEN amp var TYPE NVARIABLE When SYSTAT encounters the token amp var in the command file a dialog prompting the user to select a variable appears If no data file is currently open SYSTAT prompts the user to open a file before proceeding to variable selection 164 Chapter 5 Select Variable Replace amp war with Available variable s Add Selected variable COUNTRYS POP_1990 POP_13903 POP_13506 POP_1330 POP_2020 URBAN BIR TH_ 2 BIR TH_AT DEATH_82 ers Se Continue Cancel Select a variable and click Add Click Continue to continue command processing The list of available variables corresponds to the dialog type The variable list contains only string variables if the token type equals CVARIABLE The NVARIABLE type lists numeric variables for token substitution To list all variables use TYPE VARIABLE Multiple Variable Tokens To substitut
23. Example 9 Chi Square Test Using Choice Tokens In this example we perform chi square test by offering four different choices for specifying expected frequencies You will be prompted to open a data file and select the variable on which the test is to be performed The computations are performed based on your choice of the form and way in which the expected frequencies are stored TOKEN TOKEN ON TOKEN amp filename TYPE OPEN PROMPT Select the file to use IMMEDIATE USE amp filename TOKEN S ovar TYPE NVARIABLE IMMEDIATE PROMPT Select the variable you want to analyze IMMEDIATE TOKEN TYPE CHOICE PROMPT Specify the form and way in which you want to input expected frequencies Equal expected frequencies MMrscel baneous Chorcel sye Equal expected frequencies with missing values as a separate category Miscellaneous choice2 syc Unequal expected frequencies specified in a data file Miscellaneous choice3 syc Unequal expected frequencies specified through the keyboard Miscellaneous choice4 syc 184 Chapter 5 If you select the first choice chi square test is performed using a one way crosstabulation of the input variable by assuming equal expected frequencies across cells The second choice does the same while treating missing values of the input variable as a separate category The third choice accepts unequal expected frequencies in the form of a column in the data file The
24. Examples Use the Examples tab to conveniently execute command scripts given in the user manual with just a click of the mouse The SYSTAT Examples tree is organized by folders and nodes the folders corresponding to each volume of the user manual Double click the nodes to run the underlying commands You can also open these command scripts in the Commandspace for editing and create links to your own command files for easy execution You can even add example nodes to this tab using the Utilities menu See Chapter 5 to know more about the Examples tab Dynamic Explorer The Dynamic Explorer becomes active when there is a graph in the Graph editor and the Graph editor is active Use the Dynamic Explorer to m Rotate and animate 3 D graphs Zoom the graph in the direction of any of the axes See SYSTAT Graphics for more information about the Dynamic Explorer 28 Chapter 2 Commandspace The Commandspace has three tabs m Interactive Batch Untitled E Log Interactive Selecting the Interactive tab enables you to enter commands in the interactive mode which issues the command after you press the Enter key You can save the contents of the interactive tab excluding the gt prompts and then use the file to submit a sequence of commands Batch Untitled Selecting the Batch Untitled tab enables you to work with command files in the batch mode You can open any number of existing command fiels and edit or submit any
25. Scatterplot The input is USE INSTRDM PLOT ACHIEVE APTITUDE GROUP INSTRUCTS OVERLAY BORDER NORMAL ELL SMOOTH LINEAR FCOLOR GRAY SYMBOL 1 8 FILL TITLE Effect of Instructional Methods on Exam Achievement The output is Effect of Instructional Methods on Exam Achievement ACHIEVE 8 5 20 INSTRUCT GENERAL Ml SPECIFIC 50 APTITUDE 307 Applications Toxicology Concentration of nicotine sulfate required to kill 50 of a group of common fruit flies The WILLMSDM data contains the results of a bioassay conducted to determine the concentration of nicotine sulfate required to kill 50 of a group of common fruit flies The experimenters recorded the number of fruit flies that are killed at different dosage levels Variable Description RESPONSE The dependent variable which is the response of the fruit fly to the dose of nicotine sulfate stimulus LDOSE The logarithm of the dose COUNT The number of fruit flies with that response In bioassay it is common to estimate the dose required to kill 50 of a target population For example a toxicity experiment may be conducted to establish the concentration of nicotine sulfate required to kill 50 of a group of common fruit flies The goal is to identify the level of stimulus required to induce a 50 response rate where response may be any binary outcome variable and the stimulus is a continuous variate In bioassay stimuli include drugs toxins
26. Similar observations can be made for the chicken meals 97 SYSTAT Basics Summary The first step in any data analysis is to look at your data SYSTAT provides a wide variety of graphs that can help you identify possible relationships between variables spot outliers that may unduly effect results and reveal patterns that may suggest data transformations for more meaningful analysis SYSTAT also provides a wide variety of statistical procedures for analyzing your data We have covered some of the most common and basic statistical techniques in this chapter and we have still barely scratched the surface Chapter Data Analysis Ouick Tour This chapter provides a quick tour of SYSTAT s capabilities using data from a survey of uranium found in groundwater Groundwater Uranium Overview The U S Department of Energy collected samples of groundwater in west Texas as part of a project to estimate the uranium reserves in the United States Samples were taken from five different locations called producing horizons and then measured for various chemical components In addition the latitude and longitude for each sample location were recorded Several questions are of interest m Does the uranium concentration vary by producing horizon m Is the presence of uranium correlated to the presence of other elements m What is the overall geographic distribution of uranium in the area 99 The data for the groundwater uranium study a
27. The data set consists of measurements of an enzymatic reaction measuring the effects on an inhibitor on the reaction velocity of an enzyme and substrate ENZYME Greco et al 1982 These data measure competitive inhibition for an enzyme inhibitor V is the initial enzyme velocity S is the concentration of the substrate and 1s the concentration of the inhibitor ESTIMs The data set consists of the estimated parameters for each sample of the data set ENZYMDM EURONEW A subset of the WORLD data These data include 27 European countries The variable LABLAT is the latitude measurement of the capital and LABLON is the longitude EX1 Wheaton Muth n Alwin and Summers 1977 The data file is a covariance matrix of 6 manifest variables The original data are attitude scales administered to 932 individuals in 1967 and 1971 The attitude scales measure anomia ANOMIA powerlessness POWRLS and alienation ALNTN They also include a variable for socioeconomic index SET socioeconomic status SES and years of schooling completed EDUCTN EX2 Duncan Haller and Portes 1971 The data is a correlation matrix of manifest variables The original data measure peer influences on ambition These data include the respondent s parental aspiration REPARASP socioeconomic status RESOCIEC intelligence REINTGCE occupational aspiration REOCCASP and educational aspiration REEDASP These data also include the respondent s best frie
28. This example is a demonstration of the use of Design of Experiments DOE in the product development process A four factor two level fractional design is used to minimize the data collection needed to analyze the factors affecting the performance of a fuel gauge SPRING POINTER VENDOR and ANGLE ANOVA The input is USE DESIGNDM ANOVA CATEGORY SPRING REPLACE DEPEND READING ESTIMATE ANOVA CATEGORY POINTER REPLACE DEPEND READING ESTIMATE ANOVA CATEGORY VENDOR REPLACE DEPEND READING ESTIMATE ANOVA CATEGORY ANGLE REPLACE DEPEND READING ESTIMATE 263 Applications The output is Effects coding used for categorical variables in model The categorical values encountered during processing are Variables Levels jen a ae a i ye E a RS a eS 4 SPRING 2 levels 1 000 1 000 Dependent Variable READING N l 16 Multiple R i 0 386 Squared Multiple R 0 149 Estimates of Effects B coo xy Factor Level READING Se ee ee ee 4 CONSTANT 10 500 SPRING 1 1 250 Source Type III SS df Mean Squares F ratio p value aam eni a pa a a a 4 ee ee eee SPRING 25 000 1 25 000 2 448 0 140 Error 143 000 14 10 214 Least Squares Means 16 12 O Z Q lt Lu x 8 El 4 1 1 SPRING Durbin Watson D Statistic tha T03 First Order Autocorrelation 0 404 Effects coding used for categorical variables in model Categorical values encount
29. To impose restrictions on token replacement values define tokens using the TOKEN command with the TYPE option as follows TOKEN amp tokl TYPE tokentype Valid tokentype values include MESSAGE OPEN SAVE VARIABLE NVARIABLE CVARIABLE MULTIVAR NMULTIVAR CMULTIVAR STRING NUMBER INTEGER and CHOICE 160 Chapter 5 During processing when a token is encountered SYSTAT scans for a definition If SYSTAT finds an associated TOKEN definition a dialog consistent with the token type appears Otherwise a default dialog prompts the user for information Resetting Tokens Tokens can be reset individually or globally To clear all tokens use TOKEN without arguments or options Any tokens used in subsequent command lines result in prompting for replacement values To reset an individual token redefine the token using a new TOKEN command For example BAR amp y amp X TOKEN amp x DOT amp y amp X initially prompts for two token values DOT however only prompts for a value for amp x the token reset between the BAR and DOT commands Message Tokens In contrast to all other token types message tokens do not function as substitution markers Instead the message token yields a dialog designed to provide the user with information about the template To define a message token include a command line having the following form in your command file TOKEN amp msg TYPE MESSAGE PROMPT Prompting text appears here MM I
30. ae pos 7 8 c 3 400 0 04 9 3 S O g O g 300 0 03 4 w 1000 0 10 2 2 200 0 02 100 0 01 0 0 00 0 0 0 0 05 0 06 0 07 0 08 0 09 0 10 0 11 0 12 0 13 0 05 0 06 0 07 0 08 0 09 0 10 0 11 0 12 0 13 1200 0 12 2000 0 2 1000 0 10 1500 I I 800 0 08 3 3 jo jo 3 Ss E 2 2 600 0 06 Q 2 1000 0 18 O ke O o D D 400 0 04 YU w 2 g 500 200 0 02 0 0 00 0 0 0 0 60 0 65 0 70 0 75 0 60 0 65 0 70 0 75 RR RBER Applications Maximum likelihood estimates of p g and r evaluated by the scoring method or the EM algorithm are 0 26444 0 09317 and 0 64239 With the available prior information the estimates of p g and r are approximated by the Gibbs Sampling method The empirical estimates of p q and r are 0 25407 0 09003 and 0 65589 respectively Rao Blackwellized estimates are 0 26470 0 09564 and 0 63966 respectively 276 Chapter 8 Manufacturing Quality Control The BOXES data consists of daily measurements of five randomly selected computer components Variable Description DAY The day the sample was taken SAMPLE The sample number for the day 1 5 OHMS The resistance of the component in ohms Quality control charts are used regularly in manufacturing environments to keep track of manufacturing processes diagnose problems and improve operations Potential analyses include descriptive statistics quality control charts ANOVA and time series R Chart of Ohms vs Days The input is USE BOXES QC SHEWHART OHMS DAY TYPE R
31. between them yielding a valid MODEL statement The first HYPOTHESIS command generates a test for each coefficient in the model The second HYPOTHESIS omits the selected variables from the regression model and compares the result with the original model The EFFECT statement for this test requires an ampersand between terms so we define the separator for this token to be amp Example 7 Graph Option Template The Graph tab of the Global Options dialog defines several appearance features for subsequently created graphs As an alternative the following template prompts for scaling percentages line thickness and character size before submitting a command file As a result all graphs created by the specified file use common values for these three global graph characteristics TOKEN ON TOKEN amp xyscale TYPE INTEGER PROMPT Enter the reduction or enlargement for graphs Values below 100 result in reduction Values above 100 result in enlargement TOKEN amp charsize TYPE NUMBER 181 Command Language PROMPT Enter the factor by which to scale graph characters A value of 2 doubles the character size A value of 5 halves the character size TOKEN amp linethickness TYPE NUMBER PROMPT Enter the factor by which to scale line thickness A value of 2 doubles the line thickness A value of 5 halves the line thickness TOKEN amp cmdfile TYPE OPEN PROMPT Open a command file for creating graphs SCALE am
32. birth rate the ratio of birth rate to death rate infant mortality gross domestic product per capita female and male literacy rates average calories consumed per day and the percentage of the population living in cities WORLDDM Wilkinson Blank and Gruber 1996 This data set contains 1990 information on 30 countries including birth and death rates life expectancies male and female types of government whether mostly urban or rural and latitude and longitude The variables are COUNTRY Country name BIRTH RT Number of births per 1000 people in 1990 DEATH RT Number of deaths per 1000 people in 1990 MALE Years of life expectancy for males FEMALE Years of life expectancy for females 359 Data Files GOV Type of government URBAN Rural or urban LAT Latitude of the country s centroid LON Longitude of the country s centroid YOUTH Harman 1976 It is a correlation matrix consisting of measurements recorded for 305 females aged seven to seventeen height arm span length of forearm length of lower leg weight bitrochanteric diameter the upper thigh torso girth and torso width References Afifi A A and Azen S P 1974 Statistical analysis A computer oriented approach New York Academic Press Afifi A A May S and Clark V 2004 Computer aided multivariate analysis 4th ed New York Chapman amp Hall Akima H 1978 A method of bivariate interpolation and smoth surface fitting for
33. editing 137 154 lists 226 opening 141 printing 143 saving 139 submitting 107 137 143 154 Command folder 41 243 command pane 205 Command pushbuttons 35 command shortcuts 135 135 ellipsis 135 command syntax 129 argument 129 Index module name 129 option 129 option value 130 command templates see templates commands 127 abbreviating 130 case sensitivity 130 clipboard submission 154 cold 130 comments 146 controlling output 146 creating command files 137 delimiters 130 DOS 153 editing 137 entering 127 files 126 137 hot 130 interactive 126 127 log 126 150 long filenames 130 multiline commands 130 multiple transformations 135 quotation marks 132 recalling 130 running 126 spaces in filenames 132 submitting 137 143 150 154 syntax 129 130 tokens 156 Commandspace 28 60 126 batch 28 107 126 closing tabs 34 context menu 34 customization 205 docking 205 fonts 126 hiding 205 interactive 28 interactive tab 126 127 keyboard controls 220 log tab 28 126 150 moving 205 resizing 205 209 shortcut keys 220 showing 205 undocking 205 untitled tab 28 126 137 comments 11 146 REM 146 computer graphics metafiles 197 context menu 33 150 212 216 225 batch tab 151 Commandspace 34 144 data editor 33 Examples 34 Examples tab 144 Graph Editor 34 Log tab 150 output editor 33 Output Organizer 34 Startpage 3
34. es is Pi eye se E fe se set E A Fe mies es N of Cases 15 000 15 000 15 000 15 000 15 000 Minimum l 290 000 7 000 14 000 0 000 0 000 Maximum 550 000 34 000 31 000 100 000 40 000 Median 340 000 16 000 22 000 10 000 6 000 Arithmetic Mean 366 000 16 800 22 133 22 267 11 800 95 0 Lower Confidence Limit 327 873 12 247 19 748 6 231 4 735 95 0 Upper Confidence Limit 404 127 2 OS 24 ee 38 302 18 865 69 SYSTAT Basics IRON COST RS EN A a E a a vt a a 4 N of Cases he 15 000 15 000 Minimum 4 000 1 600 Maximum 25 000 3 500 Median ty ER GOO 2 850 Arithmetic Mean 11 800 LAND 95 0 Lower Confidence Limit 87597 Z207 95 0 Upper Confidence Limit 15 003 2 939 The median grams of protein for the 13 diet dinners is 17 the mean is 16 8 For the 15 regular dinners these statistics are 22 and 22 1 respectively Later we will request a two sample 1 test to see if this is a significant difference A 95 confidence interval for the average cost of a diet dinner ranges from 2 27 to 2 75 The confidence interval for the average cost of the regular dinners is larger 2 21 to 2 94 The BY GROUPS variable DIET remains in effect for subsequent graphical displays and statistical analyses To disengage it return to the By Groups dialog box and select Turn off A First Look at Relations among Variables What are the correlations among calories fat content protein and cost We can use correlations to qu
35. hormones and insecticides responses include death weight gain bacterial growth and color change Potential analyses include logistic regression and survival analysis Logistic regression The input is USE WILLMSDM FREO COUNT LOGIT MODEL RESPONSE CONSTANT LDOSE REF O ESTIMATE ONTL LET LDOSEB LDOSE 4895 MODEL RESPONSE LDOSEB REF 0 ESTIMATE LET LDOSEB LDOSE 2 634 MODEL RESPONSE LDOSEB REF 0 ESTIMATE 308 Chapter 8 The output is Case frequencies determined by value of variable COUNT The categorical values encountered during processing are Variables Levels A A Daa paa a es ia ee 4 RESPONSE 2 levels 0 000 1 000 Dependent Variable RESPONSE Analysis is Weighted by COUNT Sum of Weights 25 000 Input Records Paes Records for Analysis 2 Sample Split Category Count A A Secs a 0 RESPONSE 1 REFERENCE Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood 15 000 10 000 Iteration History at at at at at Iterationl Iteration2 Iteration3 Iteration4 Iteration5 Information Criteria AIC Schwarz s BIC 3 5 0 224 0 618 Parameter Estimates Parameter Estimate CONSTANT 0 564 LDOSE 0 919 Odds Ratio Estimates Parameter Overall Model Odds Ratio Fit O LT 133114 pa eo reall T3 z112 Hoe be i Standard Error A iay ee ee a 4
36. procedure However the Clipboard only accesses the last copied item Be sure the most recently copied text corresponds to the commands to be submitted Because the Commandspace itself is a text editor you can also copy commands from any of the tabs for subsequent submission via the Clipboard However other submission methods Submit Window Submit from Current Line to End Submit Current Line and Submit Selection offer the same functionality without replacing the contents of the Clipboard Moreover the command prompt gt prevents successful submission of two or more command lines copied from the Interactive tab 146 Chapter 5 Comments in Command Files The or REM command can be used for inserting comments in command files and for making a command inactive during the current run All text following or REM on the same line is ignored REM Now we merge files side by side REM MERGE filel file2 MERGE filel file3 The text following the first REM command remains in the command file The MERGE statement in the second line is not invoked The command can also be used at the end of another command line You can use this to append comments to a command line The comments could indicate what the command line does why it was written which step of a procedure it is or even the name of the person who has written it Tip To add comments that appear in your output use the NOTE command Commands to Control Output SYSTAT p
37. response surface methods estimation optimization and plotting path analysis conjoint analysis multidimensional scaling perceptual mapping partially ordered scalogram analysis test item analysis signal detection analysis network analysis spatial Statistics and C amp RT 33 Introducing SYSTAT Quick Access Use the Quick Access menu to quickly access all the commonly used statistical procedures You may want to customize this menu to contain those analyses that you frequently use so that you may access all of them in a single location Window Use the Window menu to cascade stack show side by side or arrange the tabs of the Viewspace Help Use the Help menu to access SYSTAT s online Help system Contents Index or Search Acronym Expansions Frequently Asked Questions FAQ demos and tutorials on various SYSTAT features a Quick Reference guide on SYSTAT commands and a list of new and modified commands Through this menu you may also update the license for running SYSTAT beyond the specified period check for updates to the current version of SYSTAT access the SYSTAT website and display the copyright version number and license information of your copy of SYSTAT Context Menus SYSTAT provides several context menus that appear on right clicking in various components tabs or nodes in the three panes of its interface The available menus are listed below with a brief description of each Startpage You can specif
38. sub folders and files in your hard disk There is a check box besides each item to indicate whether or not you want 1t to be included in the Examples tab Click on the check box beside a folder twice 1f you want to include 1t along with all 1ts sub folders and files in the Examples tab The check box changes to Ii when you do so Click on 1t once 1f you want to include just the folder and the files in 1t Click on a file once 1f you want to insert a node corresponding to the file in the Examples tree Clicking again will allow you to uncheck an item When you check a folder ensure that you have expanded all the nodes that belong to 1t so that all the filenames therein are seen Once you have made your selections enter an Example node caption This caption will be set for the top level folder that 1s to contain the links to your example command files Then press Select so that the corresponding tree structure is displayed in the right hand side of the dialog box You can review this tree and make any further changes if desired Once you have finalized your selections press Close This will trigger the creation of an initialization file corresponding to your selections Close the current session of SYSTAT and reopen it to see the newly added examples If you need to replace an examples tree that you have created specify the same Example node caption when you create the new tree 208 Chapter 7 Note You can also customize the tree structure direc
39. the difference in square root units is 4 124 p value 0 001 m diet meals DIETS yes the difference in average CALCIUM content between chicken and pasta is not significant 1 570 p value 0 247 E pasta meals the difference in average CALCIUM content between the DIET yes and no groups is not significant 1 888 p value 0 336 m chicken meals the difference in average CALCIUM content between DIETS yes and no groups is not significant 0 667 p value 1 000 It will be more clear if you see a dot display of these means 96 Chapter 3 E Select Graph Summary Charts Dot Choose CALCIUM as the Y variable and D ET as the X variable Specify FOODS as the grouping variable Select Overlay multiple graphs into a single frame Click the Error Bars tab choose Standard error from the Type group and specify a value of 0 9545 Click Options tab and select Line connected in left to right order m Click OK CALCIUM w HB A Oo NO FOOD chicken X pasta no yes DIET For the regular meals DIETS no the error bars do not overlap indicating a significant difference in calcium content between pasta and chicken However for the diet meals DIETS yes the overlapping error bars suggest no significant difference between the meal types Focusing on the pasta meals the average calcium content for the diet meals 1s within two standard errors of the average calcium content for the regular meals
40. which is consistent with the significant main effect for FOOD but it also shows that the highest values are also regular DIET no dinners This suggests that further investigation might be warranted 95 SYSTAT Basics Bonferroni Pairwise Mean Comparisons Since we have a significant DIET by FOODS interaction we should be cautious about interpreting main effects Let us use SYSTAT s advanced hypothesis testing capability to request Bonferroni adjusted probabilities for tests of pairwise mean differences m From the menus choose Analyze Analysis of Variance Pairwise Comparisons m Specify DIET FOODS under Groups and select Bonferroni under Test group m Click OK Post Hoc Test of CALCIUM Using least squares means Using model MSE of 1 262 with 18 df Bonferroni Test DIETS i FOODS DIETS 3 FOODS 1 3 Difference p Value 95 Confidence Interval Lower Upper no chicken noO pasta 4 124 0 000 6 478 1 770 no chicken yes chicken 0 667 1 000 2 464 1 131 no chicken yes pasta 225236 0 025 4 252 O Zk no pasta yes chicken 3 457 0 002 1 204 SUL no pasta yes pasta 1 888 0 201 0 543 4 318 yes chicken yes pasta O 0 148 3 467 0 328 We are interested in four of the six differences and probabilities in these panels First we look within diets and then within food types For the m regular meals DIETS no the difference in average CALCIUM content between chicken and pasta meals is highly significant
41. 000 1 000 659 000 0 569 20 000 1 000 749 000 0 536 18 000 1 000 803 000 0 501 16 000 1 000 1020 000 0 464 15 000 1 000 1042 000 0 427 Group size 362000 Number Failing 21 000 Applications 284 Chapter 8 Cumulative Hazard Plot 2 0 o gt N o Cumulative Hazard e P K_M PROBABILITY i E KE 0 0 O 1000 2000 3000 4000 5000 6000 7000 8000 Time Log Rank Test Stratification on SEX Strata Range 1 to 2 Chi Square Statistic Method with 1 df p Value A a ee a ey E Mantel 0 568 0 451 Breslow Gehan 1 589 0 207 Tarone Ware 1 167 0 280 Stratified Kaplan Meier Estimation The input is USE MELNMADM SURVIVAL MODEL TIME CENSOR CENSOR STRATA SEX ESTIMATE LTAB The output is Time Variable TIME Censor Variable CENSOR Input Records A Records Kept for Analysis 69 Censoring Observations at at fe te a ae 4 Exact Failures 36 Right Censored 33 285 Type 1 Exact Failures and Right Censoring Overall Time Range Failure Time Range 72 000 ZOO Stratification on SEX specified Nonparametric Estimation Table of Kaplan Meier Probabilities With stratification on SEX The following results are for SEX 0 Number at Risk Group size Number Failing PREP RP RRP PPP PPE Lp Number Failing Product Limit Likelihood 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 Mean Survival Tim
42. 12 Cluster Analysis 13 Fitting Distributions 14 Hypothesis Testing for Two Sample Data in Columns 15 Least Squares Regression 16 Logistic Regression 17 Mixed Models Descriptions for each of the above items are given in the following pages New Features 1 Autohide Spaces You can autohide the Workspace and Commandspace by clicking the 4 button For details about customizing the SYSTAT window refer Chapter 7 Customization of the SYSTAT Environment in the Getting Started volume of the user manual Choice Tokens SYSTAT now allows you to define choice tokens using a new type of token dialog box where you may specify between 2 to 10 choices Each choice may be linked to a SYSTAT command script so that depending on the user s choice the corresponding script will be executed This gives you the ability to incorporate several up to 10 sets of scripts covering various possible scenarios for a given analysis into a single SYSTAT command script Depending on the user s choice any given set may then be executed Data Edit Bar The Data Variable Editor has a new toolbar called the Data Edit Bar This allows you to navigate to any cell in the Data Editor and view edit data values For more details about the Data Edit Bar refer Chapter 3 Entering and Editing Data in the Data volume of the user manual 6 Chapter 1 10 Data File Information You can click the button in the bar beside the Data and Variables tabs
43. 25 Graph menu 31 help 38 Help menu 33 Output Organizer 27 View menu 31 Viewspace 21 workspace 21 User Menu 144 Utilities menu 31 217 Examples 31 Macro 31 Recent Dailogs 31 Theme Menus 22 User Menu 31 y Variable Editor 33 context menu 34 processing conditions 25 variable properties 25 variables adding 173 177 substituting for tokens 163 164 173 179 VDISPLAY 244 view data 24 View menu 31 Commandspace 31 commandspace 31 processing conditions 31 Startpage 31 Workspace 31 Index workspace 31 Viewspace 22 data editor 22 24 full screen 31 Graph Editor 25 maximizing 208 output editor 22 23 tile 208 W Window 144 Window menu 33 arrange 33 Arrange Icons 33 Cascade 33 Tile 29 Tile Vertically 29 windows tiling 29 WMF 196 Workspace 27 customization 206 Dynamic Explorer 27 Examples tab 27 hiding 206 Output Organizer 27 resizing 209 wrapping text 239
44. 4 ee er re rr or ee eee re 95 Upper 95 Confidence Interval Lower Confidence Interval Lower Upper Upper 311 Applications Plot of Logistic Model The input is USE WILLMSDM FREQ COUNT LOGIT MODEL RESPONSE CONSTANT LDOSE REF 0 ESTIMATE SAVE QUANT ONTL REM CREATES PLOT OF LOGISTIC MODEL WITH LIMIT LINES ADDED AT THE REM UPPER REM AND LOWER LIMITS FOR THE LDOSE VALUE CORRESPONDING TO A REM PROBABILITY HAS 50 USE QUANT BEGIN PLOT PROB LDOSE SIZE 0 XLAB YLAB XLIMIT 3 364 0 746 XMIN 5 XMAX 5 XTICK 4 ACOLOR RED YTICK 4 YMAX 1 YMIN 0 PLOT PROB LDOSE SIZE 0 SMOOTH SPLINE TENSION 0 500 XMIN 5 XMAX 5 XTICK 4 XLAB LDOSE YLABR Probability YLIN T 0 5 YTICK 4 YMAX 1 YMIN 0 USE WILLMSDM LET PDEAD COUNT 5 SELECT RESPONSE 1 PLOT PDEAD LDOSE SYM 2 YTICK 4 YMAX 1 YMIN 0 XMIN 5 XMAX 5 XTICK 4 XLABS YLAB Y SCALES NONE TITLE Logistic Model END 312 Chapter 8 The output is Logistic Model 0 75 Probability o 0 25 0 00 5 0 2 5 0 0 2 5 5 0 LDOSE Data References Anthropology Data Sources Original Source Thomson A and Randall McIver R 1905 Ancient races of the Thebaid Oxford Oxford University Press Data Reference Hand D J Daly F Lunn A D McConway K J and Ostrowski E 1994 A handbook of small data sets New York Chapman amp Hall pp 299 30
45. 72000 1 26 2 256 000 33152 101 798 TPO 1 000 12330 7 600 744 000 372 000 12000 6100 124000 3 27 2 258 000 33 068 101 859 TPO 1 000 6730 6100 500 000 100 000 6000 o400 75 000 28 2 260 000 33 123 101 877 TPO 1 000 11 070 6 700 1 000 000 150 000 10000 0100 200000 3 2 262 000 33 171 101 932 TPO 1 000 16 080 6400 660 000 13 000 7 000 0100 66 000 7 2 264 000 33 168 101 921 TPO 1 000 0100 0 800 4 980 000 2000 33000 0200 2 000 1 3 2 267 000 33 282 101 912 TPO 1 000 0100 73 500 2 000 000 150 000 60 000 0 200 500 000 32 2 270 000 33 263 102 001 TPO 1 000 8620 12 000 1 250 000 76 000 50 000 2800 250 000 1 33 2272 000 33123 101 915 TPO 1 000 11 430 28 000 1 155 000 63 000 66 000 1100 264 000 7 34 2 273 000 33118 102 002 TPO 1 000 17 960 12 600 1 750 000 60 000 40 000 0 600 200 000 1 0 35 2 275 000 33 215 101 980 TPO 1 000 15520 9400 1 500 000 50 000 100 000 0 800 200 000 1 36 2276 000 33 228 101 805 TPO 1 000 21 4909 6 200 837 000 36 000 7 000 0800 45 000 1 2 37 2278000 33 281 101 854 TPO 1 000 9460 16 300 1 155 000 66 000 66 000 6400 132 000 4 38 2 281 000 33 217 100 588 PGWC 4000 2050 7 30015 000 000 2000 2000 0100 2000 35 an 0 e ERE NN A SYSTAT dynamically links data across graphs and the Data editor These cases are now selected If you were to run a statistical analysis or plot another graph at this point it would use only these two cases As pointed out earlier SYSTAT manages data and graphics globally M
46. A M 1985 Data A Collection of Problems from Many Fields for the Student and Research Worker 123 126 Springer Verlag New York Chapter Command Language Revised by Rajashree Kamath Most SYSTAT commands are accessible from the menus and dialog boxes When you make selections SYSTAT generates the corresponding commands Some users however may prefer to bypass the menus and type the commands directly at the command prompt This is particularly useful because some options are available only by using commands not by selecting from menus or dialog boxes Whenever you run an analysis whether you use the menus or type the commands SYSTAT stores the processed commands in the command log A command file is simply a text file that contains SYSTAT commands Saving your analysis in a command file allows you to repeat it at a later date Many government agencies for example require that command files be submitted with reports that contain computer generated results SYSTAT provides you with a command file editor in its Commandspace You can also create command templates A template allows customized repeatable analyses by allowing the user to specify characteristics of the analysis as SYSTAT processes the commands For example you can select the data file and variables to use on each submission of the template This flexibility makes templates particularly useful for analyses that you perform often on different data files or for combi
47. Case labels m Click OK to execute the program CALORIES 600 al Oo o gt jo o oO O o 10 20 FAT 30 40 O Select variable Select variable BRANDS e 600 e chicken gt al jo oO o o CALORIES w z i af pasta 0 10 20 30 40 59 SYSTAT Basics The top point in each plot is a chicken dinner made by sw it must be fried chicken Notice that the beef dinner by gor at the far right close to the 300 calorie mark contains considerably more fat than other dinners in the same calorie range Do diet dinners really have fewer calories and less fat than regular dinners The dinners in the sample were selected from shelves where both regular and diet dinners were featured DIET no and yes respectively m Return to the Scatterplot dialog box m Select DIET as the grouping variable m Select Overlay multiple graphs into a single frame a Deselect Display case labels in the Symbol and Label tab and select None as the Smoother method in the Smoother tab Click the Options tab in the Scatterplot dialog box Select Confidence kernel and enter a p value of 0 75 for a 75 confidence region Click OK 600 500 09 uI 400 Y O Z 300 O 2005 DIET no 100 ne 0 10 20 30 40 FAT It is clear from the sample that the DIETS yes dinners have fewer calories and less fat than the regular dinners 60 Chapter 3 Using Commandspace Each time you use a dialog
48. Ctrl click the other variables that you want Avoid the name area while clicking and dragging To select all the variables in a list click inside the list and press Ctrl A or right click and select Select All You can also right click on a variable or a highlighted set of variables and use the menu that pops up to add them to the desired target list or remove them from the list Additional Features Several additional features have been provided for the dialog boxes They are Keyboard shortcuts as an alternative to check boxes and radio buttons Hold down the Alt key and press the underlined letter in the caption The Tab key to navigate between items For an edit text taking numeric values tooltips indicating the valid range displayed while pausing the mouse on the edit text Edit texts taking integer values not accepting the decimal separator as input Edit texts taking nonnegative values not accepting the negative sign as input Edit texts to contain filenames of files to be opened or saved for features that uire or support such options Type the desired filename with path or press the button and select a file 38 Chapter 2 Getting Help SYSTAT uses the standard HTML Help system to provide information you need to use SYSTAT and to understand the results This section contains a brief description of the Help system and the kind of help provided with SYSTAT The best way to find out more about the Help syste
49. EXIDE ra ds iS AAA 172 Working with Output 185 Oia IIA a 186 FOCO got la de a is AAA ed ee 186 o A E ON 188 Output Editor Right Click Menu 188 Output Organizer es 6 ao ck e a Ae a rs aa 189 To Move Output Organizer Entries 190 To asen Tree POE te a A 191 Configuring the Output Organizer 191 Output Organizer Right Click Menu 193 Saving Output and Graphs 193 To save JPL ss ao ao a a r A 194 To Save Results from Statistical Analyses 195 TOS aye LADOS cdi AA a a 196 To Export Results to Other Applications 197 vi A E 199 PAC Preview eka a a AAA 200 PACS CULO ww der AY de hoe Ne ri Le os 200 Printing Graphs Using Commands 201 Customization of the SYSTAT Environment 203 Commandspace Customization 000048 205 Hiding the Commandspace 048 205 Workspace Customization 2 0 0 a a a ee ee eee 206 Customizing the Output Organizer 206 Adding Examples 20 000 e eae 206 Viewspace Customization 2 0 000 eee eee 208 Maximizing the Viewspace 0 00004 208 Startpage Customization 2 aoa o e e e a a a 209 SA BaT oe oea inet ac eee a O E ee a 209 Status Bar Customization o oo oa a 00 0004 211 Customizing Menus and Toolbars in SYSTAT 212 Menu Customization ooo a id ddd 212 Commands C
50. FOODS and within the foods by fat content m From the menus choose Data Sort File m In the Sort dialog box select FOODS and FAT as the variables and then click OK 61 SYSTAT Basics h Data Sort File Available variable z Selected variable z BRANDS TITTY FOODS eer Far CALORIES A FAT ao F emove PROTEIN MATARLA IA A Order Ascending Save file Descending E From the menus choose Data List Cases m Select FOODS FAT CALORIES PROTEIN and BRANDS as the variables m Inthe Format group enter 7 for Column width and 0 for Decimal places q Data List Cases Ayvallable variable z BRANDS FOODS CALORIES FAT Remove BRANDS PROTEIN ATA RADA Format All cases Col e MW pati O Cases 1 through E Decimal places 0 O Late abels m Click OK 62 Chapter 3 Case FOODS FAT CALORIES PROTEIN BRANDS ee i a 4 ee eee 1 beef 8 290 18 gor 2 beef 9 330 25 SW 3 beef 14 330 24 ty 4 beef 19 370 24 st 5 beef 24 390 20 st 6 beef 34 300 22 gor i chicken 0 190 12 WW 8 chicken 1 160 13 WW 9 chicken 2 200 ES he 10 chicken 3 280 24 he 11 chicken 4 260 21 WW 12 chicken 5 240 19 LE 13 chicken 5 240 18 LE 14 chicken 6 270 22 LG LS Chicken 7 340 3L ty 16 chicken 8 400 27 ty 17 chicken 10 320 27 st 18 y chicken 16 330 18 st 19 Chicken 24 430 20 ty 20 chicken 25 550 22
51. G W 1976 Northwest Texas pilot geochemical survey Union Carbide Nuclear Division Technical Report K UR 1 Ott R L and Longnecker M 2001 Statistical methods and data analysis Sth edition Pacific Grove CA Duxbury p 223 Pearson K and Lee A 1903 On the laws of inheritance in man I Inheritance of physical characters Biometrika 2 357 462 Prentice R L 1973 Exponential survival with censoring and explanatory variables Biometrika 60 279 288 Rao C R 2002 Linear Statistical Inference and its Application 2nd ed John Wiley amp Sons Reisby N Gram L F Bech P Nagy A Petersen G O Ortmann J Ibsen I Dencker S J Jacobsen O Krautwald O Sondergaard I and Christiansen J 1977 Imipramine clinical effects and pharmacokinetic variability Psychopharmacology 54 263 272 Robinson D 1987 Estimation and use of variance components The Statistician 36 3 14 Rothkopf E Z 1957 A measure of stimulus similarity and errors in some paired associate learning tasks Journal of Experimental Psychology 53 94 101 Rousseeuw P J and Leroy A M 1987 Robust regression and outlier detection New York John Wiley amp Sons Ryan T P 2002 Statistical methods for quality improvement New York John Wiley amp Sons Schiffman S S Reynolds M L and Young F W 1981 Introduction to multidimensional scaling Theory methods and applications New York
52. Haller A O and Portes A 1971 Peer influence on aspirations a reinterpretation Casual Models in Social Sciences H M Blalock ed 219 244 Aldine Atherstone Efron B and Tibshirani R 1993 An Introduction to the bootstrap Chapman and Hall New York London Ekman G 1954 Dimensions of color vision Journal of Psychology 38 467 474 Fellner W H 1986 Robust estimation of variance components Technometrics 28 51 60 Fisher R A 1935 The design of experiments 7th ed New York Hafner Fisher R A 1936 The use of multiple measurments in taxonomic problems Annals of Eugenics 7 179 188 Flury B and Riedwyl H 1988 Multivariate statistics A practical approach London Chapman and Hall 362 Chapter 9 Franses P H and Dick van Dijk 2000 Non linear time series models in empirical finance Cambridge University Press Datastream Frets G P 1921 Heredity of head form in man Genetica 3 193 384 Gaver D P and O Muircheartaigh I G 1987 Robust empirical bayes analysis of event rates Technometrics 29 1 15 Gibbons J D and Chakraborti S 2003 Nonparametric statistical inference 4th ed Boca Raton Florida CRC Press Gilfoil D M 1982 Warming up to computers A study of cognitive and affective interaction overtime In Proceedings Human factors in computer systems Washington D C Association for Computing Machinery Goldstein H 1987 Mult
53. Huber la Power Mean Trimmed Residuals Aes O DWLS O Mode Tension 0 5 All Ases Limit smoother domain to data range Layout Contidence interval on regression line Sii gt Color Fill Symbol and Label Surface and Line Style m Click OK to execute the program The resulting line displays a typical calorie value for each value of FAT without fitting a mathematical equation to the complete sample 57 SYSTAT Basics 600 al O O CALORIES w D O O O O 20045 100 0 10 20 30 40 FAT The smoother indicates not surprisingly that foods with a higher fat content tend to have more calories You may wonder what foods and what brands have the most calories The fewest calories The highest fat content The lowest fat content Return to the Scatterplot dialog box m Click the Symbol and Label tab in the Scatterplot dialog box click Display case labels in the Case labels group select BRANDY to label each plot point with the brand of the dinner and set the case label size to 1 3 Repeat these steps for FOODS 58 Chapter 3 B Graph Scatterplot Main Options Smoother Residuals Coordinates els Als All Axes Layout Color Surface and Line Style Symbol type 2 Automatic symbol selection Select symbol o Symbol size 2 Default symbol size fal Es Enter size fal ES Name CIRCLE Enter character a Select variable
54. OSAVE S amp OUTPUT amp PUT amp ROOT amp SAVE amp SUBMIT amp USE amp WORK amp HTML amp TEMPDIR amp RTF Token Value Folder to which data will be exported Folder containing ASCII data for import by BASIC Folder to which graphs will be saved Folder from which data will be imported Folder to which SYSTAT output will be saved Folder to which ASCII output will be saved Folder to which ASCII data will be exported by BASIC Folder to which SYSTAT is installed Folder in which SYSTAT data files will be saved Folder from which SYSTAT comand files will be submitted Folder from which SYSTAT data files will be opened Folder to which temporary SYSTAT data files will be saved Folder to which HTML or MHT output will be saved Folder to which temporary files created by SYSTAT are saved Folder to which RTF output will be saved Most of the built in tokens are directly associated with the corresponding SYSTAT commands You can use these appropriately in your command scripts so that files are opened from or saved to paths other than the assigned ones without changing the default path For example the command SUBMIT amp WORK filenamel syc submits filenamel syc from the path assigned to amp WORK without changing the path specified in amp SUBMIT 172 Chapter 5 In the case of the USE command SY STAT first searches in the path assigned to amp SAVE If the file 1s not found there then it searches in the amp USE
55. PLIMITS 025 975 The output is Number of Lines of Input Data Read 100 00000 Number with Missing Data or Zero Weight 0 00000 Number of Samples to be Plotted 20 00000 Only Subgroups Containing Data are Plotted Estimated Population Mean Se T9 93100 Estimated Population Standard Deviation 0 90730 Total N Excluding Missing Data 100 217 Applications R Chart for OHMS with Alpha 0 05 UCL 3 80798 Center 2 11032 LCL 0 77091 X bar Chart of Ohms vs Days The input is USE BOXES QC SHEWHART OHMS DAY TYPE XBAR The output is Number of Lines of Input Data Read 100 00000 Number with Missing Data or Zero Weight 0 00000 Number of Samples to be Plotted 20 00000 Only Subgroups Containing Data are Plotted Estimated Population Mean 19 93100 Estimated Population Standard Deviation 0 90730 Total N Excluding Missing Data 100 X BAR Chart for OHMS with Alpha 0 0027 UCL 21 1483 Center 19 931 LCL 18 7137 0 5 10 15 20 25 DAY 278 Chapter 8 Medical Research Clinical Trials The CANCERDM data set contains information from a study of the effects of supplemental Vitamin C as part of routine cancer treatment for 100 patients and 1000 controls that is 10 controls for each patient Variable Description CASE Case ID ORGAN Organ affected by cancer SEXS Sex of patient AGE Age of the patient SURVATD Survival of patient measured from first
56. SW 21 pasta 3 250 20 hc 22 pasta 4 210 9 lc 23 pasta 4 220 14 ww 24 Pasta 6 220 15 WW 25 pasta 8 260 tS LG 26 pasta 12 300 14 SW 2T pasta 16 370 20 gor 28 pasta 26 440 20 gor Within each type of food the fat content varies markedly The diet brands ww lc and he are the first entries under chicken and pasta If the data file were larger you would have to scan pages and pages of listings and it would be hard to see relationships see the descriptors in the next section Note that you can sort and list data in any procedure A Quick Description As an early step in data screening it is useful to summarize the values of grouping variables and to scan summary descriptors of quantitative variables Frequency Counts and Percentages The One Way Frequency Tables on the Analyze menu features many Print options that allow you to customize exactly what reports appear in your output For example the Frequency distribution option reports the number of times frequency each category of a grouping variable occurs and expresses it as a percentage of the total sample size Cumulative frequencies and percentages are also available In our grabbing sample 63 SYSTAT Basics strategy we are interested in knowing what foods and how many of each brand and diet type we have E From the menus choose Analyze One Way Frequency Tables m Inthe Tables group of the One Way Tables dialog box select Frequency dist
57. This can be changed Also it 1s possible to change the display to Normal Exponential notation or Date and time 50 Chapter 3 m Click the top left data cell under the name of the first variable and enter the data m To move across rows press Tab after each entry To move down columns press the Enter key or down arrow key The Data editor will look like this AE 4 SYSTAT Output TE USE gt al CALORIES Lean Cuisine 240 000 Weight Watchers 220 000 Healthy Choice 250 000 Stouffer 370 000 Gourmet 440 000 Tyson 330 000 Swanson 300 000 1 2 3 4 5 6 Y 8 9 y o gi aE Fr i gt a For Help press F1 m When you have finished entering the data from the menus choose File Save As m Select the location for saving the file m Type SAMPLE as the name for the data file SYSTAT adds the suffix SYZ SAMPLE SYZ 51 SYSTAT Basics Reading an ASCH Text File This section shows you how SYSTAT reads raw ASCII data files created in a text editor or word processor SYSTAT can import ASCII files of the type txt dat and csv SYSTAT can read alphanumeric characters delimiters spaces commas or tabs that separate consecutive values from each other and carriage returns SYSTAT cannnot read an ASCII file which contains any unusual ASCII characters or page breaks control characters column markers or similar formatting codes See your word processor s
58. a new location in the user interface without releasing the mouse button Release the mouse button do not press the Ctrl key while you do this when the outline is at the desired position and touches either one of the edges of the user interface or that of the Viewspace Double click the title bar of an undocked Commandspace to reattach it at its last docked position Hiding the Commandspace An undocked Commandspace always appears in front of the rest of the user interface and may obscure output In such a situation it can be hidden until needed Selecting Commandspace from the View menu pressing Ctrl W right clicking in the toolbar area and selecting Commandspace or clicking the Close button after undocking it toggles the visibility of the Commandspace Alternatively you can hide the Commandspace and use a text editor like Notepad for command entry The Commandspace can be collapsed by clicking the pin 4 button 206 Chapter 7 Tip Users who favor dialog use over typing commands should hide the Commandspace to maximize the area available for output Workspace Customization The technique to customize the Workspace 1s analogous to that explained for the Commandspace The Workspace can also be hidden either by invoking the View menu and selecting Workspace by right clicking on the toolbar area and selecting Workspace or by clicking the Close A button after undocking the Workspace You can collapse auto hide the Workspac
59. access the Graph Gallery and to create function plots summary charts like pie doughnut bar line profile pyramid cone cylinder and high low close density displays like histograms dot densities and box plots distribution plots like density functions probability plots and quantile plots scatterplots 32 Chapter 2 scatterplots matrices parallel coordinate displays Andrews s Fourier plots icon plots and maps You can also overlay various graphs in a single frame When the Graph editor is active with a graph in it you can realign any displaced graph frames with their original positions edit various properties of the graph like font attributes of graph frame titles axes tick mark bar and case labels zoom rotation layout position size and arrangement title background color type for summary and density charts and coordinate system of graphs axes scale type tick mark style and location label limit lines grid lines transformations line style and scale ranges on the graph s axes titles labels location and layout of graph legends colors and fill patterns for the graph s elements style and size of plot symbols surface gradient and wireframe styles and various options for each graph type The Graph menu also allows you to copy graphs define text annotation font and graph annotation attributes select the pointer tool or any of the annotation tools select the panning or zooming tools reset any panning or zoom
60. automobile manufacture unit PLANKS Netmaster Statistics Courses After drying beech wood the humidity level at any given point inside a plank typically depends on the depth of the point To study the relation between the humidity levels measured as a percentage the depth and twenty different randomly selected beech planks were measured for humidity level at five depths and three widths The variables are PLANK WIDTH DEPTH and HUMIDITY PLANTS SYSTAT created this file to demonstrate regression with ecological or grouped data The variables are CO2 SPECIES and COUNT PLOTS The split plot design is closely related to the nested design In the split plot however plots are often considered a random factor Thus different error terms are constructed to test different effects Here is an example involving two treatments A between plots and B within plots The numbers in the cells are YIELD of the crop within plots These data also use PLOT PLOT 1 and PLOT 2 as variables 346 Chapter 9 POLAR These data show the highest frequency FREQ in 1000 s of cycles per second perceived by a subject listening to a constant amplitude sine wave generator oriented at various angles relative to the subject ANGLE POLYNOM The following variables were created in SYSTAT using the equations X uti l0 Y 2 3 X 44X 58 500 z where u is a uniform random variable is an index running from 1 to 20 and z is a standard normal
61. axis TOKEN amp yvarlab TYPE STRING IMMEDIATE PROMPT Enter a label For the y axis TOKEN amp zvar TYPE NVARIABLE IMMEDIATE PROMPT Select a variable for the z axis TOKEN amp zvarlab TYPE STRING IMMEDIATE PROMPT Enter a label for the z axis TOKEN amp pltitle TYPE STRING PROMPT Enter a title for the plot TOKEN amp symlabel TYPE CVARIABLE PROMPT Select a variable to use for labeling the plot points TOKEN amp symsize TYPE NVARIABLE PROMPT Select a variable to use for sizing the plot points PLOT amp zvar amp yvar amp xvar SIZE amp symsize LABEL symlabel TITLE amp pltitle XLAB amp xvarlab YLAB amp yvarlab ZLAB amp zvarlab We use the IMMEDIATE option to ensure that the axis labeling prompts occur immediately after the corresponding axis assignment In the PLOT command we enclose the string tokens in quotation marks Doing so preserves the case of the entered value and prevents potential syntax errors resulting from spaces in the replacement text Variable Creation The VARIABLE NVARIABLE CVARIABLE MULTIVAR NMULTIVAR and CMULTIVAR types of the TOKEN command allows the user to select a variable or variables from those found in the current data file These types cannot be used to create new variables Instead use the STRING type for variable creation 175 Command Language In this example we create ten new variables Each variable contains 100 cases drawn
62. be used to transform both the Alkalinity and Standard Mercury variables so that they meet the assumptions of linear regression The graph below has X Power 0 7 Y Power 0 4 271 Applications Measured Mercury Levels in Freshwater Fish vs Alkalinity 0 50 100 Alkalinity Genetics Bayesian Estimation of Gene Frequency Note This example will work with the Monte Carlo add on module version 1 Rao 1973 illustrated maximum likelihood estimation of gene frequencies of O A and B blood groups through the method of scoring McLachlan and Krishnan 1997 used the EM algorithm for the same problem This application illustrates Bayesian estimation of these gene frequencies by the Gibbs Sampling method Consider the following multinomial model with four cell frequencies and their probabilities with parameters p q and rwithp qw r 1 Letn no9 ny nNg Nygp Data Model no 176 Ny 182 NB 60 NAB 17 212 Chapter 8 Let us consider a hypothetical augmented data for this problem to be no n44 1 40 NBB ngo nag With a multinomial model n 1 p q p 2p 1 p q 4 2q 1 p q 2pq With respect to the latter full model n44 ngg could be considered as missing data MODEL X Multinomialg 435 1 p q p 2p 1 p q 2g 1 p 9 2py Prior information p q r Dirichlet a P y The full conditional densities take the form 2 a p n Binomial n gt pe a 2 line Binomial
63. being tested in this case the normality of the residuals is to be accepted or rejected The smaller the p value the stronger is the evidence against the hypothesis Since in this case the value is near 0 0 up to 3 places of decimal the normality hypothesis of residuals is rejected When the assumption of normal residuals cannot be justified even for a transformed variable we may consider nonparametric methods which do not depend on such assumptions Nonparametric Tests Now we see how the question earlier answered by using ANOVA with normality assumption on residuals can be answered by a nonparametric test which does not make this assumption Now you might ask Why then bother with ANOVA at all The answer is If the normality assumption actually holds then ANOVA is a more powerful method but it is not valid when the assumption fails If we do not have a good distribution model for URANLOG or a transformed variable then it is safer to use a 115 Data Analysis Quick Tour distribution free nonparametric method even 1f 1t 1s not powerful For a nonparametric test for the equality of URANLOG levels at various horizons From the menus choose Analyze Nonparametric Tests Kruskal Wallis m Select URANLOG as the Selected variable s and HORIZON as the Grouping variable R Analyze Nonparametric Tests Kruskal Wallis Ed es Available variable z Selected variable z Resampling SAMPLE URANLOG LATITUDE LONGTUDE HO
64. beyond thrice this number 1 e 45 characters 240 Chapter 7 You can set a different number here as desired You can even uncheck this option to prevent wrapping Truncate text in tables Apart from wrapping the text in tables can also be truncated By default in each cell the truncation will happen at 45 characters You can change this number or even turn off truncation Display statistical Quick Graphs You can turn the display of the Quick Graphs on and off By default SYSTAT automatically displays Quick Graphs Echo commands in output Includes commands in the Output Editor before the subsequent output Use SYSTAT classic output style Displays all subsequent statistical output as ASCII text using the Courier font With this option selected no output appears in formatted tables Variable label display If a variable label is defined for a variable it will be used to identify the corresponding variable in the output instead of the variable name itself Select Both if you want both variable names and labels to be used or Name if you want just the variable names to be used Value label display If value labels are defined for a variable they will be used to represent the underlying data values in the output You can select Both to display both value labels and data values and Data to display just the data values Image format The graphs created by SY STAT in the Output Editor are in the portable network graph
65. characters like the plus minus asterisk hash and exclamation mark are not used as they may be used in other parts of a command Interactive Command Entry Commands can be issued automatically when the Interactive tab 1s selected in the Commandspace To issue a command type the command and press the Enter key SYSTAT s commands can be categorized into four broad categories general commands data related commands graph related commands and statistical commands The statistical commands are in turn grouped by module While the other commands are available for use at any time the statistical commands will only function after you enter or in other words load the relevant module The modules are as follows ANOVA BAYESIAN BETACORR BLOGIT CFA CLOGTT CLUSTER CONJOINT CORAN CORR DELOGHT DESIGN DISCRIM EXACT FACTOR FITDIST GAUGE GLM IIDMC LOGLIN MANOVA MCMC MDS MISSING MIX MIXED MLOGIT MS IGMA NETWORK NONLIN NPAR PERMAP PLS POLY POSAC POWER PROBIT QC RAMONA RANDSAMP RANKREG RDISCRIM REGRESS RIDGEREG ROBREG RSM SAVING SERIES SETCOR SIGNAL SMOOTH SPATIAL SURVIVAL TESTAT TESTING TLOSS TREES TORS VC XTAB Note 1 There are three other modules in SYSTAT that are not listed above viz BASIC MATRIX and STATS Commands related to these modules will work directly without having to load the modules In other words they function just like the general commands 2 Some of these modules are available only as add
66. companies produce certain pesticides Company A produces three such products companies B and C produces two such products each and company D produces four such products No company produces a product exactly like that of another The treatment structure is a two way with COMPANYY as one factor and PESTICIDE as the other To compare these we use 33 glass containers that are randomly grouped into eleven groups of three The pesticides are assigned randomly to the groups The assigned pesticide is applied to the inside of each box in its group A box with 400 mosquitoes and soil with bluegrass is put inside each container and the number of live mosquitoes in each box was counted after 4 hours Y PESTRESIDUE Kuehl 2000 A comparison was made among two standard pesticide methods to compare and test the amount of residue left on cotton plant leaves is the same for the two methods METHOD To test these six batches BATCH of plants were sampled from the field Two plants were used in the experiment from each batch Thus there were twelve plants in the experiment SAMPLE The plants inside each batch were from the same field plot Method one was applied to three randomly selected batches and the remaining three batches 345 Data Files were given method two The amounts of residue on the leaves were measured after a specified amount of time for each of the twelve plants Y PHONECAL Rousseeuw and Leroy 1987 The data set which comes f
67. contents paste the copied portion to the batch Untitled tab or Interactive tab edit the pasted commands and submit the resulting syntax Recording Scripts SYSTAT provides you an option to reuse a part or whole of the log file of the current session To start stop recording the scripts m From the menus choose Utilities Start Stop Recording Or m Click on the Record Script tool CI provided in the Standard toolbar The Record Script dialog pops up when you stop the recording Record Script Save to file Add to user menu You can save the recorded script to a file and or you can add it to the User Menu for use in subsequent sessions For more information on the User Menu see Chapter 7 Customization of the SYSTAT Environment Quit the dialog by pressing Cancel if you do not want to save the recorded script There is also another way to reuse the recorded commands m From the menus choose Utilities Macro Play Recording 152 Chapter 5 Or m click on the Play Recording tool button Note The Play Recording option can only play the latest recording So a recording will be lost if you start recording another set of commands without saving it Rescuing Sessions The command log records only the commands from your current session You cannot use the command log to recover commands from a previous session unless you saved those commands in a command file before exiting SY STAT However in th
68. data set contains the DIA of 16 adaptor bodies produced over a period of 16 hours one in each hour The total time period 1s divided into two periods of eight hours each and the variable EIGHT takes value 1 or 2 depending upon the period of its production Similarly variables FOUR and TWO are constructed Thus the design is a nested one with four nested inside EIGHT and TWO nested inside FOUR The variables are DIA EIGHT FOUR TWO ADJADAPTOR The data set consists of the outer diameter of a component named adaptor body before and after correction The two variables are BEFORE AFTER ADMIRE Cohen and Brook 1987 In a large scale longitudinal study of childhood and adolescent mental health data were obtained on personal qualities that the subjects admired and what they thought other children admired as well as the sex and age of the subjects The admired qualities were organized into scales for antisocial materialistic and conventional values for the self and as ascribed to others In one phase of the investigation the researchers wanted to study the relationship between the sets of self versus others However several of these scales exhibited sex differences were nonlinearly specifically quadratically related to age and or were differently related to age for the sexes For the self other association to be assessed free of the confounding influence of age sex and their interacti
69. effects Unfold and Slide Select context menu Select the context menu that you want to customize Press Reset to reset any changes you may have made to the selected context menu to the installation default Popup menu Use this to create new popup menus in the Menu Bar Enter the name of a the popup menu and press Create The new menu gets added as the first 1tem in the Menu Bar Drag and drop the menu to whatever location you want it to be in Command File Lists Command files can be saved in any folder If you elect to organize your files by projects each folder will most likely contain data output and command files This approach groups related command files together but may result in similar files appearing in several project folders On the other hand you can store files by type resulting in a single folder containing only command files In either situation finding a particular command file can be a difficult task The Command File List dialog provides a command file classification scheme that 1s independent of your folder structure Using this dialog box you create lists of command files having some element in common such as Charts with Error Bars A list can then be associated with the Submit From File List toolbar button or menu item for immediate processing of any file contained therein To open the Command File List dialog box from the menus choose Utilities User Menu Command File List 22 Customization of
70. from commands by a slash For example CSTATISTICS urban babymort MEAN SEM MEDIAN 130 Chapter 5 The command specifies the task in this case to display statistics The arguments are the names of the variables URBAN and BABYMORT for which statistics will be computed m The options following the slash specify which statistics you want to see If you do not specify any options SYSTAT displays a default set of statistics In general the argument may be one or more variables numbers or strings separated by a space or comma variable lists separated by the asterisk file names folder names a specific keyword that may or may not be equated to a number an expression an equation or an inequality Each option is a keyword that may or may not be equated to an option value the equal sign is compulsory The option value has the same possibilities as the argument Hot versus Cold Commands Some commands execute a task immediately while others do not We call these hot and cold commands respectively Hot commands These commands initiate immediate action For example if you type LIST and press the Enter key SYSTAT lists cases for all variables in the current data file Cold commands These commands set formats or specify conditions For example PAGE WIDE specifies the format for subsequent output but output is not actually produced until you issue further commands Similarly the SAVE command in modules specifie
71. group Click the Fill tab select Select fill from the Fill pattern group and select solid Fill Pattern 91 CALCIUM T Graph Bar Chart Man o Available variable z Options BRANDS FOOD CALORIES Coordinates FAT PROTEIN dig VITAMINA fee CALCIUM Z Axis ie COST All Axes DIET S Error Bars Layout Color SYSTAT Basics arnable s e lt Remove Y sanable DIET Add gt lt Remove vanable s CALCIUM Co unte of Y E Matris columns Display az 3 0 t Grouping anrable s a B i Mirror Dual Remove MultiF lot Overlay multiple graphs into a single frame Stack bars of multiple variables Range between two variables Click OK 50 92 Chapter 3 Suggestion Try using the Dynamic Explorer to rotate this 3 D bar chart The box plot in the two sample t test example shows that the distributions of calcium for the yes and no groups are skewed and have unequal spreads Let us use a root transformation of CALCIUM to make its distribution symmetric Before requesting the analysis of variance we will transform CALCIUM taking the square root of each value E From the menus choose Data Transform Let In the Let dialog box select CALCIUM as the variable select SQR from the list of mathematical functions and select CALCIUM from the variable list and add it to the expression The Expression box should now look like this SQR CALCIUM I Data Transf
72. hospital attendance CNTLATD Survival of control group from first hospital attendance SURVUNTR Survival of patient from time cancer deemed un treatable CNTLUNTR Survival of control from time cancer deemed untreatable LOGSURVA Logarithm of SURVATD LOGCNTLA Logarithm of CNTLAD LOGSURVU Logarithm of SURVUNTR LOGCNTLU Logarithm of CNTLUNTR Clinical trials of this sort are the basis for evaluating the effectiveness of any new drug or medical treatment They are a critical part ofthe FDA approval process in the U S and similar evaluations in virtually all developed countries Potential analyses include descriptive statistics transformations ANOVA and survival analysis 279 Applications Box Plot of Selected Cancer Types The input is USE CANCERDM SELECT ORGANS Breast OR ORGANS Bronchus OR ORGANS Colon OR ORGANS Ovary OR ORGANS Stomach THICK 3 CATEGORY ORGANS BEGIN DEN LOGSURVA ORGANS DOX SIZE 1 2 FILL 1 FCOLOR BLUE COLOR YELLOW YLAB Log Survival XLAB Organ HEI 5IN WID 5IN TITLE Survival by Cancer Type PLOT LOGSURVA ORGANS SMOOTH LOWESS TENSION 0 SIZE 0 COLOR 1 YLAB XLAB HEI 5I1N WID 5IN TITLE END THICK 1 The output is Log Survival Survival by Cancer Type Transformation of Survival Variable The input is USE CANCERDM PPLOT SURVATD 280 Chapter 8 The output is Normal 0 0 1 0 Quantile 3 0 1000 2000 3000 4000 500C SURVATD To p
73. icon selects all entries contained in that folder With the Organizer entry selected copying via the Edit menu or right clicking results in the output corresponding to the selection being copied to the clipboard Select a new entry and paste to insert the copied output at the new location Note that although the Organizer represents an outline of what will be copied from the Output editor the Output editor itself does not show the selection n Transformations Because transformations do not produce output they do not generate Output Organizer entries To note when transformations occur echo the commands or add notes to the output However echoed commands still do not yield an entry in the Organizer To Move Output Organizer Entries You can reorganize SYSTAT s output simply by selecting and dragging Organizer entries to new locations Use the Shift key to select a range of entries or the Ctrl key to select multiple but nonconsecutive entries Selecting a folder entry causes all items within the folder to be selected The Organizer places selected items immediately after and at the same level as the location to which you drag them If you select items at differing levels and drag them to a new location SY STAT places the entries at the level of the target location 191 Working with Output To Insert Tree Folder SYSTAT generates Output Organizer entries for all statistical and graphical procedures You can also create custo
74. include other keys with this Opens SYSTAT executes any commands the user may give and on exit saves all the text output generated during the ses sion into filename xxx Opens SYSTAT executes the command file given with x and saves the output in the MHT format to filename mht Opens SYSTAT performs the actions stip ulated by any other switches specified and quits SYSTAT systat elog c data prompt Error Log dat Systat gexit x c data prompt name4 syc Systat m x c data nameS syc systat out c data prompt testN dat systat x c data prompt name6 syc mht c data prompt outfile6 mh t systat x c data prompt name6 syc mht c data prompt outfile6 mh t q systat x c data prompt name7 syc out c data prompt outfile6 txt q Note In the command file you submit any GSAVE OSAVE and EXPORT commands will save the graph output and data respectively into a filename of your choice which can be later used for further processing by SY STAT or other programs after this session of SYSTAT has quit Environment Variables SY STAT provides environment variables in the STATS module These are variables that contain the computed values of various statistics for a given session a given data file and given variables The following environment variables are available s lt Stalistie gt lt by Group gt lt variable name gt s 155 Command Language where lt statistic gt 1s as follows
75. line unless the IMMEDIATE option is specified This can result in undesirable sequences of prompting dialogs Consider the following set of commands TOKEN amp xvar TYPE VARIABLE TOKEN amp xvarlabel TYPE STRING TOKEN amp yvar TYPE VARIABLE TOKEN amp yvarlabel TYPE STRING PLOT yvar gxvar YLAB amp yvarlabel XLAB xvarlabel First SYSTAT prompts for amp yvar the y variable in the scatterplot Next a prompt for the x variable appears Prompting continues by asking for a label for the y axis and 170 Chapter 5 finally for a label for the x axis Notice that the dialog sequence does not correspond to the order of the TOKEN statements but instead corresponds to the ordering of the actual tokens in the PLOT command Rather than prompting in the order that the tokens are encountered you can define a sequence for dialog prompting using the IMMEDIATE option Instead of prompting when encountering the token the prompting dialog appears when SYSTAT processes the TOKEN statement For example to prompt for the y variable the y axis label the x variable and the x axis label in that order specify the following TOKEN amp yvar TYPE VARIABLE IMMEDIATE TOKEN amp yvarlabel TYPE STRING IMMEDIATE TOKEN amp xvar TYPE VARIABLE IMMEDIATE TOKEN amp xvarlabel TYPE STRING IMMEDIATE PLOT yvar gxvar YLAB amp yvarlabel XLAB amp xvarlabel In this case SY STAT prompts for information in the order of the TOKEN
76. may discover that an alternative window organization would better match the way you work The interface for SYSTAT can be completely restructured to create a comfortable analytical environment in which you can be maximally productive 203 204 Chapter 7 SYSTAT Startpage E ag Ele Edit View Data Utilities Graph Analyze Advanced Quick Access Addons Window Help le x pif i z Workspace 44 SYSTAT Output BLUSE SYSTAT 13 Recent Command Files Recent Output Files To edit the graph format scales thickness etc from the menus choose Edit gt Options Or press F6 to open the Edit Options dialog box Then dick the Graph tab Themes Current Theme Default Classic Default Hi gt Next Tip Introductory_Statistics Market_Research Medical_Statistics Scratchpad MYSTAT 1 Manuals A GettingStarted pdf ra Data pdf Graphics pdf Statistics I II II IV pdf LanguageReference pdf MonteCarlo pdf QualityAnalysis pdf lt IS ExactTests pdf Show at startup 0 Bx For Help press F1 HTM OGRAPH ECHO ID SEL 8 ST FRO CAT OVR CAP NUM You can m resize hide and reorganize windows and panes NW create reposition and modify toolbars NW assign sets of command files to a toolbar button allowing quick submission of commonly used commands add menu items for frequently used commands and command files define settings
77. menu View Page View You can see that you have the capabilities from the Dynamic Explorer rotation animation and zoom available in Page as in Graph view In addition you can position the chart by dragging it around on the page 121 Data Analysis Quick Tour SYSTAT GraphGDI1 a SYSTAT Output TH USE a HE USE imanti ls PLOT URANI Acka Uranium and Kiging Sanco her by Geography Contour Plot of the Kriging Smoother So far we have looked at this data by producing horizon and by latitude and longitude SYSTAT allows us to combine these two pieces of information by tailoring and coloring symbols As a final analysis we will use another advanced graphing technique a contour plot of the kriging smoother This final plot consists of successive vertical slices through the surface of the kriging smoother overlaid on the data coded by producing horizon From the menus submit the file GDWTR3DM m From the menus choose File Submit File E Select GDWTR3DM from the Miscellaneous subfolder of the command directory and click Open 122 Chapter 4 The following graph is displayed Actual Uranium and Kriging Smoother by Geography 33 8 33 7 33 6 33 5 0 O 33 4 HORIZON 7 33 3 Ogalla m Dockum a Quartermaster 33 1 a Whitehorse e El Reno 33 0 100 0 1005 1010 1015 1020 1025 Longitude The plot is simply a different view of the 3 D plot but now we c
78. menus choose Utilities Command Translate Legacy Command Files Alternatively you can right click in an untitled tab of the Commandspace and select Translate Legacy Command Files To translate just some selected commands select the commands in the untitled tab right click on the selection and then click Translate Legacy Command Files 148 Chapter 5 Translate Legacy Commands f Mx C Program Files SYST z IF TRIAL 2 THEN LET YEAR 1937 ORDER SITES SORT Waseca Crookston Morris U ORDER VARIETYS SORT Svansota No 457 Manch SCALE 250 125 kS nT Commands are from O Version 12 Translate _ IF TRIAL 2 THEN LET YEARS 1932 ORDER SITES SORT Waseca Crookston Mor ORDER VARIETY SORT Svansota No 462 M SCALE 250 125 lt 4 Ul gt From file Specify a file to read the legacy commands from The contents of the file are displayed in the box below Command s You can type the legacy commands that you want to translate in this box If you have chosen a file to translate from you can edit the contents shown in the box before you request a translation Commands are from Select the version of your commands command file Translate Press Translate to translate the commands The translated commands are displayed in the box below You can select and copy a part or the whole of the translated commands for pasting to the desired location Save to You should sav
79. name of any other module or global command for example HELP CLUSTER You can also start help by choosing Index from the Help menu and selecting the desired command from the list Yet another alternative is to type the command in any tab of the Commandspace and either clicking on it and pressing Ctrl F1 or right clicking on it and selecting the HELP command Command Files A command file is a text file in Unicode or ANSI format that contains SYSTAT commands Saving your analyses in a command file allows you to repeat them at a later date You can create a command file by selecting the batch Untitled tab in the Commandspace This tab corresponds to a simple text editor type the desired commands line by line When you are done save the commands to a file or submit them to SYSTAT for processing In contrast to the Interactive tab no interactive prompt gt appears on the batch tab commands are not processed until the resulting command file is submitted to SYSTAT 138 Chapter 5 x A XTAB USE ourworld TABULATE LEADERS GROUP MEAN POP_ 1983 XTAB USE OURWORLD TABULATE leader group MEAN pop 1983 If you find any of the SYSTAT examples relevant to your analysis you can open this example command file in the SYSTAT Command folder edit it to suit your data and save it under a different filename You can in fact simultaneously create or open any number of command files copy paste among them edit any of them and s
80. of cars and dogs The variables are CAR DOG C1 C2 D1 D2 CARS The data set reflects the attributes of the selected performance cars The variables are ACCEL BRAKE SLALOM MPG SPEED NAMES CEMENT Birkes and Dodge 1993 The data set consists of four kinds of ingredients INGREDIENTI INGREDIENT2 INGREDIENTS INGREDIENT4 corresponding to the temperature HEAT CHOICE McFadden 1979 The data set consists of hypothetical data The CHOICE variable represents the three transportation alternatives AUTO POOL TRAIN each subject prefers The first subscripted variable in each CHOICE category represents TIME and the second COST Finally SEX represents the gender of the chooser AGE represents the age of the chooser CHOLESTEROL The data set records the age and blood cholesterol levels for two groups of women Women in the first group use contraceptive pills women in the second group do not A PILL value of 1 indicates that the woman takes the pill a value of 2 indicates that she does not Each case has the cholesterol value CHOL for a pill user and for her age matched control AGE CITIES Hartigan 1975 The data set is a dissimilarity matrix consisting of airline distances in hundreds of miles between ten global cities BERLIN BOMBAY CAPETOWN CHICAGO LONDON MONTREAL NEW YORK PARIS SANFRAN and SEATTLE CITYTEMP These data consist of low and high July temperatures for eight U S cities in 1992
81. of the Data Editor to enter or edit comments related to the corresponding data file Simply pause the mouse on the button to view the file comments currently entered for the data file Default Format for Saving Command Files Earlier versions of SYSTAT saved command files in the ANSI format and the previous version saved them in the Unicode format SYSTAT now allows you to specify the format to save command files There is also a setting in the Edit Options dialog box where you may specify the default command file format Drag and Drop Data You may now drag and drop text into SY STAT s Data Editor from editors that support dragging of content This includes dragging and dropping text entered in the Commandspace of SYSTAT itself Embedded Toolbars The Format Bar the Data Edit Bar and the Graph Editing toolbar are now embedded in the Output Editor Data Editor and Graph Editor tabs respectively Open Legacy Command Files You may now directly open and execute legacy command files if a VERSION command is inserted as the first line The syntax is VERSION n where n may be either 11 or 12 Apart from this the Translate Legacy Commands dialog box and the SYSTAT Command Translator also allow you to specify the version whether itis 11 or 12 of the command file you want to translate View Toolbars You may now load one or more of SYSTAT s toolbars through the View menu The entries corresponding to the toolbars that are loaded are prefix
82. path Now there may be occasions where files with the same name exist in both these paths but you specifically need to open one of them For example suppose a file named MYDATA exists in the amp USE path and you issue the following commands USE mydata DSAVE mydata This saves a copy of the data file MY DATA in the default amp SAVE path Suppose a file by name MY DATA also exists in the amp USE path Now if you need to open the original file that is in the amp USE path you will either have to issue the USE command with the full path or USE amp USE mydata Refer Chapter 7 Customization of the SYSTAT environment for details about SYSTAT s file locations Examples The examples presented here illustrate some practical implementations of token substitution For more examples examine the command files used in the Graph Gallery 173 Command Language Example 1 Automatic Substitution in Exploratory Analysis In this example automatic token substitution defines the input file to use SYSTAT then prompts for a variable and creates a bar graph TOKEN amp infile survey2 TOKEN amp catvar TYPE VARIABLE PROMPT Select the variable appearing in the bar graph USE amp infile NONAMES NOTE File in use amp infile CATEGORY amp catvar BAR amp catvar CATEGORY amp catvar OFF The path to the file contains spaces and must therefore be enclosed in quotes when defining the token However the quot
83. percent of students within each school who are eligible to participate in a free meal program PFSM VRA A verbal reasoning ability level from 1 to 3 INCOME The data here were collected from a class of students There are two variables SCORES represents the percent score of students in a statistics test and INCOME the monthly family income in thousand dollars INSTRDMe Huitema 1980 This data set consists of measures of achievement on a biology exam for two groups of students One group was simply told to study everything from a biology text in general and the other was given terms and concepts that they were expected to master An additional covariate the student s aptitude is also included in the data set The variables are STUDENT Student ID INSTRUCTS Type of instruction given INSTRUCT Coded variable for IVSTRUCT APTITUDE Student s underlying ability to learn ACHIEVE Student s score on the exam IRIS Anderson 1935 These data measure sepal length SEPALLEN sepal width SEPALWID petal length PETALLEN and petal width PETAL WID in centimeters for three species SPECIES of irises 1 Setosa 2 Versicolor and 3 Virginica JOHN John 1971 These data are from an incomplete block design with three treatment factors A B and C a blocking variable with eight levels BLOCK and the dependent variable Y JUDGEHILL Judge et al 1988 This data set is obtained on appending data for the two models It c
84. random variable The variable ESTIMATE was estimated from a cubic regression model Finally the variables UPPER and LOWER were computed UPPER corresponds to two standard errors above the estimated value and LOWER corresponds to two standard errors below POWER Ott and Longnecker 2001 The data set consists of deviations from target power POWER using monomers from three different suppliers SUPPLIER with a total number of 27 cases PRENTICE Prentice 1973 This is a survival time data of 137 advanced lung cancer patients The data file contains following variables TRTMNT Two treatments 1 standard 2 test SURVTIME Survival time measured from the start of the treatment for each patient STATUS Censoring status where 1 censored 0 failed TMRTYPE Types of tumor 1 squamous 2 small 3 adeno and 4 large Karnofsky score 1s a performance status assigned to the patient at the time of diagnosis AGE Age of the patient MONTHS Diagnostic period THERAPY Prior therapy status where 0 no prior therapy and 10 with prior therapy KSCORE PROCESS gt Breyfogle 2003 The data set consists of the number of units checked and the number of defects found in 10 operations step in a production process PULPFIBER Lee 1992 The data set contains 62 measurements on the properties of pulp fibers and the paper made from them Four types of pulp fiber characteristics are XI Arithmetic fiber length X2 Long fiber fraction X3 Fine fr
85. randomly from a standard normal distribution TOKEN amp v TYPE STRING PROMPT Enter a name for the new variables Names should be 256 characters long or less NEW DIM amp v 10 REPEAT 100 FOR i 1 TO 10 LET 6 v 1 ZRN NEXT The DIM statement reserves memory for ten subscripted variables assigning a root name supplied by the user REPEAT generates 100 cases The FOR NEXT loop assigns standard normal deviates to each of the ten variables Notice that although we are dealing with variables the VARIABLE type refers to existing variables and thus cannot be used for our purposes namely to create new variables Example 3 Token Substitution for Numbers and Integers The following commands generate a t distribution with a reference line at a specified location The output includes the cumulative area up to and the probability of obtaining a value as extreme as the given value TOKEN amp df TYPE INTEGER PROMPT Enter the degrees of freedom for Che t dIstrIDULCLON TOKEN amp tval TYPE NUMBER PROMPT Enter a t value FPLOT Y TDF t amp df XLIMIT amp tval XLAB t YLAB Density TITLE t Distribution with df DF TEMP tarea TCF amp tval amp df PRINT Area to the left of amp tval tarea IF amp tval gt 0 then TEMP pval 2 l tarea IF amp tval lt 0 then TEMP pval 2 tarea PRINT Two tailed p value DY op ON ele The degrees of freedom for a t distribution must be an int
86. samples claimed to be of two types was sent to each of six commercial laboratories to be analyzed for fat content Each laboratory assigned two technicians who each analyzed both types The variables are FAT LAB TECHNICIAN SAMPLE Fat content as a percentage Lab which ran the experiment Technician code Sample type used EGYPTDMe Thomson and Randall Maciver 1905 This data set consists of four measurements of male Egyptian skulls from five different time periods ranging from 4000 B C to 150 A D The four measurements of male Egyptian skulls are MB Maximal breadth of skull BH Basibregmatic height of skull BL Basialveolar length of skull NH Nasal height of skull YEAR Time of measurement EKMAN Ekman 1954 These data are judged for similarities among 14 different spectral colors The variable names are the colors wavelengths W584 W600 W610 W628 W651 W434 W445 W465 W472 W490 W504 W537 W55 and W674 The judgments are averaged across 31 subjects ELECSORTe This data set is obtained by sorting the data file ELECTION by variable NAMES 329 Data Files EMF The data set consists of counts emfs of patients in urban and suburban areas affected by cancer or not The variables are CANCER EMF RESIDENCE COUNT ENERGY SYSTAT created this file to demonstrate error bars The variable SE determines the length of the error bar ENERGY is determined as low medium and high ENZYMDM Greco et al 1982
87. school AVGPAY Average annual pay for a worker in 1989 TOTALSLE Total sale VIOLRATE Violent crime rate per 100 000 people in 1989 PROPRATE Rate of property crimes per 100 000 people in 1989 PERSON Number of persons who commit crimes POP90 Population in thousands in 1990 as cited in the New York Times ID Name of each state in the United States COUNT Number associated with the state MSTROKE and FSTROKE Risk fsioke pe 100 000 malos an females adjusted to weight INCOMES89 Median household income in 1989 INCOME Income in 1991 BUSH PEROT and CLINTON Vote count in 1000 for each candidate in the 1992 presidential election 397 Data Files Number of electoral votes each state received in the 1992 presi EE KOLE dential election PRES 888 Number of electoral votes each state received in the 1988 presi dential election GOV 93 Newly elected governor s political party in each state after win ning the 1993 gubernatorial races GOV 928 Winning political parties in the 1992 gubernatorial races Census Bureau s estimate of the percentage of Americans living Pee below the poverty level in 1991 POVRTY90 Poverty estimates for 1990 TORNADOS oe of tornados per thousand square miles from 1953 to HIGHTEMP Average high temperature LOWTEMP Average low temperature RAIN Average annual rainfall SUMMER Average summer temperature WINTER Average winter temperature POPDEN Population density LABLON LABLOT GOVSLRY Longi
88. statements rather than in the order that the tokens themselves appear Note SYSTAT always processes MESSAGE tokens first these tokens do not require the IMMEDIATE option Viewing Tokens As you develop your own library of templates it may become useful to have one template file submit another template file However if tokens have the same name in the two files undesired output can result To help correct any token conflicts you can list all current tokens with their defining characteristics by specifying TOKEN LIST You will get a list of predefined tokens as well as user defined tokens For each token SYSTAT displays m the token m the type m the current assigned value E text appearing in the prompting dialog 171 Command Language Generating this listing for each template identifies tokens common to both files Differences should be examined closely two tokens sharing a name but defined as different types are likely to yield odd behavior Predefined tokens SYSTAT has default file locations for opening and saving files which can be set through the File Locations tab of the Edit Options dialog box or the FPATH command When a command like USE filename or SUBMIT filename is executed without an explicit file path SY STAT looks for the file in the corresponding locations The default file locations are assigned to built in tokens as follows Token Name amp EXPORT amp GET amp GSAVE amp IMPORT amp
89. tab in the Scatterplot Matrix dialog box m Select Confidence kernel and enter the value of p as 0 75 75 SYSTAT Basics 4 Graph Scatterplot Matrix SPLOM fx Confidence ellipse Connectors partitions o ti None C Line connected in case order sale Sample El p I C Traveling salesman path me Centroid Elm p Minimum spanning tree Vertical spikes to Y ffs Sener E a we rod MiConlidence kemel p 0 75 2 Bl vector lines from C Convex hull around all points y E Color ue C Influence on correlation coefficient Fil Overlapping data Delaunay triangulation Symbol and Label Points overlap oronoi tessellation Legend Slight random jitter Line Style Sunflower symbols m Click OK CALORIES PROTEIN FAT CALORIES COST CALORIES 76 Chapter 3 For CALORIES and FAT look at the separation of the univariate densities on the diagonal of the display Notice that the price range COST at the bottom right for the diet dinners is within that for the regular dinners COST is the Y variable in the bottom row of plots Within each group COST appears to have little relation to CALORIES or FAT It is possible that COST has a positive association with PROTEIN for the regular dinners open circles in the COST versus PROTEIN plot Is there a relationship between cost and nutritive value as measured by the percentage daily value for vitamin A calcium and ir
90. the Menu Bar simply click on the checkmark to uncheck its name Likewise to display a toolbar check the corresponding name in the list Apart from making use of the 32 built in toolbars you can create your own toolbars Press the New button enter the desired name and press OK The toolbar appears in front of the dialog Drag it to the desired location or leave it floating in front of the interface Drag and drop the desired menu menu items or toolbar buttons from other toolbars or the Commands list in the Commands tab into the new toolbar Toreset any toolbar to its default state select its name in the Toolbars list and press the Reset button To reset all toolbars just press the Reset All button m To rename or delete a toolbar that you have created press the Rename or Delete buttons respectively The Toolbars tab also offers optional button appearance features Show tooltips Displays the button name when the mouse pauses on a button 220 Chapter 7 With shortcut keys Displays the shortcut key sequence to be pressed to invoke the same feature along with the button tooltip Keyboard Shortcuts Although SYSTAT runs in a Windows environment many users find manipulating the mouse to be an annoyance Fortunately for these users every menu item can be accessed using the keyboard The F10 key activates the File menu Once activated use the arrow keys to navigate through the menu system The up and down arrows s
91. the preceding period Other variables include number assigned to the cow COW and the Latin square number SQUARE WILLMSDM Hubert 1984 This data set contains the results of a bioassay conducted to determine the concentration of nicotine sulfate required to kill 50 of a group of common fruit flies The experimenters recorded the number of fruit flies that are killed at different dosage levels The variables are The dependent variable which is the response of the fruit fly to the dose of Ree nicotine sulfate stimulus LDOSE The logarithm of the dose COUNT The number of fruit flies with that response WINER Winer 1971 The data are from a design with two trials DAY 2 one covariate AGE and one grouping factor SEX WORDS gt Caroll Davies and Richmond 1971 The data set contains the most frequently used words WORDS in American English Three measures have been added to the data The first is the most likely part of speech PART The second is the number of letters LETTERS in the word The third is a measure of the meaning WEANING This admittedly informal measure represents the amount of harm done to comprehension 1 a little 4 a lot by omitting the word from a sentence WORLD Global mapping The variables include MAPNUM MAXLAT MINLAT MINLON MAXLON LABLAT LABLON and COLORS WORLD95M For each of 109 countries 22 variables were culled from several 1995 almanacs including life expectancy
92. this artificial data set in an analysis of covariance in which Y is the dependent variable X is the covariate and TREAT is the treatment COVSTRUCTe It is a hypothetical data The variables are P O Y COX Cox 1970 These data record tests for failures among objects after certain times TIME FAILURE is the number of failures and COUNT is the total number of tests CRABS Wilkinson 2005 These data record the location of 23 fiddler crab holes in an 80 x 80 centimeter area of the Pamet River marsh in Truro Massachusetts The variables are CRAB X Y 326 Chapter 9 CRIMERW Clausen 1998 These data show the information case by case about crimes in three different areas in Norway The following is a list of the three different areas and three crimes The SYSTAT names are within parentheses PLACES CRIMES Mid Norrway Mid N Burglary North Norway NorthN Fraud Oslo Area Oslo Vandalism CRIMESTAT FBI Uniform Crime Reports 1985 The data set consists of arrests by sex for selected crimes in United States in 1985 The variables are CRIME MALES FEMALES CROPS Milliken and Johnson 1984 It is an agricultural data consists of yields in pounds YIELD of two varieties of wheat VARIETY grown in four different fertility regimes FERT To compare four fertilizers and two varieties of crops four whole plots were grouped into two blocks BLOCK The two varieties were assigned randomly to the two whole plots in each
93. three drugs DRUG with weight loss measured in grams for the first and second weeks of the experiment WEEK 1 and WEEK 2 SEX was another factor MELNMADMe Wilkinson and Engelman 1996 This data set contains reports on melanoma patients The variables are TIME The survival time for melanoma patients in days CENSOR The censoring variable WEIGHT The weight variable ULCER Presence or absence of ulcers DEPTH Depth of ulceration NODES Number of lymph nodes that are affected SEXS The sex of the patient SEX The stratification variable coded for analysis METOX Fellner 1986 The data set is about metallic oxide analysis where two types of metallic oxides eighteen lots from the first type and thirteen from the second were used Two samples were drawn from each lot A pair of chemists was randomly selected for each sample The variables are TYPE SAMPLE CHEMIST and Y MILKe Brownlee 1960 The data set pertains to bacteriological testing of milk Twelve milk samples SAMPLE were tested in all six combinations of two types of bottles BOTTLES and 339 Data Files three types of tubes TUBE Ten tests were run on each combination and the response was the number of positive tests in each set of ten Y MINIWRLD gt This data file is a subset of OUR WORLD MINTEMP Barnett and Lewis 1967 The data set consists of a variable TEMP that is annual minimum temperature F of Plymouth in Britain for 49 years MISSLES
94. use at just a click of the mouse 25 Introducing SYSTAT Variable editor Each data file active or inactive has a Data tab and a Variable tab The Data tab allows you to edit data values directly in the grid that you see by default The Variable tab allows you to edit the properties of variables directly We will henceforth refer to the Variable tab as the Variable editor The Variable editor has one row corresponding to each variable and the row includes all the items that are in the Variable Properties dialog With it you can m Set any of the properties for any variable with a single click of the mouse m View and set the processing conditions in effect for the current data set viz information regarding frequency weight category and grouping variables defined if any and any case selection conditions You can navigate to any specified column or row of the Data editor and veiw edit the value stored in any cell using the Data Edit bar that 1s embedded in the Data editor See Chapter 3 Entering and Editing Data of the SYSTAT Data volume for more information about the Data editor Graph editor Double clicking a graph in the Output editor or just clicking the Graph tab after drawing a graph opens the Graph editor SYSTAT Graph1 i File Edit View Data Utilities Graph Analyze Advanced Quick Access Addons Window Help y SYSTAT Examples ma payn 5 i W Applications Gallery 19 od a aE a RT iso ES y S P Demonstratio
95. ways of coping with stressful events of 275 college undergraduates The variables are PI P3 Problem solving Cl C3 Cognitive restructuring E1 E3 Express Emotions Sl S3 Social Support SUBWORLD The data in the file SUBWORLD are a subset of cases and variables from the OURWORLD file 352 Chapter 9 SUBWRLD 2 The dataset is a transformation of SUBWORLD data set The variables are standardized and sorted in descending GDP CAP order and transformed them to log base 10 units to symmetrize the distributions before they are standardized only cases with values for all the variables have been included SUB_OURWORLD It s a subset of data set OURWORLD in SYSTAT The variables are CTEDUC Expenditure in US dollars per person for education in the city CTHEALTH Expenditure in US dollars per person for health in the city RUEDUC Expenditure in US dollars per person for education in rural area RUHEALTH Expenditure in US dollars per person for health in rural area SUNSPTDMe Andrews and Herzberg 1985 The data set consists of a calculated relative measure of the daily number of sunspots compiled from the observations of a number of different observatories YEAR The year the observations JAN DEC The relative measure of sunspots for the indicated month ANNUAL The mean relative measure of sunspots for the entire year SURVEY In Los Angeles circa 1980 interviewers from the Institute for Social Science Research at UCLA surve
96. week patients received a score on the Hamilton depression rating scale A diagnosis of endogenous or non endogenous depression was made for each patient Although the total number of subjects in this study was 66 the number of subjects with all measures at each of the weeks fluctuated 61 at week 0 start of placebo week 63 at week 1 end of placebo week 65 at week 2 end of first drug treatment week 65 at week 3 end of second drug treatment week 63 at week 4 end of third drug treatment week and 58 at week 5 end of fourth drug treatment week The variables are ID HAMD CONSTANT WEEK ENDOG ENDOGWK RLONGLEY Longley 1967 The data were originally used to test the robustness of least squares packages to multicollinearity and other sources of ill conditioning The variables in his data set are TOTAL DEFLATOR GNP UNEMPLOY ARMFORCE POPULATN and TIME 349 Data Files ROCKET Components 4 B and C are mixed to form a rocket propellant The elasticity of the propellant ELASTIC was the dependent variable The other variable is RUN ROHWER Timm 2002 The data set is based on the performance of 32 kindergartens in three standardized tests peabody picture vocabulary test PPVT Raven progressive matrices test RPMT and a student achievement test S47 The independent variables are named N still S named still VS named action VA sentence still SS ROTATE Metzler and Shepard 1974 These data measu
97. while quitting SYSTAT By default SYSTAT prompts you to save all open documents including any new unsaved data and commands that you may have entered when you quit the application You may want to uncheck this option when you run the application unattended in the batch mode 238 Chapter 7 Output Options The Output tab of the Global Options dialog determines the format and content of subsequently created output Edit Options General Numeric display format Data Field width Decimal places Exponential notation Out t Locale PE System default wt Sample number ADS x 123456789 00 C Suppress digit grouping Output results Wrap text in tables at characters Length Truncate text in tables at characters Display statistical Quick Graphs Graph File Locations width Default fort J Echo commands in output Proportional output Use SYSTAT classic output style Monospaced output Variable label display Label Value label display Label Image format PNG Numeric display format These settings control the default display of numeric data in the output Field width is the total number of digits in the data value including decimal places Exponential notation is used to display very small values This is particularly useful for data values that might otherwise appear as 0 in the chosen data format For example a value of 0 00001 is displayed as 0 000 in the default 12 3 format but is displayed a
98. will scroll The Scroll Lock should be off if you want to use the arrow keys for navigation around the Data Editor Status Bar Customization Of the status bar items mentioned above the QGRAPH HTM ECHO SEL BY WGT FRQ ID CAT and OVR items appear by default You can add or remove items from the status bar by right clicking on it In the context menu that appears check the items you want to keep and uncheck the items you do not use You can get all the items 2 12 Chapter 7 to appear by selecting All Items all the 1tems will disappear 1f you select No Items To revert to the default set of items select Default Items If you simply do not need the status bar or need more area available for a window from the menus choose View Status Bar Repeat the above steps to bring back the status bar Customizing Menus and Toolbars in SYSTAT Menu Customization SYSTAT has a default organization for the menus and toolbars based on similarity of features However users can customize these according to their needs and preferences using the Customize dialog box To open the Customize dialog box from the menus choose View Customize Alternatively right click in the Toolbar area and select Customize The four tabs in the Customize dialog box can be used to customize menus including right click or context menus toolbars and keyboard shortcuts A context menu is also available to customize menu items and toolbar butt
99. 0 YEAR FLEA Lubischew 1962 The data set consists of measurements on the following four variables on two species SPECIES of flea beetles XI X2 X3 X4 Distance of the transverse groove to the posterior border of the paradox in microns Length of the elytra in mm Length of the second antennal point in microns Length of the third antennal joint in microns FLEABEETLE Hand et al 1996 Data were collected on the genus of flea beetle Chaetocnema which contains three species SPECIES concinna Con heikertingeri Hei and heptapotamica Hep Measurements were made on the width and angle of the aedeagus of 74 beetles The goal of the original study was to form a classification rule to distinguish the three species The data set consists of only measurements of angle of aedeagus of beetles The variables are ANGLE SPECIESS FOOD These data were gathered from food labels at a grocery store The variables are BRAND Shortened name for brand FOOD Type of dinner chicken pasta or beef CALORIES Calories per serving FAT Grams of fat PROTEIN Grams of protein VITAMINA CALCIUM IRON Percentage of daily value of vitamin A calcium and iron COST Price per dinner DIET Yes 1f low in calories no if standard FORBES Bringham 1980 The data are various characteristics of financial performance in chemical companies reported by 30 largest companies The variables are PE RATIO Price to earning ratio which i
100. 0 THEN LET x 5 New Features 41 42 Graphics Locales and Digit Grouping You may now select the locale that SYSTAT should use while displaying numbers in the Output Editor SYSTAT also determines the format of the number s you type in the Data Editor from this setting That means you can now type numbers using the decimal and digit grouping symbols of the selected locale The default locale corresponding to the entry System default is determined from the Regional and Language Settings in the Windows Control Panel Node and Link Captions You may now set Output Organizer node and collapsible link captions using the NODE command Run HELP NODE to know the command syntax for accessing this new feature New Features 43 44 45 Color using RGB Values SYSTAT now offers you the option of specifying colors in terms of their Red Green Blue component values This is available for specifying the color of elements axes and frame colors Gradient colors for surfaces through the dialog box SYSTAT now allows you to specify the gradient style for surfaces through the dialog box This is available in the Surface and Line Style tab of the dialog boxes for the relevant graph types Label Dots in Dot Summary Charts SYSTAT now offers the option of labeling dots in dot Summary charts 15 What s New and Different in SYSTAT 153 Modified Features 46 47 48 49 50 51 Built In Colors SYST
101. 00 000 100 000 23 2250 000 33312 101 978 TPO 1 000 15 780 15 500 1 500 000 75 000 20 000 7 900 200 000 115 000 24 2253 000 33 271 101 805 TPO 1 000 21190 10 700 2 000 000 100 000 200 000 0 700 200 000 435 000 25 2255 000 331687 101 450 TPO 1 000 13160 18 200 559 000 33 000 9 000 0 600 72 000 115 000 26 2256 000 33152 101 798 TPO 1 000 12 330 7 500 744 000 372 000 12 000 5 100 124 000 325 000 27 2258 000 33 068 101 859 TPO 1 000 5 730 6 100 500 000 100 000 5 000 0 400 75 000 60 000 a IIAN NNN 22172 4n1 277 TPN 4 nan 11 a7n A TAN 1 NA Ana 18N NAN an nan nann NN NNN A16 NNN vi ET CeT a At A Define properties for selected variable s GRAPH HTM ECHO SEL BY WGT FRO ID CAT OVR CAP NUM SCRL 103 Data Analysis Quick Tour Graphics Distribution Plot Since we will be looking extensively at uranium levels it is a good idea to take a look at the distribution of this variable and make sure 1t meets assumptions for future analyses To plot a histogram of URANIUM m Click the Histogram icon E in the Graph Toolbars m Choose URANIUM and add it to the X variable s list m Click OK SYSTAT displays the following plot in the Graph editor 80 0 6 70 0 5 60 0 42 z S 3 0 35 O 30 w 0 29 20 0 1 10 0 0 0 0 50 100 150 URANIUM We can see that the distribution of URANIUM is skewed To properly apply most statistical analyses the histogram should show a bell shaped normal dist
102. 1 Manly B F J 1986 Multivariate statistical methods New York Chapman amp Hall STATLIB http lib stat cmu edu DASL Datafiles EgyptianSkulls html Astronomy Data Source Original Source Waldmeir M 1961 The sunspot activity in the years 1610 1960 Zurich Schulthess and International Astronomical Union Quarterly Bulletin on Solar Activity Tokyo Data Reference Andrews D F and Herzberg A M 1985 Data pp 67 76 Springer Verlag 313 Applications Biology Data Source Data Source Carey J R Liedo P Orozco D and Vaupel J W 1992 Slowing of mortality rates at older ages in large med fly cohorts Science pp 258 457 461 Data Reference STATLIB http lib Stat cmu edu DASL Datafiles Medflies html Data Source Allison T and Cicchetti D V 1976 Sleep in mammals Ecological and constitutional correlates Science pp 194 732 734 Chemistry Data Sources Original Source Adapted from a conference session on statistical computing Greco et al 1982 Data Reference Wilkinson L and Engelman L 1996 SYSTAT 6 0 for Windows Statistics pp 487 488 SPSS Inc Engineering Reference Devor R E Chang T and Sutherland J W 1992 Statistical quality design and control pp 756 761 New York MacMillan Environmental Science Sources Original Source Lange Royals and Connor 1993 Transactions of the American fisheries society Data Reference STATLIB http lib Stat cmu ed
103. 1 0 873 0 075 4 0 154 1 134 0 845 0 029 5 i 0 014 1 260 0 847 0 027 6 0 014 L259 0 847 0 027 7 0 014 1 260 0 847 0 027 8 0 014 1 260 0 847 0 027 Dependent Variable VELOCITY Sum of Squares and Mean Squares Source i SS df Mean Squares A A A A A A Regression 1 15 404 3 5 135 Residual 0 014 43 0 000 Total 15 418 46 Mean corrected 5 763 45 R squares Raw R square 1 Residual Total 0 099 Mean Corrected R square 1 Residual Corrected 0 998 R square Observed vs Predicted 0 998 Parameter Estimates Wald 95 Confidence Interval Parameter Estimate ASE Parameter ASE Lower Upper A a areas 4 ee ee ee rr oe ee oo ee re eee VMAX 1 260 0 012 104 191 PELAS 1 284 KM i 0 847 0 027 314 8716 05793 0 900 KIS i 0 027 0 001 31033 0 025 0 029 260 Chapter 8 12 gt 0 8 E O O Lu gt 04 o 9 Y QS Zo Co a NV 2 e C oi con DWLS Smoother The input is USE ENZYMDM THICK 1 7 BEGIN PLOT VELOCITY INH CONC SUB CONC SIZE 0 SMOOTH DWLS TENSION 0 500 TITLE XLABEL YLABEL ZLABEL AXES CORNER ACOLOR BLACK YGRID ZGRID FCOLOR gray ZMAX 1 1 HE IGHT 3 75 WIDTH 3 75 ALTITUDE 3 75 FACET XY PLOT VELOCITY INH CONC SUB CONC SIZE 0 SMOOTH DWLS TENSION 0 500 TITLE XLABEL YLABEL ZLABEL AXES no SC no legend no FCOLOR white ZMAX 1 1 tile HEIGHT 3 75 WIDTH 3 75 ALTITUDE 3 75 FACET PLOT VELOCITY INH CONC SUB CONC SIZE 0 SMO
104. 2 000 14 000 13 000 19 000 15 000 19 EGermany 16 700 16 700 16 307 16 800 76 000 14 000 12 000 14 000 12 000 10 700 7 000 20 Czecho 15 400 15 500 15 683 17 300 67 000 15 000 14 000 12 000 11 000 15 600 11 000 21 a 0 700 0 800 0 848 1 700 18 000 49 000 48 000 28 000 18 000 193 000 140 000 Mi 4a HIS gt Each row is a case and each column is a variable You can type new data into an empty Data editor or you can edit and transform data m To define a variable right click on a column and choose Variable Properties This opens the Variable Properties dialog box and allows you to name the variable supply a label for it select the variable type indicate whether it is categorical set display options and specify comments m Use the Edit menu to cut copy delete and paste rows columns and blocks of data m Use the Data menu to transform data and select subsets of cases The data file that you create or open for use is called the active data file You can open any number of data files using the File menu a new tab 1s created in the Data editor for each file that you open The currently active file automatically goes into the view mode when you create or open another file You need to make it active only if you want to perform any data transformation or analyses on it You can make a data file active using its context menu or the Output Organizer You can thus have any number of data files available in the Data editor ready for
105. 3 toolbar area 34 variable editor 33 correlation 69 crosstabulation 64 CTRL key 220 Customize dialog 30 Commands tab 213 Keyboard tab 224 Toolbars tab 218 customizing menus and toolbars 212 D data 243 entering 47 data editor 24 30 cell entry 217 context menu 33 first case 217 Invert Case Selection 217 last case 217 next case 217 previous case 217 data files 24 active 24 viewing 24 Data folder 243 Data menu 31 Descriptive Statistics 66 dialog boxes 35 additional features 37 check boxes 37 command pushbuttons 35 command templates 158 edit texts 37 pushbuttons 36 radio buttons 37 right click 38 selecting variables 37 source variable list 36 special lists 36 tabs 35 target variable list s 36 directories file locations 243 DOS commands 144 153 errors 153 graphs 153 mht 154 minimized 154 opening 153 output 154 quitting 154 saving 154 submitting 153 switches 153 drag and drop 212 213 219 Dynamic Explorer 27 dynamic explorer 92 E ECHO 210 echo commands 240 Edit menu 30 Data Editor 30 Find 30 Graph Editor 31 output editor 30 Output Organizer 31 Redo 30 Replace 31 Undo 30 Index EMF 196 encapsulated postscript files 196 entering data 47 EPS 196 Examples 27 Examples tab 34 206 Collapse All 34 context menu 34 customizing 206 Expand All 34 ini file 208 opening commnad files 34 run 34 Exc
106. 4 000 0 604 0 080 22 000 1 000 645 000 0 577 0 081 21 000 1 000 659 000 0 549 0 081 20 000 1 000 749 000 0 522 0 082 18 000 1 000 803 000 0 493 0 082 16 000 1 000 1020 000 0 462 0 083 15 000 1 000 1042 000 0 431 0 083 Group size 38 000 Number Failing 21 000 Product Limit Likelihood 89 404 Mean Survival Time Mean 95 0 Confidence Interval Survival Time Lower Upper 3404 857 2282 604 4527 110 Survival Quantiles Survival 95 0 Confidence Interval Probability Time Lower Upper 0 250 5 0 500 803 000 465 000 0750 362 000 142 000 584 000 Survival Plot po 20 D gt L O 2 2 gt K M Probability Lower Limit Upper Limit o 0 0 A 1 0 4 1000 2000 3000 4000 5000 6000 7000 8000 Time Log Rank Test Stratification on SEX Strata Range 1 to 2 Chi Square gt Statistic Method With 1 df A es ce eee i is 4 Mantel 0 568 Breslow Gehan 1 589 Tarone Ware 1 167 OOOO 0O0 0 0 0 0 sold 459 431 lt 3 18 333 297 209 OoOo0o0o0o0o0o000O 807 1103 LSO JLS cool 066 612 584 287 Applications Weibull Estimation The input is USE MELNMADM SURVIVAL MODEL TIME ULCER DEPTH NODES CENSOR CENSOR ESTIMATE EWB ONTL The output is Time Variable TIME Censor Variable CENSOR Input Records z 09 Records Kept for Analysis 69 Censoring Observations A a e a A A a 4 Exact Failures 36 Right Censored 33 Covariate
107. 50 Stouffer 370 Gourmet 440 Tyson 330 Swanson 300 Fat 5 6 3 19 26 14 12 Viewing entering and editing data occurs in the Data editor To open the Data editor either choose Data editor from View menu or click on the Data editor tab Untitled1 syz in the Viewspace 48 Chapter 3 SYSTAT Unti E SYSTAT Output HE USE on nn amp WN E For Help press F1 Open the Variable Properties dialog box either from the menu Data gt Variable Properties or by right clicking on first column 49 Variable Properties Varable name WAR 1 Variable label Warnable type Display options Numeric O String Numeric display options Categorical Characters Decimal places O Exponential notation Date and time Comments Save changes while navigating SYSTAT Basics m Type BRANDS for the variable name The dollar sign at the end of the variable name indicates that the variable contains character information Note Variable names cannot exceed 256 characters In the Variable label edit box you can type the alias for the variable name Select String as the Variable type Choose 15 from the drop down list width edit box Click OK to complete the variable definition Repeat this process for the remaining variables selecting Numeric as the variable type Note In Numeric display options the default decimal places are 3
108. 54 0 424 Canonical Loadings Correlations between Conditional Dependent Variables and Dependent Canonical Factors 1 2 E E No EE ek DINNER 0 068 0 918 STRANGER 0 852 0 404 PROBLEM 0 736 0 037 Information Criteria AIC 1878 445 AIC Corrected 1893 445 Schwarz s BIC LODOS 296 Chapter 8 Scatterplot Matrix SPLOM The input is USE DAYCREDM LABEL SETTING 1 Parent 2 Sitter 3 Center SPLOM DINNER STRANGER PROBLEM GROUP SETTING DEN NORM ELL DASH 1 7 10 COLOR 13 1 2 FILL SYMBOL 11 1 8 OVERLAY TITLE Social Competence Measures Across Settings The output is Social Competence Measures Across Settings DINNER STRANGER PROBLEM DINNER Y3NNI0O STRANGER YAONVALS SETTING Parent Sitter Center PROBLEM N3I1IOYA DINNER STRANGER PROBLEM A scatterplot matrix can be used to check the assumptions of MANOVA 1 e that the variance and covariances are homogeneous across settings From the SPLOM there does not seem to be any systematic violations of the assumptions which might require a variable transformation 297 Applications Analysis of Fear Symptoms of U S Soldiers using Item Response Theory COMBATDM data contains reports of fear symptoms by selected U S soldiers after being withdrawn from World War II combat There are nine symptoms that are included for analysis and the number of soldiers in each profile of symptom is reported Variable C
109. 722 926 PROBLEM 230259 126 149554 411 33741 074 Univariate F Tests Source i Type III SS df Mean Squares F ratio p a DINNER iY ASEOS SOL 0 L 75105991 386 261 257 Error 12936578 626 45 287479 525 STRANGER 20910836 774 gi 20910836 774 245 450 Error 3833722 92 6 45 85193 843 PROBLEM 117347 118 1 117347 118 156 504 EXPO 33 741 2074 45 749 802 Applications 294 Chapter 8 Multivariate Test Statistics Statistic i Value F ratio df p value Se A Soe Si ee u es eee Se ee Wilks s Lambda GeO 128 489 Sp 43 0 000 Pillai Trace i 0 900 128 489 3 43 0 000 Hotelling Lawley Trace 8 964 128 489 Sy 43 0 000 Test of Residual Roots Chi square df 1 through 1 102 306 3 Canonical Correlations 0 948 Dependent Variable Canonical Coefficients Standardized by Conditional within Groups Standard Deviations DINNER MES TA STRANGER 0 523 PROBLEM 0 204 Canonical Loadings Correlations between Conditional Dependent Variables and Dependent Canonical Factors DINNER i 0 805 STRANGER 0 780 PROBLEM i 0 623 Information Criteria AIC 1878 445 AIC Corrected 1893 445 Schwarz s BIC CL A OO Oa 5S Test for effect called SETTING Null Hypothesis Contrast AB DINNER STRANGER PROBLEM 1 166 479 62 116 2 207 2 Ap 109 908 126 189 LAO Inverse Contrast A X X 1A 1 2 WS A A a a a a o e a Dd Oion 2 0 028 0 056 Hypothesis Sum of Product Matrix H B A A X X 1A 1AB DI
110. 957 Experimental designs New York John Wiley amp Sons Cohen J 1988 Set correlation and contingency tables Applied Psychological Measurement 12 425 434 Cohen P and Brook J 1987 Family factors related to the persistence of psychopathology in childhood and adolescence Psychiatry 50 332 345 Conover W J 1999 Practical nonparametric statistics 3rd ed New York John Wiley amp Sons pp 371 373 Cook R D and Weisberg S 1990 Confidence curves in nonlinear regression Journal of The American Statistical Association 85 544 551 Cornell J A 1985 Mixture Experiments In Koltz S and Johnson N L Eds Encyclopedia of Statistical Sciences Vol 5 569 579 New York John Wiley amp Sons Cox D R 1970 The analysis of binary data New York Halsted Press Crowder M J and Hand D J 1990 Analysis of repeated measures London Chapman amp Hall DASL 2005 Available at http lib stat cmu edu DASL Stories SteppingandHeartRates html Databook 2005 Available at http stats unipune ernet in Databook DatasetsPUNE Waterquality xls Davis D J 1977 An analysis of some failure data Journal of the American Statistical Association 72 113 150 Devor R E Chang T Sutherland J W 1992 Statistical Quality Design and Control New York MacMillan Draper N R and Smith H 1998 Applied regression analysis 3rd ed New York John Wiley amp Sons Duncan O D
111. ACHER Timm 2002 The data set was obtained at the University of Pittsburgh by J Raffaele to analyze the reading comprehension and reading rate of students The teachers were nested within classes The classes were noncontract and contract classes The variables are CLASSES Types of classes TEACHERS Teachers READRATE Reading rate READCOMPRE Reading comprehension TETRA These data are from a bivariate normal distribution Variables include X Y and COUNT frequency THREAD Taguchi et al 1989 The data set consists of the tensile strength STRENGTH in kilograms per millimeter squared of thread samples collected every day for two months MONTH of production TRANSAMSTERDAM Franses and Dick van Dijk 2000 The data utilized the index of the stock markets in Amsterdam EOE The exchange rate is Dutch guilder The sample period for the stock index runs from January 6 1986 until December 31 1997 The original series is sampled 5 days in a week The variables are AMSTEOE Daily indices of stock data of Amsterdam in Netherlands There is 5 days in a week opening date 1 06 1986 ending date 12 31 1997 TRAMSTOCK Simple difference transforms series of AMSTEOE TIME Time is sample case number TRIAL These data contain six variables X 1 X 5 and SEX 355 Data Files TVFSP Hedeker and Gibbons 1996 The data set is from the Television School and Family Smoking Prevention and Cessation Project Hedeker and
112. AT now provides 45 built in colors as against the 12 available in previous versions Colors for overlaid graphs pie and stacked charts Overlaid graphs pie charts and stacked bar charts will now be colored in such a way as to provide more contrast between adjacent elements Stacked Bar Charts with Grouping Variable You may now stack bars in the case of grouped bar charts as well A stacked chart is drawn for each group and all the charts are laid out in the same frame Individual Border Displays on Plots SYSTAT now provides options to separately specify the border displays for individual borders This allows you to suppress the display along any given border or specify different kinds of border displays along the two borders in all two dimensional plots Multiple Slices in Pie Charts You may now request separating multiple slices from a pie chart Request specified slice numbers or all slices Numeric Case Labels SYSTAT now allows you to specify a numeric variable for setting labels in plots multivariate displays and maps In prior versions you could only use string variables for labeling elements 16 Chapter 1 Statistical Features New Features 1 ARCH and GARCH Models in Time Series As part of 1ts Time Series feature update SYSTAT now offers m Fitting of ARCH and GARCH models through BHHH BFGS and Newton Raphson implementations of the maximum likelihood method Various options for setting convergence criteri
113. Academic Press Simonoff J S 2003 Analyzing categorical data New York Springer Verlag Smith G M 2001 Statistical process control and quality improvement Upper Saddle River NJ Prentice Hall p 474 Stouffer S A Guttmann L Suchman E A Lazarsfeld P F Staf S A and Clausen J A 1950 Measurement and prediction Princeton N J Princeton University Press Strahan R and Gerbasi K C 1972 Short homogeneous versions of the Crowne Marlowe social desirability scale Journal of Clinical Psychology 28 191 193 Swets J A and Pickett R M 1982 Evaluation of diagnostic systems New York Academic Press Swets J A Tanner W P and Birdsall T G 1961 Decision processes in perception Psychological Review 68 301 340 366 Chapter 9 Taguchi G El Sayed E A and Hslang T 1989 Quality engineering in production systems New York McGraw Hill pp 32 41 The Open University 1981 S237 The Earth Structure composition and evolution Thomson A and Randall Maciver R 1905 Ancient Races of the Thebaid Oxford Oxford University Press Timm N H 2002 Applied multivariate analysis New Y ork Springer Verlag Walser P 1969 Untersuchung ber die Verteilung der Gerburtstermine bei dermehrgebarenden Frau Helvetica Paediatrica Acta Suppl XX ad vol 42 fasc 3 1 30 Wheaton B Muthen B Alwin D F and Summers G F 1977 Assessing reliability and s
114. BABYMT82 BABYMORT LIFI 1 3 400 3 600 3 500 5 000 56 000 20 000 15 000 9 000 9 000 10 500 6 000 2 Austria 7 500 7 600 7 644 7 300 55 000 12 000 12 000 12 000 11 000 12 000 6 000 3 3 Belgium 10 200 9 900 9 909 9 500 l 12 000 12 000 11 000 11 000 11 300 6 000 E 4 Denmark 5 100 5 100 5 131 4 600 83 000 10 000 12 000 11 000 11 000 8 200 6 000 5 Finland 4 800 4 900 4 977 4 900 60 000 14 000 13 000 9 000 10 000 6 000 6 000 6 France 54 400 55 400 56 358 59 300 73 000 15 000 14 000 10 000 9 000 9 000 6 000 7 Greece 9 900 10 000 10 028 11 600 65 000 14 000 11 000 9 000 9 000 14 900 10 000 8 Switzer 6 500 6 500 6 742 5 900 58 000 12 000 12 000 9 000 9 000 7 700 5 000 9 Spain 38 200 38 800 39 269 43 400 91 000 13 000 11 000 7 000 8 000 9 600 6 000 10 UK 56 300 56 600 57 366 56 600 76 000 13 000 14 000 12 000 11 000 10 100 7 000 11 Italy 57 500 57 200 57 664 54 800 69 000 11 000 10 000 10 000 9 000 12 400 6 000 12 Sweden 8 300 8 400 8 526 7 400 83 000 11 000 13 000 11 000 11 000 7 000 6 000 13 Portugal 10 100 10 100 10 354 12 100 30 000 16 000 12 000 10 000 10 000 19 800 14 000 14 Netherl 14 400 14 500 14 936 13 900 88 000 12 000 13 000 8 000 8 000 8 400 7 000 15 WGermany 61 400 60 700 62 168 51 800 94 000 10 000 11 000 12 000 11 000 10 100 6 000 16 Norway 4 100 4 200 4 253 4 100 70 000 12 000 14 000 10 000 11 000 7 800 7 000 17 Poland 36 600 37 500 37 777 44 700 59 000 19 000 14 000 9 000 9 000 19 300 13 000 18 Hungary 10 700 10 600 10 569 10 500 54 000 12 000 1
115. C 1996 Desktop data analysis with SYSTAT Upper Saddle River NJ Prentice Hall p 738 Statistics Data Sources Original Source Huitema B E 1980 The Analysis of covariance and alternatives New York John Wiley amp Sons Data Reference Wilkinson L Blank G and Gruber C 1996 Desktop data analysis with SYSTAT Upper Saddle River NJ Prentice Hall p 442 Toxicology Data Source Hubert J J 1991 Bioassay 3rd ed Dubuque Iowa Kendall Hunt Appendix 9 Data Files SYSTAT software comes with a folder of data files which can be accessed through the File gt Open gt Data dialog The folder contains over 350 files of data used in the nearly 600 examples provided in the user manual and online help This Appendix gives details of these files with sources of data a brief description of the study which generated the data and a description of the variables in the file These data files not only contain the data but also a great deal of information on the data file The information given in this Appendix is available in the data file itself When you have clicked on the data file name in the dialog and opened it in the Data editor by hovering the mouse over the corner rectangle the top left cell you will see the general information on the file Then in the Variable Properties dialog ofa variable which can be opened by Data gt Variable Properties with the variable name selected by clicking on it or by sim
116. CLINCOV Hocking 2003 This example is based on a clinical data set where a pharmaceutical firm wants to test a new drug for a particular disease The response is a measure of the improvement in the patients status A sample consisting of three clinics CLINIC is selected at random from a large population of clinics From each clinic a sample of ten patients with 324 Chapter 9 the particular disease are selected The drug is applied to each patient and we record the response Y of the drug as well as a relevant physical characteristic Z for each patient CLOTH Montgomery 2005 Here the occurrences of nonconformities DEFECTS in each of 10 rolls of dyed cloth were counted ROLL The rolls were not all the same size in square meters Thus the sample unit was defined as 50 square meters of cloth and roll sizes were expressed in these units UNITS COBDOUG Judge et al 1988 The data set is related to the Cobb Douglas production function in Econometrics The Cobb Douglas Production function considers the effect of Labor L and Capital invested K over the output Q The data set consists of 20 observations containing the variables Y XI and X2 where we have Y n0 and X InL and X2 nK CODDER These data contain the percentage of reader attention PERCENT in a certain geographical area LOCUS for the local newspaper COFFEE Hand et al 1996 The data set contains the prices in pence of a 100gm pack of a particula
117. Coefficients B X X xXx Y Factor MB BH BL NH anm emni i i aa a 4 CONSTANT 136 004 131 545 93 901 51 542 YEAR i 0 001 0 001 0 001 0 000 Information Criteria AIC l 3468 115 AIC Corrected 3473 336 Schwarz s BIC 3522 3006 250 Chapter 8 Multiple Correlations BH BL NH MB 0 181 0 425 Drs TO 0 371 148 and df 150 1 1 R N 1 df where N Adjusted R2 Adjusted R BH BL NH MB 0 026 0 175 0 022 0132 RESIDUAL 1 RESIDUAL 2 RESIDUAL 3 Plot of Residuals vs Predicted Values Wivnalisas Ivnaisays e Ivngalis38 RESIDUAL 4 py IWNGISSY ESTIMATE 2 ESTIMATH3 ESTIMATH4 ESTIMATE 1 251 Applications Astronomy Sunspot Cycles SUNSPTDM data consists of a calculated relative measure of the daily number of sunspots compiled from the observations of a number of different observatories Variables Description YEAR The year the observations were made JAN DEC The relative measure of sunspots for the indicated month ANNUAL The mean relative measure of sunspots for the entire year Sunspots exhibit cyclical behavior on a 10 to 11 year cycle These cycles have potentially important effects on the earth s ecosystem including weather and the growth and development of living organisms Understanding the natural causes and effects of sunspot behavior are all important areas of scientific exploration Potential analyses include Time
118. Confidence Interval r Estimate Lower Upper A eee ee ee A oe ee a A es ee A eS ee ee ae l 1 20162 0 88635 1 51689 Pad OS 5 84938 8 70496 i 0 77647 0 06911 1 62204 0 15354 0 26604 0 04104 0 06307 0 10217 0 02398 ce Matrix B 1 B 2 ULCER DEPTH NODES ee fee es 0 02587 0 00284 0 53068 0 00750 0 28760 0 18613 0 00122 0 02138 0 00720 0 00329 0 00025 0 00290 0 00068 0 00002 0 00040 B 1 B 2 ULCER DEPTH NODES a a a EAU ODO 0 02421 1 00000 0 10803 0 91511 1 00000 0 13193 0 51120 0 29073 1 00000 0 07699 0 19929 0 07878 0 02046 1 00000 289 Log Log S t 100 Probability Plot Time 1000 Table of Estimated Quantiles for Last Accelerated Weibull Model Covariate Vector ULCER 1 507 DEPTH 2 562 NODES 3 246 Survival Probability Estimated Time 95 0 Confidence Interval Log of Estimated Time Applications Standard Error of Log Time 299 s995 990 s913 950 900 750 667 500 333 230 100 050 DAI 010 005 001 Oo0o0O0O0O0O0O00O0O0O00000Oo0Oo0OoOo Lower Upper 0 079 Su 166 0 895 21 825 23949 40 769 10 186 re OZ 294169 1794023 84 262 349534 353 087 932 437 560 840 13394193 1101 241 2474 271 T861 913 4426 540 2386 677 6039 263 3989 200 T223 1245 5152 747 17822 869 6287 225 24087 403 7752 889 33292 060 8840 918 40892 701 T1313 122 00452 1377 H O Www 10 00 JJ 7J 0 0014 0 NRO Oo0o0O0O0
119. Ctrl Alt F Ctrl 1 Ctrl 2 Ctrl 3 Ctrl Tab Ctrl Alt Tab Ctrl Alt Shift Tab Ctrl Home Ctrl End F10 Esc Ctrl Shift Alt P Ctrl Alt P Ctrl P Ctrl Z Alt Back space Ctrl Y Ctrl F F3 Ctrl H Ctrl R Ctrl A Ctrl Shift F Ctrl Shift Alt P Ctrl Alt P Customization of the SYSTAT Environment activate the Output Editor activate the Data Editor activate the Graph Editor invoke the Customize dialog invoke the Graph Gallery invoke the Graph Function Plot dialog activate the Workspace activate the Viewspace activate the Commandspace move the focus between the three spaces of the user interface This shortcut will not cycle between the three tabs of the Commandspace cycle forward to the right through the tabs of the active space cycle backward to the left through the tabs of the active space move the cursor to the top of the active tab move the cursor to the end of the active tab activate the File menu closes an open dialog box specify the printer paper size source and orienta tion to be considered while printing preview the content of the Output Editor before printing print the content of the Output Editor undo step by step a few steps of editing done redo step by step a few steps of editing done find text find the next instance of the text specified for the search replace text select enti
120. E 149554 411 331741074 Residual Covariance Matrix SY X DINNER STRANGER PROBLEM O E AN DINNER LY 287479 525 STRANGER 46647 669 85193 843 PROBLEM 5116 869 3323 431 749 802 Residual Correlation Matrix RY X DINNER STRANGER PROBLEM ts et DINNER i 1 000 STRANGER 0 298 1 000 PROBLEM 0 349 0 416 1000 Information Criteria AIC 1878 445 AIC Corrected 1893 445 Schwarz s BIC L T906 513 293 SETTING ji N of Cases 19 Least Squares Means i DINNER STRANGER PROBLEM aap ak ee oe S a 4 LS Mean 1142 316 628 474 49 526 Standard Error 123 006 66 962 6 282 SETTING 2 N of Cases 10 Least Squares Means i DINNER STRANGER PROBLEM WS is LS Mean 1418 700 564 400 39 200 Standard Error 169 552 92 301 8 659 SETTING L 3 N of Cases 19 Least Squares Means E DINNER STRANGER PROBLEM A A LS Mean 1365 368 878 895 66 474 Standard Error 123 006 66 962 6 282 Test for effect called CONSTANT Null Hypothesis Contrast AB DINNER STRANGER PROBLEM 1308 795 690 589 51 733 Inverse Contrast A X X 1A 0 023 Hypothesis Sum of Product Matrix H B A A X X 1A 1AB i DINNER STRANGER PROBLEM A ee H o DINNER 75105991 386 STRANGER 39629901 926 20910836 774 PROBLEM 2968749 169 1566469 415 117347 118 Error Sum of Product Matrix G E E i DINNER STRANGER PROBLEM a ee ee ee a DINNER 12936578 626 STRANGER 2099145 095 3833
121. EATH RT BIRTH RT XMIN 0 XMAX 60 YMIN 0 YMAX 30 E XTICK 6 KERNEL CONTOUR ZTICK 10 ZPIP 0 AX 0 SC 0 TITLE Birth and Death Rates for 30 Countries END The output is Birth and Death Rates for 30 Countries S S Deaths per 1000 People 1990 S 0 10 20 30 40 50 6C Births per 1000 People 1990 304 Chapter 8 Statistics Instructional Methods The INSTRDM data consists of measures of achievement on a biology exam for two groups of students one group simply told to study everything from a biology text in general and the other given terms and concepts that they were expected to master An additional covariate the students aptitude is also included in the data set Variable Description STUDENT Student ID INSTRUCT Type of instruction given INSTRUCT Coded variable for INSTRUCT APTITUDE Student s underlying ability to learn ACHEIVE Student s score on the exam From an education theory standpoint this data set is interesting because it demonstrates the effect on achievement due to different study instructions A student is likely to show a higher level of achievement when given specific instructions on what to know for an exam than a student who gets only general instructions From a statistical standpoint it demonstrates the importance of considering covariates when using ANOVA models A straight ANOVA of ACHIEVE on INSTRUCT shows no significance at the 95 confidence level but after separating out s
122. ENDOR Durbin Watson D Statistic 1 645 First Order Autocorrelation 0 137 Effects coding used for categorical variables in model The categorical values encountered during processing are Variables Levels A A E O A y ii a G E Ho gt gt gt ANGLE 2 levels 1 000 T000 Dependent Variable READING N l 16 Multiple R i 0 463 Squared Multiple R 0 214 Applications 266 Chapter 8 1 Estimates of Effects B X X X Y Factor Level READING ae aa e a CONSTANT 10500 ANGLE 1 1 500 Analysis of Variance Source Type III SS df Mean Squares F ratio p value a et a a a ia ANGLE 36 000 1 36 000 3 818 0 071 Error i 132 000 14 9 429 Least Squares Means 16 12 O Z Q lt LLI x 8 4 1 1 ANGLE Durbin Watson D Statistic 1769 First Order Autocorrelation 0 023s Creating the Four Factor Two Level Design Matrix The input is DESIGN SAVE XDESIGN FACTORIAL LEVELS 2 FACTORS 4 REPS 1 Once the design matrix is created the following steps complete the DOE process m Assigning variable names m Assigning factor level labels 267 Applications m Collecting and entering data m Performing analyses The output is B C AXDESIGN syz Dot Plots 1 1 1 1 1 1 1 1 The input is USE DESIGNDM CATEGORY SPRING POINTER VENDOR ANGLE THICK 6 CSIZE 2 DOT READING SPRING POINTER VENDOR ANGLE LINE SERROR 95 COLOR 1 FCOLOR 2 TIT
123. ENSOR CENSOR STRATA SEX ESTIMATE COX LTAB CHAZ The output is Time Variable TIME Censor Variable CENSOR Input Records 20069 Records Kept for Analysis 69 Censoring Observations A A ee Ho gt Exact Failures 36 Right Censored 33 282 Chapter 8 Covariate Means ULCER 1 507 DEPTH 2 562 NODES 3 246 Type 1 Exact Failures and Right Censoring Overall Time Range 720007 73 0700 0 Failure Time Range 72 000 1606 000 Stratification on SEX specified 2 levels Cox Proportional Hazards Estimation With stratification on SEX Iteration Step Log Likelih ood 0 0 112 564 1 0 108 343 2 0 103 570 3 0 HO 32553 4 0 103 533 Results after 4 Iterations Final Convergence Criterion 0 000 Maximum Gradient Element 0 000 Initial Score Test of Regression 32 533 with 3 df Significance Level p Value 0 000 Final Log Likelihood VO 533 AIC E LL DOS Schwarz s BIC 2Li 8L6 Standard Parameter Estimate Error Z p Value A eee H gt ULCER i 0 817 0 385 ae E23 0 034 DEPTH 0 083 0 053 LAS OS 112 NODES Cane it 0 057 22309 0 022 Life Table for Last Cox Model With stratification on SEX The following results are for SEX 0 Evaluated at Mean Values of Covariates ULCER 1 507 DEPTH 2 2 962 NODES 3 246 No Tied Failure Times Model Number at Number Survival Model Hazard Cumulative Risk Failing Time Probability Rate
124. Gibbons looked at the effects of two factors on tobacco use for students in 28 Los Angeles schools One factor involved the use of a social resistance curriculum or not The other factor was the presence or absence of a television intervention Crossing these two factors yields four experimental conditions which were randomly assigned to the schools Students were measured on tobacco and health knowledge both before and after the introduction of the two factors TYPING These data show the average speeds of typists in three groups using typing speed SPEED and a character or numeric code for the machine used EOUIPMNTS US State and Metropolitan Area Data Book 1986 Bureau of the Census The World Almanac 1971 POPDEN People per square mile FBI reported incidences per 100 000 people of personal crimes murder rape rob bery assault PROPERTY Incidences per 100 000 people of property crimes burglary larceny auto theft PERSON INCOME Per capita income SUMMER Average summer temperature WINTER Average winter temperature LABLAT Latitude in degrees at the center of each state LABLON Longitude at the center of each state RAIN Average inches of rainfall per year USCORR The data set is a correlation matrix among 16 variables from the USSTATES data file Following are the variable names ACCIDENT CARDIO CANCER PULMONAR PNEW FLU DIABETES LIVER VIOLRATE PROPRATE AVGPAY TEACHERS TCHRSAL MARRIAGE DIVORCE HOSPITAL DOCTO
125. Hazard 31 000 1 000 133 000 0 967 0 032 0 033 30 000 1 000 184 000 0 934 0 034 0 069 29 000 1 000 251 000 0 900 0 036 0 106 28 000 1 000 320 000 0 865 OOS 0 146 27 000 1 000 391 000 0 829 0 041 0 188 26 000 1 000 414 000 0 793 0 042 0 232 25 000 1 000 434 000 0 758 0 043 On ZTT 23 000 1 000 471 000 0 721 0 048 0 327 22 000 1 000 544 000 0 682 07053 0 383 20 000 1 000 788 000 0 638 0 062 0 449 19 000 1 000 812 000 0 596 0 065 O STS 283 OOOO 079 236 308 Model Hazard Oo0o0O0O0O0O0O00O0o00O00O0O000o0O0Oo0O0Oo0OoOo Rate RHRPOO 9803 gol Ll 018 468 Cumulative Hazard Oo0o0O0O0O0O0O00O0o00O00O0O000Oo0O0Oo0O0Oo0OoOo 15000 1 000 1151 000 0 547 T37 000 1 000 1239 000 0 491 5 000 1 000 1579 000 0 361 4 000 1 000 1606 000 0 230 Group size ot OOO Number Failing 15 000 The following results are for SEX 1 Evaluated at Mean Values of Covariates ULCER 1 507 DEPTH 124562 NODES 3 246 No Tied Failure Times Model Number at Number Survival Risk Failing Time Probability 38 000 1 000 72 000 0 998 37 000 10 00 125000 0 973 36 000 1 000 127 000 0 949 35 000 1 000 142 000 0 923 34 000 1 000 T5000 0 898 33 000 1 000 154 000 0 873 32 000 1 000 176 000 0 848 31 000 1 000 229 000 0 823 30 000 1 000 256 000 0 798 29 000 1 000 362 000 OSTIZ 28 000 1 000 422 000 0 747 27 000 1 000 441 000 00 220 26 000 1 000 465 000 0 692 25 000 1 000 495 000 0 663 23 000 1 000 584 000 0 634 22 000 1 20 00 645 000 0 603 21
126. IAN MINIMUM SD VARIANCE N PTILE 2 5 50 97 5 BEGIN DENSITY PP RBEP HIST XMIN 0 20 XMAX 0 35 LOC 0 0 DENSITY QQ RBEQ HIST XMIN 0 05 XMAX 0 13 LOC 0 3 DENSITY RR RBER HIST XMIN 0 60 XMAX 0 75 LOC 0 6 END FORMAT CLEAR function Taly fc2 fc3 fc4 274 Chapter 8 The output is PP QQ RR RBEP RBEQ A A A eti a a a ii Ho N of Cases i 10000 10000 10000 10000 10000 Minimum Ih OT le 0 06147 0 58743 0 22834 0 09148 Maximum SN E 0 12441 0 68789 0 26139 0 12460 Median 0 26412 0 09108 0 64442 0 24480 0 10736 Arithmetic Mean 0 26461 0 09119 0 64420 0 24486 0 10753 Standard Deviation 0 01516 0 00922 0 01334 0 00436 0 00448 Variance 0 00023 0 00009 0 00018 0 00002 0 00002 Method CLEVELAND 2 500 0 23545 0 07382 0 61744 0 23642 0 09916 50 000 LUO 2 eel 0 09108 0 64442 0 24480 0 10736 97 500 0 29642 0 11032 0 67021 0 25359 0 11642 RBER A A e e ti Vi a y ii ii 4 N of Cases i 10000 Minimum 0 62002 Maximum 0 67243 Median 0 64772 Arithmetic Mean 0 64761 Standard Deviation 0 00671 Variance 0 00005 Method CLEVELAND 2 500 0 63412 50 000 0 64772 97 500 E 0 66050 215 1000 0 10 2000 0 2 900 800 0 08 1500 700 y 3 600 0 06 8 8 E 3 E 5 2 500 Q 2 1000 0 18 O ke O o 400 0 04 0 300 S 5 500 200 0 02 100 0 0 00 0 0 0 0 20 0 25 0 30 0 35 0 20 0 25 0 30 0 35 PP RBEP 800 0 08 3000 700 0 07 600 0 06 3 2000 023 o lt O
127. Inhibitor concentration Understanding how reaction rates depend on the various reaction conditions 1s critical to optimizing the yield of a reaction Also the functional form of the rate on reaction parameters serves as a test of the theoretical models used to interpret a chemical reaction Potential analyses include nonlinear modeling bootstrapping and smoothing 258 Chapter 8 Estimation using Bootstrap Method The input is USE ENZYMDM NONLIN MODEL VELOCITY VMAX SUB CONC KM 1 INH CONC KIS SUB CONC ESTIMATE SAMPLE BOOT 100 o E Next the ESTIM file is used to draw the density plots ESTIM contains the estimated parameters for each sample USE ESTIM CBSTAT MEAN SD SEM DENSITY VMAX KM KIS The output is Arithmetic Mean Standard Error of Arithmetic Mean Standard Deviation 30 35 30 0 3 U 25 U 20 0 26 O 8 8 g 3 g 20 0 25 5 3 o 5 o 5 O jo O45 ne o o s s 10 0 1 0 y D 10 0 12 5 0 0 0 0 0 0 D gt Na o A D D 0 7 0 8 0 9 1 0 NY nY ny wv xv ny ny KM VMAX 30 TU 20 0 23 Lo 5 5 5 O po o 10 0 1 WO D s 0 0 0 PES g E S D Y AYYY 259 Nonlinear Analysis The input iS USE ENZYMDM NONLIN MODEL VELOCITY VMAX SUB CONC KM 1 INH CONC KIS SUB_ CONC ESTIMATE The output is Iteration History Applications No Loss VMAX KM KIS a oe se O i 3 568 1 010 1 020 1 030 L a 3192 1 009 0 988 Oe Go 2 2 897 TOLA 0 961 0 481 SOTA 1 02
128. K l yes regularly 2 no HEALTHY General health 1 excellent 2 good 3 fair 4 poor CHRONIC Any chronic illnesses in last year 0 no 1 yes SURVEY3 Marascuilo and Levin 1983 and Cohen 1988 This is a fictitious data set consisting of responses of 640 men COUN T to the question Does a woman have the right to decide whether an unwanted birth can be terminated during the first three months of pregnancy The response alternatives were cross tabulated with religion RELIGION and RESPONSES are represented by ordinal numbers in the data SWEAT Johnson and Wichern 2002 The data set consists of perspiration measurements from 20 healthy females on three variables sweat rate SWEAT RATE sodium content SODIUM and potassium content POTASSIUM SWETSDT Ae Swets Tanner and Birdsall 1961 and reported by Swets and Pickett 1982 This example shows frequency data for two detectors in a study Each of the subjects in the experiment used a six category rating scale RATING to indicate his or her confidence that a signal was present on each of 597 trials when the signal was present and on 591 randomly mixed trials on which the signal was not present The COUNT variable shows the number of times a subject gave a particular rating to a given signal state The identifier SUBJ is a numeric variable in this case SYMPe The dataset consists of 18 representative symptoms that have been taken and tallied for how many times they
129. LATITUDE Latitude at which the sample was taken LONGTUDE Longitude at which the sample was taken HORIZON Initials of producing horizon HORIZON ID of producing horizon URANIUM Uranium level in groundwater ARSENIC Arsenic level in groundwater BORON Boron level in groundwater BARIUM Barium level in groundwater MOLYBDEN Molybdenum level in groundwater SELENIUM Selenium level in groundwater VANADIUM Vanadium level in groundwater SULFATE Sulfate level in groundwater TOT ALK Alkalinity of groundwater BICARBON Bicarbonate level in groundwater CONDUCT Conductivity of groundwater PH pH of groundwater URANLOG Log of uranium level in groundwater MOLYLOG Log of molybdenum level in groundwater GRADES The variables in this data set are marks in four quiz QUIZ1 QUIZ2 QUIZ3 QUIZ4 of six students VAME and their marks in MIDTERM and FINAL exams GROWTH Each case in this file represents a group of plants receiving the same dose DOSE of a growth hormone GROWTH is the mean growth measure for each group and SE is the standard error of the mean HARDDIA Taguchi 1989 The data set consists of measurements on 20 units of two characteristics of a product Brinell hardness number BAN and circular diameter DIAMETER HEAD Frets 1921 The data consists of measurements on the following characteristics of two sons of 25 families The variables are HLEN Head length of the first son HBREADI Head breadth of the first son 333
130. LD are invalid shortcuts whereas the following 1s a valid one USE OURWORLD Commas and spaces Except when used to continue a command from one line to the next and in the case of functions commas and spaces are interchangeable as delimiters For example the following are equivalent CSTAT Urban babymort pop 1990 CSTAT urban babymort pop 1990 CSTAT urban babymort pop 1990 Quotation marks You must put quotation marks around any character string data that belongs to a string variable a string that needs to be case sensitive or contains spaces For example type NOTE Statistical Analysis to display a note in the output in title case and on a single line If your data file has a string variable for country names written in title case the following command will select the case corresponding to Sweden SELECT countrys Sweden non You can use either double or single quotes If you are using a dialog box to generate commands involving strings in general you may not need to specify quotation marks 133 Command Language In certain commands that involve values taken by string variables if you do not use quotes around a value SYSTAT looks for the value written in uppercase For example SPECIFY gov Democracy urban city will be interpreted as SPECIFY gov DEMOCRACY urban CITY whereas SPECIFY gov Democracy urban city will be interpreted as SPECIFY gov Democracy
131. LE Fuel Gauge Designed Experiment Results THICK 1 268 Chapter 8 The following plots assume that we have collected data in accordance with a generated experimental design The output is Fuel Gauge Designed Experiment Results 2 2 _ READING READING 1 1 1 1 SPRING POINTER READING READING 1 1 1 1 VENDOR ANGLE Environmental Science Mercury Levels in Freshwater Fish The MRCURYDM data consists of measurements of largemouth bass in 53 different Florida lakes to examine the factors that influence the level of mercury contamination The pH level amount of chlorophyll calcium and alkalinity were measured from water samples that were collected The age of each fish and the mercury concentration in the muscle tissue were measured older fish tend to have higher concentrations from a sample of fish taken from each lake To make a fair comparison of the fish in different lakes the investigators used a regression estimate of the expected mercury concentration in a three year old fish as the standardized value for each lake Finally in 10 of the 53 lakes the age of the individual fish could not be determined and the average mercury concentration of the sampled fish was used 269 Variable ID LAKES ALKLNTY PH CALCIUM CHLORO AVGMERC SAMPLES MIN MAX STDMERC AGEDATA LNCHLORO Applications Description Lake ID Lake name Measured alkal
132. LUSTER IDVAR COUNTRYS JOIN BERTH RU DEATH RT 302 Chapter 8 The output is Distance Metric is Euclidean Distance Single Linkage Method Nearest Neighbor Clusters Joining Haiti Jamaica France Italy Haiti Ecuador France Canada Algeria Somalia Trinidad Italy Hungary Barbados Brazil Ecuador Somalia Jamaica Jamaica Mali Somalia Yemen Algeria Jamaica Jamaica Yemen Jamaica Finland Sweden Ethiopia Chile UK Spain Sudan Turkey Germany France Libya Haiti CostaRica Canada Italy Argentina Trinidad Brazil Gambia Barbados Hungary Guinea Mali Somalia Bolivia Ecuador Algeria Iraq Yemen at Distance O NM Hs YY 0 NNNNNNDNRRR SRP SP 2 PPP R2PRRPRPPOOOO 083 No of Members WAA 01 UY N O 0 N ds N 0 NN OU N BN N W NN NE R 00 WON QD O Xe Clustering Countries by Birth and Death Rates Hungary Spain Italy Germany UK Sweden Finland France Canada Barbados Argentina Chile Jamaica CostaRica Trinidad Brazil Turkey Ecuador Libya Algeria Bolivia Iraq Sudan Ethiopia Haiti Somalia Gambia Quinea Mali Yemen Cluster Tree 2 3 Distances 4 303 Applications Kernel Densities Ellipses and Modal Smoothers The input is USE WORLDDM BEGIN PLOT DEATH RT BIRTH RT XMIN 0 XMAX 60 YMIN 0 YMAX 30 o o XTICK 6 SYMBOL 1 SIZE 5 LABEL COUNTRYS SMOO MODE XLAB Births per 1000 People 1990 YLAB Deaths per 1000 People 1990 DEN D
133. Means ULCER q 1 50725 DEPTH i 2 560203 NODES 3 24638 Type 1 Exact Failures and Right Censoring Overall Time Range 72 00000 7307 00000 Failure Time Range 72 00000 1606 00000 Weibull Model B 1 shape B 2 scale Extreme value parameterization Convergence 0 00000 Tolerance 0 00000 Iteration Step Log Likelihood Method 0 0 346 02864 BHHH 1 0 OO LOS BHHH 2 0 LO 12128 BHHH 3 0 o Los 69616 BHHH 4 0 o Lo ome BHHH 5 0 3 AA N R 6 0 3 0 Ta 55232 BHHH E 0 306 81388 BHHH 8 1 300 61528 N R 9 0 306 50985 N R 10 0 30 6350812 N R 11 0 306 908 L Z N R Results after 11 Iterations Final Convergence Criterion 0 00000 Maximum Gradient Element 0 00001 Initial Score Test of Regression 14 73796 with 5 df Significance Level p value 0 01154 Final Log Likelihood sr 300 590812 288 Chapter 8 AIC Schwarz Paramete ULCER DEPTH NODES 1 0 B 1 I I Vector i Coeffici Paramete DEPTH NODES Covarian ULCER DEPTH NODES Correlation Matrix 623 01624 s BIC 634 18677 r Estimate Standard Error Z p value A A A a O A A A a a 1 20162 0 16086 7 47021 0 00000 i EAS ALET 0 72848 9 98955 0 00000 l 0 77647 0 43142 1 79978 0 07190 0 15354 0 05740 2 67495 0 00747 0 06307 0 01995 3 16235 0 00156 0 83221 EXP B 2 1446 88707 Mean Failure Time Variance 1595 59198 3 71688E 006 900 37653 1 18354E 006 ent of Variation 1 20828 i 95 0
134. Models Second Edition John Wiley amp Sons Hollander M and Wolfe D A 1999 Nonparametric statistical methods 2nd ed New York John Wiley amp Sons Hosmer D W and Lemeshow S 2000 Applied logistic regression 2nd ed New York John Wiley amp Sons Hubert J J 1984 Bioassay Second Edition Dubuque Iowa Kendall Hunt 363 Data Files Huitema B E 1980 The analysis of covariance and alternatives New York John Wiley amp Sons Jackson J E 2003 A user s guide to principal components John Wiley amp Sons Jobson J D 1992 Applied multivariate data analysis Vol IT Categorical and multivariate methods New York Springer Verlag John P W M 1971 Statistical design and analysis of experiments New Y ork MacMillan Johnson R A and Wichern D W 2002 Applied multivariate statistical analysis 5th ed Engelwood Cliffs N J Prentice Hall Johnson R W 1999 The official NFL 1999 Record amp Fact Book New York Workman Publishing 435 Judge G G Griffiths W E Lutkepohl H Hill R C and Lee T C 1988 Introduction to the theory and practice of econometrics 2nd ed New York John Wiley amp Sons pp 275 318 pp 453 454 Kooijman S A L M 1979 The description of point patterns In R M Cormack and J K Ord eds Spatial and Temporal Analysis in Ecology Fairland Md International Co operative Publishing House pp 305 332 Kuehl R O
135. Montgomery D C Peck E A and Vining G G 2006 Introduction to linear regression analysis 4th ed Hoboken N J Wiley Interscience Montgomery D C and Runger G C 1993 Gauge capability and designed experiments Part 1 Experimental design models and variance component estimation Quality Engineering 6 1 115 Montgomery D C 2005 Introduction to statistical quality control 5th ed New York John Wiley amp Sons Morrison A S Black M M Lowe C R MacMahon B and Yuasa S Y 1990 Some international differences in histology and survival in breast cancer International Journal of Cancer 11 261 267 Morrison D F 2004 Multivariate statistical methods 4th ed Pacific Grove CA Duxbury Press Morrison K J and Zeppa R 1963 Histamine introduced hypothesion due to morphine and arfonad in the dog Journal of Surgical Research 3 313 317 Musa J D 1979 Software reliability data Data and Analysis Centre for Software Rome Air Development Center Rome NY Myers R H and Montgomery D C 2002 Response surface methodology 2nd ed New York John Wiley amp Sons 365 Data Files Neter J Kutner M H Nachtsheim C J and Wasserman W 2004 Applied linear regression models Homewood IL Irwin Netmaster Statistics Courses Available at http www dina kvl dk per Netmaster courses st1 13 Data datafiles planks txt Nichols C E Kane V E Browning M T and Cagle
136. NE 100 100 a a 5 o 5 al i i im o o o th th th a a a al 2 a A 2 0 2 0 ABILITY ABILITY 2 0 ABILITY 301 Sociology Applications World Population Characteristics The WORLDDM data contains 1990 information on 30 countries and includes birth and death rates life expectancies male and female types of government whether mostly urban or rural and latitude and longitude Variable COUNTRYS BIRTH RT DEATH RT MALE FEMALE GOV URBANS LAT LON Description Country name Number of births per 1000 people in 1990 Number of deaths per 1000 people in 1990 Years of life expectancy for males Years of life expectancy for females Type of government Rural or city Latitude of the country s centroid Longitude of the country s centroid Countries are often classified into categories for example developed or third world based on certain socioeconomic criteria one key group of criteria being population statistics This data set contains such criteria for 30 countries of various regions and per capita income levels allowing countries to be clustered according to population characteristics In addition variables such as the type of government and whether the country is mostly rural or urban may have an impact on these population characteristics Potential analyses include ANOVA regression cluster analysis multidimensional scaling and mapping Cluster Analysis The input is USE WORLDDM C
137. NNER STRANGER PROBLEM A A Hoo o gt DINNER 68 808 686 STRANGER y 283002039 879394 074 PROBLEM 7 11375 124 68489 589 3320393 Error Sum of Product Matrix G E E l DINNER STRANGER PROBLEM E ai ees a ee DINNER i 293695 78 626 STRANGER 2099145 095 38381226926 l PROBLEM 230259 126 149554 411 33741 074 295 Applications Univariate F Tests Source Type III SS df Mean Squares F ratio p value i ee i a u D gt gt gt gt gt DINNER i 687808 686 2 343904 343 a196 0 312 Error y 12936578 626 45 287479 525 STRANGER 879394 074 2 439697037 FLE 0 010 Error i 3833722 926 45 85193 843 PROBLEM DILO 2 2763 296 3685 0 033 Error i 33741 074 45 749 802 Multivariate Test Statistics Statistic i Value F ratio df p value A ES a e A A A a E l e a 4 Wilks s Lambda epee bed a 2519 6 86 0 027 Pillai Trace 0 290 2 488 6 88 O2 029 Hotelling Lawley Trace 0 364 2 547 6 84 0 026 THETA S M N p value 0 232 2 0 000 20 500 0 035 Test of Residual Roots Roots Chi square df a A a i yes H o gt 1 ERPOUGH 2 14 250 6 2 through 2 i 2 624 2 Canonical Correlations 0 482 0 241 Dependent Variable Canonical Coefficients Standardized by Conditional within Groups Standard Deviations i 2 a gaa ae ae ee 4 DINNER 0 341 0 980 STRANGER 0 723 0 288 PROBLEM 0 5
138. O0Oo0o00O0Oo0O000Oo0Oo0o0Oo0rR 290 Chapter 8 Quantile Plot 1 0 0 8 06 0 Q o Q 0 4 0 2 0 0 9 S N o S S S S S S S S S W Y Va D Se 291 Applications Psychology Day Care Effects on Child Development The DAYCREDM data consists of three measures of a child s social competence a measure for behavior at dinner a measure for behavior in dealing with strangers and a measure involving social problem solving in a cognitive test In addition there is a categorical variable for the setting in which a child was raised either by parents by a babysitter or in a day care center Variable Description SETTINGS Daycare setting in which child is raised SETTING Coded setting DINNER Behavioral measure of skill during dinner STRANGER Measure of skill in dealing with a stranger PROBLEM Social problem solving skill in a cognitive test An important issue in child development is whether the daycare setting in which a child is raised has a differential effect on social behavior This data set offers three measures of social competence for children in three different daycare settings some cared for during the day by parents others by a babysitter and the rest in a daycare center The data set is a good candidate for MANOVA because it offers three ways of measuring for a single latent variable social competence One critical issue is whether the data satisfy the assumptions of MANOVA especially regardi
139. OL tolerance TSLS Two Stage Least Squares TSQ chart Hotelling s T chart TXT text format U U chart chart showing defects per unit UCL upper control limit USL upper specification limit UTL upper tolerance limit y VAR variance VIF variance inflation factor W WMF Windows metafile Acronyms X XLS excel format X MR chart Individuals and moving range chart XPT TPT SAS transport files XTAB Crosstabulations Y Z A accelerator keys 220 access keys 220 223 224 active data file 24 add empty row 30 Add Examples 144 Advanced menu 32 align graphs 30 tables 30 text 30 Alt key 37 212 223 analysis of variance one way 81 post hoc tests 181 two way ANOVA 89 181 Analyze menu 32 application gallery 43 247 ASCII files 30 51 Autocomplete 237 B bar charts 84 90 bitmaps 30 196 BMP 196 Bonferroni adjusted probabilities 70 95 boxplots 81 Bubble Help 231 buttons appearance 219 customization 216 Discussion 41 in Help system 39 Reset 219 shortcut keys 220 Index toolbars 217 219 tooltips 219 C CAP 211 Case Selection 210 Invert 217 CGM 30 197 CLASSIC 240 clipboard command submission from 154 cut selection 220 export results 197 submitting commands 236 cold commands 130 collapsible link 23 collapsing 23 expanding 23 command buffer 236 command files 27 comments 146 creating 137 154
140. OR GREEN FILL TITLE Histogram for Uranium PPLOT URANIUM LOC 6in 0in FCOLOR gray FILL COLOR YELLOW TITLE Probability Plot tor Uranium END TASK al The DENS and PPLOT commands create the histogram and the probability plot respectively Between the BEGIN and END statements we can change the data file in use and plot an unlimited number of graphs Each graph can have its own attributes such as location and color Plotting Several Graphs Using Menus Plotting more than one graph can be accomplished directly from SYSTAT s menu E From the menus choose Graph Begin Overlay Mode m Choose graphs and options from menus and dialog boxes You can choose locations for the graphs in the Layout tab unless you want them overlaid on top of one another m Then from the menus choose Graph End Overlay Mode Display 109 Data Analysis Quick Tour Transforming Data and Selecting Cases In the Commandspace select and submit the line beginning with PPLOT Using the Graph Properties dialog box in the Workspace transform the URANIUM variable by clicking the down arrow of X Power until 0 is reached yielding a log transformation Probability Plot for Uranium N Normal 0 0 1 0 Quantile o ho 1 0 10 0 100 0 URANIUM ow a Notice that the probability plot is much more linear Using SYSTAT s lassoing capability you can isolate outliers m Click the Lasso icon g and lasso the two o
141. OTH DWLS TENSION 0 500 TITLE XLABEL YLABEL ZLABEL ZMAX 1 1 HEIGHT 3 75 WIDTH 3 75 ALTITUDE 3 75 PLOT VELOCITY INH CONC SUB CONC SIZE 0 SMOOTH DWLS SURF XYCUT TENSION 0 500 TITLE XLABEL YLABEL ZLABEL ZMAX 1 1 HE TLGHT 3 75 WIDTH 3 75 ALTITUDE 3 75 261 Applications PLOT VELOCITY INH CONC SUB CONC COLOR 11 FILL 1 SIZE 1 3 TITLE Enzyme Reaction Velocity by Concentration XLABEL Substrate Concentration YLABEL Inhibitor Concentration ZLABEL Reaction Velocity ZMAX 1 1 HEIGHT 3 75 WIDTH 3 75 ALTITUDE 3 75 PLOT VELOCITY INH CONC SUB CONC COLOR 2 FILL 0 SIZE 1 3 END THICK 1 The output is TITLE Enzyme Reaction Velocity by Concentration XLABEL Substrate Concentration YLABEL Inhibitor Concentration ZLABEL Reaction Velocity ZMAX 1 1 HEIGHT 3 75 WIDTH 3 75 ALTITUDE 3 75 Enzyme Reaction Velocity by Concentration Reaction Velocity 262 Chapter 8 Engineering Robust Design Design of Experiments DESIGNDM data consists of the results of a designed experiment to improve the performance of a fuel gauge Variable Description RUN The case ID SPRING Dummy variable for the type of spring used POINTER Dummy variable for the type of pointer used VENDOR Dummy variable for the vendor used ANGLE Dummy variable for the type of angle bracket used READING The reading of the fuel gauge under the designed conditions
142. OUNT POUNDING SINKING SHAKING NAUSEOUS STIFF FAINT VOMIT BOWELS URINE Description Number of soldiers in each profile of symptom Violent pounding of the heart Sinking feeling of the stomach Shaking or trembling all over Feeling sick at the stomach Cold sweat Feeling of weakness or feeling faint Vomiting Losing control of the bowels Urinating in the pants Determining which withdrawal fear symptoms are common to the soldiers after a combat and the probability of each taking place is useful in preparing the soldiers for future encounters Potential analyses include Test item analysis factor analysis multidimensional scaling and cluster analysis Classical Test Item Analysis The input is USE COMBATDM TESTAT MODEL POUNDING URINE FREQUENCY COUNT IDVAR COUNT ESTIMATE CLASSICAL 298 Index Oo0o0o0o00O0O0O0OoO Excl Item R DOO COCOCO0OO Chapter 8 The output is Case frequencies determined by value of variable COUNT Data Below are Based on 93 Complete Cases for 9 Data Items Test Score Statistics Total Average Odd Even ee Mean 4 538 0 504 2 473 2 065 Standard Deviation 2 399 0 267 1 333 LZS Standard Error tL 0 250 00 28 Un 139 MLS Maximum 25000 1 000 5 000 4 000 Minimum pu 000 Org ELT 0 000 0 000 N of Cases l 93 93 93 93 Internal Consistency Data Split half Correlation 0 690 Spearman Brown Coefficient 0 816 Guttman Rulon Coefficient 0 816 Coefficient Alpha All Items Do
143. PT respectively When used in the arguments of commands you should separate the number from the unit by a space For example DEPTH 2 CM sets the depth of a graph to 2 centimeters In the case of option values a number can be suffixed by the unit of measurement with or without a space For example the option HEIGHT 200PT sets the height of a graph to 200 points 134 Chapter 5 Reserved Keywords The following commands from the BASIC module are reserved keywords in SYSTAT You cannot use these words as variable names LET FOR IF THEN ELSE ARRAY DIM PRINT Barring these keywords you may name a file variable matrix array or user defined function by any string that you so desire However SYSTAT may encounter some name conflicts in certain commands In order to resolve such conflicts we will use a precedence rule Precedence The SYSTAT namespace which consists of all its possible module names commands arguments options and option values has the following precedence structure highest to lowest m Class 0 SYSTAT module names commands options and option values where such values are fixed keywords m Class 1 Built in function names m Class 2 User defined function matrix and array variable names m Class 3 File variable names in the currently active data file When SYSTAT encounters a potential conflict in a command line it will use the precedence rule to resolve the conflict Depending on the context a na
144. R USCOUNT Taken from the US data These data are the means of PERSON personal crimes and PROPERTY property crimes within REGIONS The COUNT variable shows the number of states over which the means were computed USINCOME S These data are on the average income INCOME of a few regions The variables are DIVISION COUNT INCOME 356 Chapter 9 USSTATES gt State and Metropolitan Area Data Book 1986 The variables are REGION and REGIONS Divide the country into four regions DIVISION and DIVISION Divide the country into nine regions LANDAREA Land area in square miles 1980 POP85 1985 population in thousands ACCIDENT Number of deaths by accident per 100 000 people CARDIO o of deaths from major cardiovascular disease per 100 000 CANCER Number of deaths from cancer per 100 000 people PULMONAR areal aes oor chronic obstructive pulmonary disease PNEU FLU ea of deaths from pneumonia and influenza per 100 000 DIABETES Number of deaths from diabetes mellitus per 100 000 people LIVER Dae of deaths from chronic liver disease and cirrhosis per 00 000 people DOCTOR Number of active nonfederal physicians per 100 000 HOSPITAL Number of hospitals per 100 000 in 1988 MARRIAGE Number of marriages in thousands in 1989 DIVORCE Number of divorces and annulments in thousands in 1989 TEACHERS Number of teachers in thousands TCHRSAL Average salary for teachers for the 1990 year HSGRAD a of public high school graduates in the 1982 83
145. RA a HESS 68 A First Look at Relations among Variables 69 Su ubpopulations sies 6 ehh how SBA A aa 73 A Two Sample t TesSt 78 A One Way Analysis of Variance ANOVA 81 A Two Way ANOVA with Interacti0d 89 Bonferroni Pairwise Mean Comparisons 95 SUMMALY e 4 4 ok Oe ie ee ee ee ee ee EO 97 4 Data Analysis Quick Tour 99 Groundwater Uranium Overview 2 00004 99 Potential Analyses arica ES SSL DROSS ESS 100 The Groundwater Data File 101 GGA NICS Ei medie re ch o Bs la hh eo a es 103 Distribution Plot aa dea HSH BES ee Ae 103 Exploring the Groundwater Data Interactively 104 Transformed Graph 2 0 0 0 0 ee 105 Histograms and Probability Plots 106 SYSTAT Windows and Commands 107 Transforming Data and Selecting Cases 109 Dynamically Highlighted Cases 110 Connections between Graphs and the Data Editor 111 Saus tes deve dE e A de ARCO a eie En 111 Graph of Mean Uranium Levels 112 Output for ANOVA 113 Outliers and Diagnostics 114 Nonparametric Tests ee 114 Advanced Graphics Vi PARR S OA ee ASS mS 116 Kniging SMOOMer sig od oka wa a aE GR eB eS 117 ROTOR aa Soh we oie ee as du ape ds te ds Pe eee eS 117 MOON a ak Y a BE Ae EA eee we 118 PASE VIEW rape o
146. RIZON HORIZON varlable UAANIUM gd OW cee Farwise Comparisons Dwass Steel Critchlow Fligner Conover Inman 116 Chapter 4 Output from Kruskal Wallis Test Kruskal Wallis One way Analysis of Variance for 127 Cases Categorical values encountered during processing are Variables Levels A A A ah es a a gt gt gt gt HORIZON 5 levels 1 000 2 000 3 000 4 000 5 000 Dependent Variable URANLOG Grouping Variable HORIZON Group Count Rank Sum 1 43 2851 500 2 18 986 000 3 21 1880 500 4 29 1455 000 5 16 955 000 Kruskal Wallis Test Statistic 15 731 p value is 0 003 assuming Chi square Distribution with 4 df From the Kruskal Wallis One way Analysis of Variance table the chi square test has a p value 0 003 meaning that there is only 0 3 chance that these data would show this much difference between the groups if the individual producing horizons have the same average level of uranium Thus we conclude that the uranium level differs significantly for producing horizons We arrived at the same qualitative conclusion from ANOVA and its Quick Graph but it was quantitatively different The p value in ANOVA was 0 014 here it is 0 003 Advanced Graphics This part of the tour explores SYSTAT s advanced graphics capabilities including 3 D rotation animation zooming using the Dynamic Explorer smoothers contour plots and Page view The graphics in this secti
147. S second table ranges from three to five There are 15 regular DIET no dinners and 13 diet DIETS yes dinners The List layout option in Two Way Tables in the Analyze menu is useful for summarizing counts that result from cross classifying two factors Let us look at combinations of DIET and BRANDS E From the menus choose Analyze Tables Two Way m Inthe Options group of the Two Way Tables dialog box select List layout and deselect Counts m Select DIETS as the row variable and BRANDS as the column variable 65 SYSTAT Basics if Analyze Tables Two Way Han Available yariable s Row anable Measures BRAND ml 3 DIETS FOODS E CALORIES Resampling FAT PROTEIN Column variable Add gt CALCIUM we c Remove C List layout Cell Statistics Crit lizt Stel T ables Counts Expected counts _ Percents C Deviates C Row percents Standardized deviates C Column percents C Combination Options C Include missing values Shade values ja Piso m Click OK Frequency Distribution for DIETS rows by BRANDS columns DIETS BRANDS Frequency Cumulative Percent Cumulative Frequency Percent A A ee eae ee no gor i 4 4 14 286 14 286 no st 4 8 14 286 28 571 no sw i 3 TA 10 714 39 286 no ty l 4 is 14 286 53 571 yes hc i 2 18 10 714 64 286 yes Te 5 23 17 857 82 143 yes WW 5 28 17 857 100 000 There are two DIET and seven BRANDS categories so there should be 14 co
148. S1 TEMP n n nu sl TEMP mean sum n PRINT The mean of the variable Sl is mean The output is 100 000 4997 915 318 SYSTAT created a temp variable sum SYSTAT created a temp variable n 100 000 500 062 493 SYSTAT created a temp variable mean The mean of the variable S1 is 5 000 Command Templates Command files provide a method for repeating analyses across SYSTAT sessions Output produced by a particular command file will be identical to output produced by any subsequent runs of the same command file assuming the data do not change If however we change the data file in use or replace the variables used for a graph or statistical analysis the results will vary from the original output but still retain the same structure Command templates provide a method for achieving this customizability 157 Command Language A command template provides a skeletal framework for graph creation statistical analysis or file management The template has the appearance of a standard command file but uses tokens in place of filenames variables numbers or strings Tokens serve as substitution markers a value must be substituted for the token for command processing to continue Every time you submit the command template you can substitute a different value for each token For example suppose we were to create a template for simple linear regression This model requires a response variable and a predictor va
149. SYSTAT I5 Getting Started SYSTAT WWW SYSTAT COM For more information about SYSTAT software products please visit our WWW site at http www systat com or contact Marketing Department Systat Software Inc 225 W Washington Street Ste 425 Chicago IL 60606 Phone 877 797 8280 Fax 312 220 0070 Email info usa systat com Windows is a registered trademark of Microsoft Corporation General notice Other product names mentioned herein are used for identification purposes only and may be trademarks of their respective companies The SOFTWARE and documentation are provided with RESTRICTED RIGHTS Use duplication or disclosure by the Government is subject to restrictions as set forth in subdivision c 1 11 of The Rights in Technical Data and Computer Software clause at 52 227 7013 Contractor manufacturer is Systat Software Inc 225 W Washington Street Suite 425 Chicago IL 60606 USA SYSTAT 13 Getting Started Copyright 2009 by Systat Software Inc Systat Software Inc 225 W Washington Street Ste 425 Chicago IL 60606 All rights reserved Printed in the United States of America No part of this publication may be reproduced stored in a retrieval system or transmitted in any form or by any means electronic mechanical photocopying recording or otherwise without the prior written permission of the publisher 1234567890 05 04 03 02 01 00 Contents 1 What s New and Diff
150. Series smoothing autocorrelation Fourier analysis ARIMA etc and Descriptive Statistics variance and distribution Autocorrelation Plot The input is USE SUNSPTDM SERIES ACF ANNUAL The output is Autocorrelation Plot 0 5 sl Correlation 10 20 30 40 50 Laa 232 Chapter 8 Biology Mortality Rates of Mediterranean Fruit Flies The FRTFLYDM data contains information on mortality rates for Mediterranean fruit flies over 172 days after which all flies died Experimenters recorded the number of flies dying each day and divided this by the number alive at the beginning of the day to measure the mortality rate for each day Variable Description DAY Day number LIVING Number of fruit flies alive at the beginning of the day MORTRATE Mortality rate of the fruit flies for each day The Mediterranean fruit fly data can be used to determine the functional form of mortality rate as a function of time A scatterplot of these two variables suggests that mortality rate might be a cubic function of time Since the number of fruit flies alive is directly determined by these two variables the mortality rate function can be substituted into an equation for the number of fruit flies living as a function of time which appears to be exponentially decreasing to estimate parameters for the nonlinear model Potential analyses include nonlinear modeling linear regression and transformations Nonlinear Modeling Showing an Ex
151. Sources Toxicology Data Source RETEFENCES s coce a e cs eh a E Acronym amp Abbreviation Expansions Index Chapter What s New and Different in SYSTAT 13 This chapter gives a summary of new features and major changes in this version relative to SYSTAT 12 in respect of GUI data commands output help graphics and statistics Under each of these items a list is given of new modified and deleted features This is followed by a brief description of each item in the same order with the same serial number More details are given in the appropriate chapters in the manual GENERAL FEATURES Graphical User Interface New Features pd Autohide Spaces Choice Tokens Data Edit Bar Data File Information Default Format for Saving Command Files Drag and Drop Data Embedded Toolbars Open Legacy Command Files Se S Sr ek age A ae View Toolbars 2 Chapter 1 10 Windows XP Style Grids 11 Trim Leading and Trailing Spaces in String Data Modified Features 12 Autocomplete Commands 13 Command Coloring 14 Dialog Boxes 15 Rescue Report 16 Shortcut Keys 17 Status Bar 18 Themes Deleted Features 19 Open Multiple Graphs View and Active Modes 20 Print Content of Data Variable Editor Data New Features 21 Close Data Files 22 Default Variable Format 23 Save View Mode Data Files 24 Import Business Objects Modified Features 25 Copy Paste to Data Var
152. Style Reg Tw Cen MT Condensed Tw Cen MT Condensed Verdana C Underline Color Use different options of the Font dialog box to change the appearance of any selected output text You can select the desired font type style and size You can also select effects like Underline and font color to be used Font Style You can change the selected output text to Bold Italicized and Underlined typefaces and also change the font color of the selected output text by selecting these options from Format in the Edit menu Alignment You can align the selected output text to the left right or centre by selecting those options in Format Bullets and Numbering Any selected text can be formatted as a Numbered list or a Bulleted list from the options in Format You can also reduce the indentation of the text or indent text by selecting Outdent or Indent respectively Inserting Image You can insert an image in the desired location of the Output editor by selecting the Insert Image option in Format Collapsible Links By selecting the Expand All option in Format you can expand all the links in the output you can collapse all those expanded links by selecting the Collapse All option in Format 188 Chapter 6 Find You can search for specific numbers or text in the Output editor To open the Find dialog box from the menus choose Edit Find C Match whole word only C Match case Search strings contain either comple
153. T LABLON RATIO ALT AZIMUTH WIDTH ID number Time in universal time at which eclipse will begin at the Latitude Longitude for that case Northernmost latitude of total obstruction Northernmost longitude of total obstruction Southernmost latitude of total obstruction Southernmost longitude of total obstruction Center latitude of total obstruction Center longitude of total obstruction Ratio of diameters of the Moon and the Sun Altitude above horizon at the given Latitude Longitude Azimuth at which eclipse will occur Width of the path of total obstruction 328 Chapter 9 TOTALITY AUG 11 1999 JUN 21 2001 DEC 14 2001 JUN 10 2002 DEC 4 2002 MAY 31 2003 APR 8 2005 OCT 3 2005 LABELS Time period of total obstruction at centerline Indicator for ellipse beginning on this date Indicator for ellipse beginning on this date Indicator for ellipse beginning on this date Indicator for ellipse beginning on this date Indicator for ellipse beginning on this date Indicator for ellipse beginning on this date Indicator for ellipse beginning on this date Indicator for ellipse beginning on this date Variable used for labeling eclipses on graphs EDUCATN This data set is a subset of the data set SURVEY2 EGGS gt Bliss 1967 An experiment was conducted to test the performance of laboratories and technicians to determine the fat content of dried eggs A single can of dried eggs was stirred well Samples were drawn and a pair of
154. T but do have a browser by simply supplying the htm or mht file 195 Working with Output Using Commands To save output enter the following OSAVE FILENAME SYO or RTF OR HTML or MHT Omitting SYO or RTF or HTML or MHT saves the output as a SYSTAT output file with an SYO extension To Direct Output to a File or Printer You can use commands to send output directly to a file or the printer OUTPUT lt FILENAME gt VIDEO or PRINTER or COMMANDS ERRORS WARNINGS For example the commands below send a listing of cases including commands to the text file MYFILE DAT The OUTPUT command at the end closes the text file so that subsequent output is sent to the screen only USE OURWORLD OUTPUT MYFILE COMMANDS LIST COUNTRYS HEALTH OUTPUT To Save Results from Statistical Analyses Many procedures include an option such as Save or Save File that saves the results of the analysis ina SYSTAT data file The contents of the file depend on the analysis For example m Correlations can save Pearson and Spearman correlations m Factor Analysis can save factor scores residuals and a number of other statistics m Linear Regression can save residuals and diagnostics for each case E Basic Statistics can save selected statistics for each level of one or more grouping variables m Crosstabs can save the count in each cell for later use as table input Check each procedure to see what is saved 196 Chap
155. TOR option of the TOKEN command TOKEN var TYPE NMULTIVAR SEPARATOR char Replace char with the desired single character separator SY STAT truncates separators longer than one character to the first character The designated character does not appear before the first variable or after the last variable 166 Chapter 5 String Tokens To substitute a text string for a token specify TOKEN text TYPE STRING When SYSTAT encounters the token amp text in the command file a dialog prompting the user for a string appears MM Enter String Replace kent with A Type the desired text string The entire string including any quotes entered as part of the string replaces the token For instance if a plot command contains a string token as an option PLOT Y X amp text you can enter a list of options such as XLAB X Variable YLAB Y Variable SYMBOL 2 as replacement text for the token Alternatively to prompt for each option setting assign each to a separate token NOTE Analysis of strl data NOTE strl Notice the tokens for the strings in the preceding command line For the first note quotes enclose the token In this arrangement the token replacement value should not include quotes but should only contain the text used to label the axis In contrast for the second note the token is not enclosed in quotes The appearance of this note depends on whether the quotes are included in the token repl
156. The data can be analyzed to determine if there are any changes in the skull sizes between the time periods The researchers theorize that a change in skull size over time is evidence of the interbreeding of the Egyptians with immigrant populations over the years Because there are four different measurements that characterize skull size multivariate techniques that allow multiple dependent variables can be used Dependent variables are the measurements MB BH BL and NH The predictor variable is YEAR Assuming that YEAR is a discrete predictor variable then data can be analyzed using MANOVA Assuming that there is a linear trend to the change in skull size then YEAR can be treated as a continuous predictor variable Potential analyses include MANOVA regression and principal components Box Plot and Regression The input is USE EGYPTDM THICK 2 5 BEGIN DENSITY MB BL YEAR BOX FCOLOR 1 FILL 1 XMAX 1000 XMIN 5000 COLOR 3 11 HEIGHT 5 5 WIDTH 4 XTIC 4 TITLE Variation of Skull Measurements by Period PLOT MB BL YEAR SMOOTH LINEAR SIZE 0 XMAX 1000 XMIN 5000 XTIC 4 COLOR 4 HEIGHT 5 5 WIDTH 4 END 249 Applications The output is Variation of Skull Measurements by Period YEAR MANOVA The input is PLENGTH SHORT USE EGYPTDM MANOVA MODEL MB BH BL NH CONSTANT YEAR ESTIMATE The output is N of Cases Processed 150 Dependent Variable Means MB BH BL NH 1334973 132 547 96 460 50 933 Regression
157. UM values are homogeneous Always take the time to think about what possible subgroups might be influencing or obscuring results A One Way Analysis of Variance ANOVA Does the cost of a dinner vary by brand Let us try an analysis of variance ANOVA to determine whether the average price of frozen dinners varies by brand After looking at the graphics earlier in this chapter we assume that differences do exist so we also request the Tukey HSD test for post hoc comparison of means This test provides protection for testing many pairs of means simultaneously allowing us to make statements about which brand s average cost differs significantly from another brand s Before we run the analysis of variance we will specify how the brands should be ordered in the output results will be easier to follow if we order the brands from least to most expensive 82 Chapter 3 E From the menus choose Data Order of Display In the Order dialog box select BRANDS as the variable Select Enter sort and type gor hc sw lc ww st ty Click OK From the menus choose Edit Options m In the Output Results group on the Output tab select Long from the Length drop down list This will provide extended results for the analysis of variance m Click OK To request an analysis of variance E From the menus choose Analyze Analysis of Variance Estimate Model m Inthe Analysis of Variance Estimat
158. USO he ty mb gO 0 001 27 035 0 465 SW re 0 421 0 548 BS oe 0 330 sw ww 0 437 0 506 1 187 0 314 SW st 0 682 Dis LL 1 466 0 103 SW ty 1 017 0 006 1 SUL 0 232 TE ww 0 016 1 000 0 666 0 634 lc st 0 261 0 874 0 950 0 428 ke ty 0396 0 120 1 285 0 093 ww st 0 245 0 903 0 934 0 444 WW ty 0 580 0 138 1 269 0 109 st ty a0 355 0 742 1 061 0 391 Let us read the Tukey results appearing above The first and second columns represent the pair and the third column indicates the difference in cost for each pair of means Differences between the gor brand and the others are reported in column 3 0 19 with hc 0 42 with sw and 1 44 with ty The fourth column reports the probability associated with each difference Gor is significantly less expensive than all brands except hc and sw In column 3 notice that on the average the hc brand costs 0 915 less than the st brand and 1 25 less than the ty brand From the probability table these differences are significant with probabilities of 0 015650 and 0 000672 respectively The only other significant difference is that the average price for the sw brand costs 1 02 less than the ty brand 89 SYSTAT Basics A Two Way ANOVA with Interaction Do nutrients vary by type of food Earlier in a scatterplot matrix we observed a small cluster of dinners that had higher calcium values than the others In the two sample t test we were unable to detect differences in average calcium value
159. We can point out that the means are ordered by increasing cost because of the Order feature This feature also pertains to graphical displays E From the menus choose Graph Bar Chart E Select BRANDY as the X variable and COST as the Y variable i Graph Bar Chart Man o Available variable z vanable s Options Error Bars Coordinates ARIS Adis All Sees Layout Color Fill BRANDS BRANDS FOODS CALORIES lt Remove FAT PROTEIN Y vanablels ITA MIA CALCIUM IRON COST DIETS vanable s Counts ofr Matris columns Display az Overlay multiple graphs into a single frame Range between two variables Cancel 85 SYSTAT Basics m Click the Error Bars tab and select Standard error from the Type group If Graph Bar Chart Main Options Coordinates KEN E YAMS All Axes Layout Color Fill Type p 0 6827 O Standard deviation i H Value of variable Set direction by Width of error bars 05 Style 0 Line with ticks Box Click the Fill tab select Select fill from the Fill pattern group and select __ as the Fill Pattern 86 Chapter 3 li Graph Bar Chart Main Fill pattern Options Default fill selection a Select fill Coordinates Fill Pattern Fill Pattern Value s AAMS 0 000000 feds All Axes Layout Select variable Color m Click OK COST gor hc sw Ic ww st ty
160. a are provided m Forecasts for error variances using the parameter estimates m Jarque Bera test for normality of errors m McLeod and Lagrange Multiplier tests for ARCH effect 2 Best Subsets Regression A new addition to SYSTAT s Regression suite this feature includes m Finding the best models choice of predictors given the number of predictors the number varying from one to the total number available in the data set m Identifying the best model by various criteria such as R Square Adjusted R Square Mallow s Cp MSE AIC AICC and BIC and m Performing a complete regression analysis on the data set chosen by the user same as the training set or different using the best model selected by any of the above criteria 3 Confirmatory Factor Analysis As part of the Factor Analysis feature SYSTAT now offers Confirmatory Factor Analysis CFA with m Maximum likelihood Generalized Least Squares and Weighted Least Squares methods of estimation of parameters of the CFA model m A wide of variety of goodness of fit indices to measure the degree of conformity of the postulated factor model to the data which include Goodness of Fit Index GIF Root Mean Square Residual RMR Parsimonious Goodness of fit Index PGFI AIC BIC McDonald s Measure of Certainty and Non Normal Fit Index NNFI 17 What s New and Different in SYSTAT 13 Environment Variables in Basic Statistics SYSTAT now provides environment variable
161. a new instance of the SYSTAT application Right click in the batch Untitled tab of the Commandspace and submit the file m Use the DOS command syntax to open or submit the file The details of this method are explained later in this chapter m Create a link to the command file in the Examples tab of the Workspace using the Add Examples dialog box that opens on clicking Add Examples under the Utilities menu Double click the link or right click and select Run to execute the command file as it is You can even use the context menu to open the command file in the batch tab edit it and then execute it Refer Chapter 7 Customization of the SYSTAT Environment to know more about adding examples m Open the command file in any external application like Notepad copy some or all commands right click anywhere in the Commandspace and select Submit Clipboard To submit a range of commands select the commands and choose Submit Selection from the context menu If the range includes the last command in the tab use Submit From Current Line to End If you choose either Submit Window or Submit From Current Line to End SYSTAT prompts you to specify whether to submit the range or not 145 Command Language Alternative Command Editors Command files are ASCII text files having an SYC filename extension and containing command syntax Hence you can use any text editor to create command files In your editor type each command on a new line and sa
162. a significant correlation with CALORIES r 0 55 p value 0 014 We are unable to detect significant correlations between COST and CALORIES FAT and PROTEIN Subpopulations The presence of subpopulations can mask or falsely enhance the size of a correlation With Correlations we could specify DIET as a BY GROUPS variable as we did previously Instead let us examine the data graphically and use 75 nonparametric kernel density contours to identify the diet yes and no groups We will also look at univariate kernel density curves for the groups m From the menus choose Graph Scatterplot Matrix SPLOM Select CALORIES FAT PROTEIN and COST as the Row variables Select DIETS as the Grouping variable Select Kernel Curve from the drop down list for Density displays in diagonal cells Select Only display bottom half of matrix and diagonal and Overlay multiple graphs into a single frame 74 Chapter 3 a Graph Scatterplot Matrix SPLOM Man o o Available variable s Row variable Options BRANDS CALORIES FAT FOODS Smoother E PROTEIN CALORIES COST E FAT PROTEIN Column variable s Anis VITAMINA Same as Aow Layout CALCIUM IRON COST Fill DIETE Symbol and Label Color Grouping variable s DIETS Legend Line Style Density displays in diagonal cells Specify separate row and column variables 7 Transpose matrix Overlay multiple graphe into a single frame m Click the Options
163. acement value 167 Command Language m Typing Response results in a note of RESPONSE Without using quotes SYSTAT displays labels in upper case m Typing Response results in a label of Response Because the command line does not include quotes around the token for the second note quotes must be included in the replacement value for the note to match the case of the supplied text string Numeric Tokens To substitute a numeric value for a token specify one of the following TOKEN amp num TYPE NUMBER TOKEN amp num TYPE INTEGER When SYSTAT encounters the token amp num in the command file a dialog prompting the user for a number or integer appears MM Enter Number Replace tnum with A After entering a value press Continue If the value is not numeric an error occurs and the user is prompted again Likewise attempts to input a decimal value for an integer result in re prompting The prompting dialog continues to appear until a valid value is entered or the Cancel button is pressed 168 Chapter 5 Custom Prompts By default the instruction appearing in substitution dialogs states Replace amp tok with To assist the user in entering valid information for a token replace the default instruction with a custom prompt using the PROMPT option of the TOKEN command For example to prompt the user for a graph title use TOKEN amp titlel PROMPT Enter the graph title When SYSTAT encounters am
164. action 347 Data Files X4 Zero span tensile The four paper properties are Y Breaking length Y2 Elastic modulus Y3 Stress at failure Y4 Burst strength PUMPFAILURES Gaver and O Muircheartaigh 1987 The data set consists of the number of failures F and times of observation T for 10 pump systems at a nuclear power plant PUNCH Cornell 1985 These data measure the effects of various mixtures of watermelon WATERMELN pineapple PINEAPPL and orange juice ORANGE on taste ratings by judges TASTE of a fruit punch QUAD Cook and Weisberg 1990 The data set is from a function which reaches its maximum at b 2c however for the data given by Cook and Weisberg this maximum is close to the smallest_X In other words little of the response curve 1s found to the left of the maximum QUAKES The Open University 1981 The data set consists of TIME in days between successive serious earthquakes worldwide QUESTABILITY Gibbons and Chakraborti 2003 In raising small children s ability an important factor is to develop their ability to ask questions in groups A study of group size and number of questions asked by preprimary children in a classroom atmosphere was conducted with a familiar person after dividing the 46 children into 4 groups Group 24 children Group2 12 children Group3 6 children and Group4 4 children The total number of questions asked QUESTIONS by all children of each group is recorde
165. age Examples Often the best way to learn about a procedure is through examples The Help system provides several examples for each statistical procedure or graph Select the example most relevant to your analysis or browse the examples to explore SYSTAT s capabilities El SYSTAT 13 E e oh Hide Back Stop Refresh Home ECAR Print Options Contents index Search Favortes Binary Logit with One Predictor Graphics Y Statistics Analysis of Variance o Bootstrapping and Sampling Classification and Regression Trees gt Cluster Analysis Conjoint Analysis Correlations Associations and Distance Measures o Correspondence Analysis o Crosstabulation Descriptive Statistics Design of Experiments gt Discriminant Analysis Factor Analysis Fitting Distributions General Linear Models o Hypothesis Testing Linear Regression t Logistic Regression 2 Overview D Binary Logit Models Multinomial Logit Models Conditional Logit Models Discrete Choice Models Examples 2 Binary Logit with One Predictor To illustrate the use of binary logistic regression we take this example from Hosmer and Lemeshow s book Applied Logistic Regression referred to below as H amp L Hosmer and Lemeshow 2000 consider data on low infant birth weight LOW as a function of several risk factors These include the mother s age AGE mother s weight during last menstrual p
166. ake sure you deselect the data before continuing Otherwise the remainder of the analyses will be done only on the selected observations To deselect the cases use the Lasso tool to select an area of the graph that contains no data points 111 Data Analysis Quick Tour Connections between Graphs and the Data Editor For those of you with a technical inclination here 1s the explanation of the connection between the graphs and the Data editor m Graphs have their own data allowing the real time transformations of the Graph Properties dialog box and the ability to save and reload them without the original data file m When a graph is plotted the data in the graph are linked to the Data editor allowing lassoing m The Data editor and the program kernel share the same data set so all data are live and what you see is what you get For example if you select data in the Graph editor and then run a regression the regression applies only to the selected data Statistics This part of the tour introduces SYSTAT s statistics capability Here we explore the question of whether the five producing horizons have varying levels of uranium by performing an ANOVA of URANLOG the log of URANIUM versus HORIZON This analysis is being done based on the visual judgment that the normal distribution for log URANIUM is a valid model m Inthe SYSTAT window click the ANOVA icon Xi on the Statistics toolbar Select URANLOG as the depende
167. alf of a 2 factorial design Each cell contains two observations on a Y variable FRTFLYDM Carey Liedo Orozco and Vaupel 1992 This data set contains information on mortality rates for Mediterranean fruit flies over 172 days after which all flies were dead Experimenters recorded the number of flies dying each day DAY and divided this by the number alive LIVING at the beginning of the day to measure mortality rate IWORTRATE for each day GAUGE 1 Smith 2001 The data set consists of repeated measurements READING of a characteristic of ten items ITEM each by three persons PERSON GAUGE 2 Montgomery and Runger 1993 Three operators measure a quality characteristic on twenty units twice each GDP The data set consists of CSO s quarterly estimates of growth rates of GDP for 1996 1997 to 2004 2005 for the following eight sectors The variables are YEAR AGRICULTURE MINING MANUFACTURE ELECTRICITY CONSTRUCTION TRADE FINANCING COMMUNITY OVERALL GDP GDWTRDM Nichols Kane Browning and Cagle 1976 The U S Department of Energy collected samples of groundwater in West Texas as part of a project to estimate U S uranium reserves Samples were taken from five different locations called producing horizons and Chapter 9 then measured for various chemical components In addition the latitude and longitude for each sample location was recorded The variables are SAMPLE The ID of the groundwater sample
168. all available commands arguments options option values Press the Esc key to close the drop down list Command autocompletion is enabled by default You can turn it off by unchecking Autocomplete commands in the General tab of the Edit Options dialog box or by clicking on AUTO in the Status Bar Command Coloring The commands variable names numbers strings and comments REM statements that you type will be colored in distinguishing colors The colors are as follows Commands Blue Command options comments Green Arguments option values Purple File variable names Black Numbers strings in quotes Pink Coloring makes it easy for you to identify the various components of a command line thereby reducing the risk of making syntax errors Command coloring is enabled 137 Command Language by default You can turn it off by unchecking Color command keywords in the General tab of Edit Options dialog box Online Help for Commands SYSTAT s online help system provides easy access to information about SY STAT commands At the command prompt type HELP followed by the name of a module or command for which you want help For example you can access help on the CORR module by typing HELP CORR If you are already in the CORR module you can type just HELP to get a list of commands available within CORR HELP followed by the name of a command that you know belongs to the CORR module for example HELP PEARSON or HELP followed by the
169. an use the contours to pinpoint the high levels of uranium with respect to the producing horizons The peaks of the kriging smoother are represented by tighter brighter yellow and red contours while the valleys are represented by dashed blue and green contours The actual data points are distinguished in color and symbol by producing horizon Notice how the peak is in the middle of the Quartermaster group this is why it had the highest value in the earlier ANOVA We can also see that the uranium level is not uniformly higher throughout this producing horizon but is highly localized Advanced Statistics The kriging smoother provided a quick geographic visualization of uranium concentrations SYSTAT also provides a comprehensive spatial statistics procedure for analyzing and modeling geographic data You can create variograms and perform stochastic simulation or kriging 123 Data Analysis Quick Tour Advanced Spatial Statistics Model Available varnablefa Dependent Yariogram SAMPLE URANLOG LATITUDE pa lt Remove ae LONGTUDE Pav HORIZON anig east URANIUM LONGTUDE ARSENIC lt Remove Resamplin BORON s axis north BARIUM Add gt iia MOLYEDEN k LATITUDE ana SELENIUM lt Remove VANADIUM SULFATE ass depth TOT_ALK RICAARAMA Variogram O riging O Simulation Save results Summary At this point we have made some significant discoveries about the groundwa
170. and reorganized table of measures 12 Cluster Analysis In Cluster Analysis the data file containing the saved results will preserve the value labels if any from the input data file 19 What s New and Different in SYSTAT 13 13 Fitting Distributions SYSTAT now performs the estimation of parameters for the beta chi square Erlang gamma Gompertz Gumbel logistic log logistic negative binomial Weibull and Zipf distributions using the maximum likelihood method 14 Hypothesis Testing for Two Sample Data in Columns For two sample z two sample t and test for two variances option for input data in a layout where the data across the samples appear in different columns This is in addition to the current indexed layout 15 Least Squares Regression The following enhancements are available in the Least Squares Regression feature m Save Standard Errors and Confidence Intervals in Least Squares Regression m A choice of bootstrapping residuals Bootstrap Estimates of the Regression Coefficients Bias Standard Error and confidence intervals are then computed based on these 16 Logistic Regression SYSTAT provides the following enhancements to its Logistic regression feature m Simplified user interface and command line structure to analyze binary multinomial conditional and discrete choice models separately m Option to specify the reference level for the binary and multinomial response models m Simpler form
171. ands 130 J JMP files 30 JPEG files 196 JPG 196 K keyboard shortcuts 220 224 232 Keyboard tab 224 L landscape orientation 200 201 LDISPLAY 244 license 33 linear regression examples 179 listing data 60 Log tab 28 logistic distribution 177 M Macintosh PICT files 196 menu animation 226 menus 30 Advanced 32 Analyze 32 data 31 edit 30 file 30 graph 31 help 33 Quick Access 33 themes 232 utilities 31 view 31 Window 33 208 metafiles 196 MHT 30 MINITAB files 30 modules 128 monospaced output 239 N normal distribution 175 176 177 NUM 209 numbers substituting for tokens 167 175 176 O one way analysis of variance 81 orientation 200 output commands 195 directing to a file 195 directing to a printer 195 HTML format 194 printing graphs 201 rich text format 194 saving 193 194 saving graphs 196 output editor 23 186 Index alignment 186 collapsible link 23 context menu 33 customization 208 find text 188 graphs 186 maximizing 208 preview 33 refresh 33 right click editing 188 tables 186 view source 33 Output format 238 output options 238 Output Organizer 27 captions 206 closing folders 189 Collapse Tree 31 configuring 191 context menu 34 customizing 206 detailed node captions 34 dragging entries 190 Expand tree 31 hiding 192 208 navigating output 189 rename 34 reorganizin
172. antify the linear relations among these variables m From the menus choose Analyze Correlations Simple m Inthe Simple Correlations dialog box select Continuous data type and select Pearson from the Continuous data drop down list m Select CALORIES FAT PROTEIN and COST as the variables 70 Chapter 3 BB Analyze Correlations Simple Man o Available variable z Selected varlable z Options BRANDS a FOODS ue Resamplin ae eee il CALORIES oe FAT PROTEIN VITAMIN CALCIUM IRON COST DIETS Types Deletion 2 Continuous date Listwise Distance measures O Pairwise Bank order date Unordered data Save matrix Binary data m Click the Options tab and select Probabilities and Bonferroni Because we study six correlations among four variables we use Bonferroni adjusted probabilities to provide protection for multiple tests 71 SYSTAT Basics MA Analyze Correlations Simple Options Bonferroni Dunr Sidak O Uncorrected Resampling EM estimation Mormal Contaminated normal eo E ee Proba Culito m Click OK Number of Observations 28 Means CALORIES FAT PROTEIN COST 303 214 10 804 19 679 2 544 Pearson Correlation Matrix CALORIES FAT PROTEIN COST eee ee EE CALORIES 1 000 FAT l 0 757 1 000 PROTEIN 0 550 0 278 1 000 COST i 0 099 0 134 0 420 1 000 Bartlett Chi square Statistic 38 865 df 6 p value 0 000 72 Cha
173. arger sample m From the menus choose Graph Scatterplot m In the Scatterplot dialog box select FAT as the X variable and CALORIES as the Y variable m Click the Fill tab in the Scatterplot dialog box and select a solid fill for the first fill pattern 54 Chapter 3 B Graph Scatterplot Ma Available variablelz e vanablels Options BRANDS add FAT FOODS CALORIES lt Remove Residuals FAT sarlable s PROTEIN 7 CALORIES Coordinates VITAMINA ia ER HARIS CALCIUM lt Remove IRON Aas COST variable DIETS r Matrix columns Display az 3 0 Smoother All Axes Layout Mirror Dual Color MultiF lot Fill Symbol and Label L Univariate density display on lt border Histogram L Univariate density display on Y border Histogram Surface and Line Style Overlay multiple graphs into a single frame Cancel m Click OK to execute the program 55 SYSTAT Basics 600 CALORIES w D O O O O NO O Oo 100 0 10 20 30 40 FAT Return to the Scatterplot dialog box by clicking the Scatterplot tool Notice that the previous settings are preserved m Click the Smoother tab in the Scatterplot dialog box and select LOWESS smoother S6 Chapter 3 29 Graph Scatterplot Main Smoother method Options None Spline Midrange C Kriging Linear tep CY Andrews Ou Os And Quadratic NEXPO Bisquare Coordinates Log Inverse
174. arlo C Quality analysis Statistics I C Statistics II 3 Statistics III E Statistics IV My Computer My Network Files of type SYE Files syc a 142 Chapter 5 Note m If you do not see the command file you are looking for you can choose a different file type in the Files of type field m You can also open a command file you used recently by clicking its name in the Recent Files quadrant of the Startpage or in the Recent Command item of the File menu Working with Text To undo your last action from the menus choose Edit Undo Or press Ctri ZzZ Or press Alt Backspace To cancel your last undo action from the menus choose Edit Redo Or press CErl Ye m To search for text from the menus choose Edit Find Or Press ERE In the Find what field enter the text you want to search for and then press Find Next To find additional instances of the same text continue to press Find Next 143 Command Language Or from the menus choose Edit Find Next Or press F3 You can search for whole words alone do a case sensitive search or search backwards m To replace a text from the menus choose Edit Replace Or Press CUP EE Find the desired text and press Replace or Replace All as desired Printing Command Files Currently the facility to print command files is not available in SYSTAT Open the command file in an alternative comman
175. ated into other strings as in this graph title Without quotes labels appear in upper case as in this tick label If quotes around the token are desired in the command file explicitly include them in the command lines Interactive Token Substitution To prompt the user for a token substitution value precede the token text with an ampersand in the command file During processing when SYSTAT initially encounters the token a dialog prompts for a replacement value 159 Command Language MM Enter String Replace der with A Entering a value and pressing the Continue button allows processing to continue Pressing the Cancel button halts further submission of the command file If subsequent commands use a token which has already been assigned a value SYSTAT substitutes that value automatically For example the command PLOT amp y amp x results in dialog prompting for the tokens amp y and amp x Suppose the current file has variables named AGE and DEPRESS If we assign DEPRESS to amp y and AGE to amp x the resulting graph plots depression score versus age If the command file continues with REGRESS MODEL amp Y CONSTANT amp X ESTIMATE SYSTAT computes the regression of depression score on age without prompting for substitution values Validating Input The Token Substitution dialog accepts any value supplied by the user However commands typically require numbers strings or filenames to execute correctly
176. atistics IV My Computer Dave as type SYC Files spc we hay E Note m To save a file under a different name click Save As from the File menu and specify the desired filename and path m To change the default command file format check ANSI under Default command file format in the General tab of the Edit Options dialog box m To save all unsaved files click from the File menu Save All and specify appropriate filenames for each m Instead of typing commands you can perform the corresponding actions through menus and dialogs and select Save or Save As with the Log tab active m The commands that you type line by line in the Interactive tab can also be saved to a command file by selecting Save or Save As with the Interactive tab active 141 Command Language To open a command file m From the menus choose File Open Command Or click the batch Untitled tab and press the Open toolbar button on the Standard toolbar Or right click on any batch Untitled tab and click Open m In the Look in field click the drive or folder that contains the command file you want to open m Double click the folder that contains the command file you want to open m Click the command file name from the list that is displayed and press Open Open SYSTAT Command File oki Of 2m 5 Data Yolurne oo _ExactTests My Recent Gettimg_Started Documents 2 GraphDeme 9 Graphics Miscellaneous 3 Montec
177. ault in any tab of the Commandspace SYSTAT displays command keywords in colored font with specific colors denoting specific kinds of keywords You may uncheck this option if you do not want commands to be colored Link data files to output file When a SYSTAT output file is saved the data files are linked to the output file That means you can open an output file saved in a previous session and continue working with it provided the underlying data files exist in the same path Uncheck this option if you do not want to use output files across sessions Save command log in output file When a SYSTAT output file is saved the command log will also be saved with it That means you can open an output file saved in a previous session and re use the commands from that session Uncheck this option if you do not use output files across sessions Perform substitutions specified by TOKEN commands With this option selected SYSTAT treats the ampersand amp character as a token indicator During processing predefined or user specified values replace every amp and the text immediately following it Deselect this option to prevent these substitutions Show Cancel dialog to terminate lengthy processing Whenever processing by SYSTAT takes some time before results can be displayed a Cancel dialog pops up so that you can cancel processing You may want to uncheck this option to avoid accidental cancellation of a process Prompt to save all documents
178. between the Viewspace and Workspace The Format Bar Data and Graph Editing toolbars can be toggled by right clicking on the Output editor Data editor and Graph editor tabs respectively and selecting Show Toolbar You can also close the Workspace Commandspace and toolbars so that more space 1s available for viewing the output data and graphs To do so m undock them and click in the upper right corner or deselect their entry on the View menu Closed items can be reopened via the View menu or using the keyboard Keyboard short cuts are explained in Chapter 7 30 Chapter 2 Menus SYSTAT has a common menu bar for all the panes and tabs There are menus for opening saving and printing files editing output transforming data matrix manipulation generating experimental designs and random samples performing statistical analyses and creating graphs At any given point of time those menu items that are relevant to the active pane or tab are enabled The menu can be customized using the Customize dialog from the View menu File Use the File menu to create or open data command and output files import from databases and save the contents of the active pane all panes or newly created data files The data file formats supported include SYSTAT Excel SPSS SAS MINITAB S PLUS Statistica Stata JMP and ASCII files You can save command files or the command log and submit commands that are in the Commandspace a command file the W
179. box to perform a step in an analysis a command is generated These commands are SYSTAT s instructions to perform the analysis Instead of using dialog boxes to generate these commands you can use the Commandspace and type them yourself Whether generated by the dialog box or typed manually the commands from each SYSTAT session can be saved in a file modified and resubmitted later Although many users will use dialog boxes exclusively we introduce commands here briefly to show how commands succinctly document the steps in your analysis If you do not expect to use commands you should skip the sections showing them You can type commands in the Commandspace of the SYSTAT window at the prompt gt on the Interactive tab When the Log tab is selected in the Commandspace the commands corresponding to your dialog box choices are also displayed in the Commandspace For example the following command was generated by the Scatterplot dialog box selections q USE Food syz REM Following commands were produced by the PLOT dialog PLOT CALORIES FAT REM End of commands from the PLOT dialog If you enter commands from Interactive tab you can recall previous commands by up and down arrow keys or by using F9 key Sorting and Listing the Cases Detailed graphics and statistics may not always be what you need sometimes you can learn a lot simply by looking at numbers This section shows you how to sort the dinners by type of food
180. can vertically through the active menu The left and right arrows open submenus or move between menus Use Enter to execute a selected item SYSTAT also offers shortcut and access keys for keyboard control of the SYSTAT interface Shortcut Accelerator Keys In general shortcut keys involve holding down the Ctrl key with a single letter to perform a specific task Most shortcut key combinations appear on the menus after the equivalent entry Shortcut key behavior may depend on the active window For example Ctrl P prints the content of the Output Editor if it is active but prints a graph if the Graph Editor is active The following shortcut keys are available Pane Tab Shortcut Key Function Any Ctrl N create a new file in the active tab Ctrl O open a file in the active tab Ctrl open data file Ctrl Shift import a data file from a database Ctrl S save the content of the active tab Ctrl Alt S save all open files Ctrl D Ctrl E save current data Ctrl Q quit the SYSTAT application Ctrl X cut selection placing contents on the clipboard Ctrl C Ctrl Insert copy selection to the clipboard Ctrl V Shift Insert paste clipboard contents at the current location Del delete the current selection F6 invoke the Global Options dialog Ctrl O launch a full screen view of the Viewspace 221 Output Editor Data Variable Editor Ctrl Shift O Ctrl Shift D Ctrl Shift G F4 Ctrl G
181. cations Interval Upper 254 Chapter 8 Asymptotic Correlation Matrix of Parameters i A B C A io a em ime ms A 1 000 B 0 952 1 000 C i 0 866 0 971 1 000 Scatter Plot 1500000 1000000 O Z 2 500000 0 0 50 100 150 200 DAY Scatterplot The input is USE FRTFLYDM PLOT LIVING DAY MORTRATE AX CORNER FILL FCOLOR GRAY COLOR RED XLAB Number of Flies Living YLAB Days Passed ZLAB Mortality Rate XGRID YGRID ZGRID TITLE Fruit Fly Mortality Rates Over Time 255 The output is Applications Fruit Fly Mortality Rates Over Time 1500000 1000000 500000 Mortality Rate Animal Predatory Danger SLEEPDM data contains information from a study on the effects of physical and biological characteristics and sleep patterns influencing the danger of a mammal being eaten by predators The study includes data on the hours of dreaming and nondreaming sleep gestation age and body and brain weight for 62 mammals Variable SPECIES BODY BRAIN SLO SLP DREAM SLP TOTAL SLEEP LIFE GESTATE PREDATION EXPOSURE Description Type of species Body weight of the mammal in kg Brain weight of the mammal in g Number of hours of non dreaming sleep Number of hours of dreaming sleep Number of hours of total sleep The life span in years The gestation age Index of predation as a quantitative variable Index of exposure as a quantitative variable 256 Chapter 8 The danger fac
182. ce Rules The SYSTAT namespace which consists of all its possible module names commands arguments options and option values now has the following precedence structure highest to lowest m Class 0 SYSTAT module names commands options and option values where such values are fixed keywords m Class 1 Built in function names m Class 2 User defined function matrix and array variable names m Class 3 File variable names in the currently active data file With the introduction of this precedence there will not be restrictions on variable names that you use in data files Depending on the context a name will be treated as coming from the lowest numbered class possible 38 String Subscripted Variables For string variable names that are subscripted you now have to prefix the dollar sign before the subscript For example what was myvar 1 in the prior version should now be myvar 1 39 Temporary Variables Temporary variable names should now be suffixed by the tilde symbol for example mytmpvar Also you need to use the TMP command to define temporary variables for example TMP mytmpvar 10 Deleted Features 40 Built In Variables The erstwhile CASE COMPLETE BOF BOG EOF and EOG are no longer available as built in variables They are now functions that you may use as before 14 Chapter 1 Output just by suffixing parentheses to the name For example SELECT COMPLETE and IF CASE lt 1
183. cedure to convergence Finally the parameter estimates standard errors standardized coefficients popularly called z ratios p values 95 confidence intervals and ratios and the log likelihood are presented The output is The categorical values encountered during processing are Variables Levels 41 Introducing SYSTAT The examples include all SYSTAT input You can copy and paste the example input also available as files in the Command folder of the SY STAT directory and having links in the Examples tab of the Workspace to the Batch tab of the Commandspace to submit the example as is or you can modify the commands to reflect your own analyses before submitting them The resulting output including graphical results follows the command input Many of the examples include Discussion buttons throughout the output Pressing any of these buttons yields a detailed explanation of the immediately preceding output There may also be examples that are explained in more than one step in which case More or Next buttons will be included in the page Example Command Files The input commands for each example in the User Manual or in the Help system are available as command files in the Command folder of the SYSTAT directory This provides an alternative way to run the examples These files are organized in terms of the printed manual Each file contains commands for one example and is named using six characters xxyyzz syc The fi
184. d and then double click the topic in the list or click and press the Display button to view 1t 39 Introducing SYSTAT Search Offers a full text search of the Help system Type the desired keyword and press the Enter key or the List Topics button The Help system returns all topics containing the specified term Double click the desired topic in the list or click and press the Display button to view 1t Check Search previous results to search for the keyword from within the previously listed topics By default all word forms of the keyword are located Uncheck Match similar words if you want just the exact keyword to be located Check Search titles only if you want to confine the search to the page titles alone Favorites Allows you to create and use a list of favorite help topics The topic that you are currently viewing will automatically appear in the Current topic You can either press Add to add this topic to the list or you can type in a page title that you know exists in the Help system and then press Add Select a topic 1n the list and press the Display button or the Enter key to view the topic Use the Remove button to remove a selected topic from the list The following buttons are available in the toolbar of the Help system Hide Show Hides or shows the Contents Index and Search tabs Back Returns to the previous Help topic Forward Moves to the next Help topic if you had pressed the Back button previously Sto
185. d editor like Notepad and use the Print option therein to print the command file Submitting Command Files When you submit a command file SYSTAT executes the commands as 1f they were typed line by line at the command prompt For example suppose you have a text file of SYSTAT commands named TUTORIAL SYC You can execute the commands in the file in eight different ways m Issue a SUBMIT command from any SYSTAT procedure SUBMIT tutorial Note Unless the command file is in the default directory for commands in the File Locations tab of the Edit Options dialog box you have to define the path for the file For information on Global Options see Chapter 7 Customization of the SYSTAT Environment 144 Chapter 5 m Inthe SYSTAT window from the menus choose File Submit File m Open the command file in the batch Untitled tab in the Commandspace using the File or context menu From the Submit sub menu of the File menu you can then submit the entire file Window submit from the cursor s location till the end of the file From Current Line to End or submit just the current line Current Line m From the menus choose Utilities User Menu Menu List and click on the item from the list For information on creating menu items in the User Menu refer Chapter 7 Customization of the SYSTAT Environment Double click the file after navigating to its location in the hard disk through Windows Explorer The file opens in
186. d for 30 minutes on each of eight different days BLOCK RAINFALL e Lee 1989 This is a data set of December rainfall Y on November rainfall X from 1971 to 1980 RANSAMPLEe The data set consists of 100 random observations on X Y Z where X follows the standard normal distribution Y given_X follows normal distribution with mean_X and standard deviation 1 Z given X Y follows normal distribution with mean_X and Y and standard deviation 1 The data set is generated by using SYSTAT RATGROWTH Milliken and Johnson 1992 This experiment involved studying the effect of a dose of a drug on the growth of rats The data set consists of the growth of fifty rats where ten rats were randomly assigned to each of the five doses of the drug The weights were obtained each week for eleven weeks The variables are DOSE RAT WEEK WEIGHT 348 Chapter 9 RATS Morrison 2004 For these data six rats were weighed at the end of each of five weeks WEIGHT 1 to WEIGHT 5 RCITY Adapted from a Swiss Bank pamphlet These data include 46 international cities CITY the name of continental region REG JON average working hours per week WORKWEEK working time in minutes to buy a hamburger and a large portion of french fries BIG_MAC average cost in U S dollars per basket of a basket of goods and services LIVECOST net hourly earnings EARNINGS and percentage of taxes security paid by worker PCTTAXES REACT These data i
187. d to the original filenames Check Retain original filename s to avoid the suffix Close Closes the application Command Log SY STAT records the commands you specify during your current session in a temporary file called the command log Select the Log tab in the Commandspace to view the command log You can view copy submit and save all of the commands stored in the command log at any time during a session However because the log serves as a command recorder you cannot edit commands using the Log tab 0 USE Ourworld syz REM Following commands were produced by the CSTAT dialog IDVAR CSTATISTICS POP_1983 BABYMORT HEALTH N MIN MAX MEAN SD IDVAR COUNTRY FREQ WEIGHT NISELECT BY REM End of commands from the CSTAT dialog Interactive Untitled syc z After selecting the Log tab you can submit commands directly from the command log in four ways Submit the entire log by choosing Submit Window from the File or context menus m Submit the most recently processed commands by moving the cursor to the desired starting point and choosing Submit From Current Line to End from the File or context menus m Submit a subset of commands by selecting the desired commands and choosing Submit Selection from the context menus m Submit the desired line by moving the cursor to the line and choosing Submit Current Line from the context menus 151 Command Language To modify commands before submission copy the log
188. dashboards and extraction transform and load operations necessary when building data warehouse The Business Objects Universe is a semantic layer which sits between the business end user and the complexities of the underlying database model End users force the universe to access all the databases to which they have been given permission This feature allows you to login to the Business Objects platform choose a universe to query build a query and process the resultant data in SYSTAT Modified Features 25 Copy Paste to Data Variable Editor SYSTAT now allows you to copy a cell and paste it into a column However it no longer supports the following 10 Chapter 1 26 27 28 m Pasting one or more cells in a row column to a block of cells encompassing more than one row column m Pasting an individual variable property to a new row in the Variable Editor m Pasting more than one property simultaneously to a block of variables Open Multiple Data Files In the previous version of SYSTAT the ability to work with multiple unmodified data files was tied to the global option to order output based on the input data file The two options have been delinked in this version and by default you may have multiple unmodified data files open with output ordered chronologically Ata time you may set any one of the files active for further processing If you still want to work with a single active data file SYSTAT provides a
189. data sets consist of a hypothetical agricultural data where the yields of crops are related to the soil type and the type of fertilizer used The variables are YIELD FERTILIZER and SOIL AIAG Breyfogle 2003 This data set originated from Automotive Industry Action Group AIAG 1995 The data set deals with measures of a critical quality characteristic MEASURE of 80 samples 5 samples collected in each of 16 subgroups SUBGROUP AIRCRAFT Bennett and Desmarais 1975 These data show amplitude of vibration FLUTTER versus time TIME in an aircraft wing component AIRLINE Box et al 1994 The variable PASS contains monthly totals of international airline passengers for 12 years beginning in January 1949 AKIMA Akima 1978 These data are topological measurements of a three dimensional surface using the variables_X Y and Z AMe Borg and Lingoes 1987 adapted from Green and Carmone 1970 This unfolding data set contains similarities only between the points delineating A and M and these similarities are treated only as rank orders Variables include A through 416 and ROW ANNEAL Brownlee 1960 The experiment seeks to compare two different annealing methods for making cans Three coils COIL of material were selected from the populations of coils made by each of the two methods METHOD Pair of samples was drawn from each of two locations LOCATION on the coil The response 1s the life LIFE of the can
190. distinct global option to close the active data file when another is opened Independent of this setting you may order the output either chronologically or based on the input data file Recode Variables SY STAT now offers an option ELSE which will allow you to recode all values other than a given set of values to a certain specified value Also when you recode into a new variable it inherits all non recoded values from the old variable Use the ELSE option if you do not want to inherit the non recoded values Store and Retrieve Current Settings SYSTAT now supports storing the current setting of the following m active data file m value label display format m variable label display format The stored settings may then be retrieved at any subsequent instant during the current session Deleted Features 29 View Data Files You will no longer be able to open data files directly in the view mode However by default data file tabs will switch to view mode as before when another file is opened or set active 11 Command Line Interface New Features 30 ACTIVE Command 31 What s New and Different in SYSTAT 13 The ACTIVE command now activates a file that is in the view mode It no longer opens the file from disk Built in Functions SYSTAT offers the following new built in functions Mathematical ACSH COSH ASNH EVEN CASE FLOOR CEIL NCASE COLUMN NVAR Multivariable COMPLETE Groups and Inte
191. documentation to find out how to save data as an ASCII text file Make sure that your text file satisfies the following criteria m Each case begins on a new line to read ASCII files with two or more lines of data per case use BASIC commands m Missing data are flagged with an appropriate code Imagine that someone used a text editor to enter 10 pieces of information variables about 28 frozen dinners BRAND Short names for brands FOODS Words to identify each dinner as chicken pasta or beef CALORIES Calories per serving FAT Total fat in grams PROTEIN Protein in grams VITAMIN A Vitamin A percentage daily value CALCIUM Calcium percentage daily value IRON Iron percentage daily value COST Price per dinner in U S dollars DIET Yes the dinner was shelved with dinners touted as diet or low in calories No it was shelved with regular dinners BRAND FOODS CALORIES FAT PROTEIN VITAMINA CALCIUM IRON COST DIET Ic chicken 270 6 22 6 10 6 2 99 yes Ic chicken 240 5 19 30 10 10 2 99 yes Ic chicken 240 5 18 4 10 8 2 99 yes Ic pasta 260 8 15 20 30 8 2 15 YES Ic pasta 210 4 9 30 10 8 2 15 yes ww chicken 260 4 2i 30 4 15 2 79 yes ww pasta 220 4 14 15 8 15 2 79 yes 52 Chapter 3 BRAND FOODS CALORIES FAT PROTEIN VITAMINA CALCIUM IRON COST DIET ww pasta 220 6 15 6 25 15 2 79 yes he chicken 200 2 17 0 Z 2 2 00 yes he chicken 280 3 24 15 4 15 2 00 yes ww chicken 160 l 13 30 2 2 2 49 yes he pasta 250 3 20 0 8 8 2 00 yes
192. e Mean Survival Time 95 Time 31 La 38 7307 000 1606 000 2 levels K M Probability Oo0o0O0O0O0Oo0o00O00O0Oo0O0O0Oo0OoOo 000 000 200 0 Confidence Interval 2399302 1270408 Survival Quantiles Probability Survival Time 8 95 0 Ls 968 e935 903 SOL 839 806 774 741 707 672 636 594 548 438 329 Standard Er OoOo0O0O0O0Oo0Oo00O00O0Oo0O0O0Oo0OoOo Confidence Interval Lower Upper 0200 0 500 02750 1 579 000 471 000 1 579 000 788 000 251 000 The following results are for SEX 1 Number at Risk Number Failing hehe ODO 000 000 000 000 000 000 000 000 000 000 Time K M Probability Oo OO 0 0000000 DRA 947 s921 995 868 842 816 199 763 STOT 711 1151 000 ror Standard Er O OGOOGO OLO OOO ror 95 03 Applications Confidence Interval Lower Upper 0 792 0 995 0 766 0 983 OSTA 0 968 0 692 0 950 0 655 0 929 0 619 0 908 0 584 0 885 0 547 0 861 0 512 0 836 0 475 0 808 0 439 0 780 0 394 Os 7A 0 346 Oe TEL OS 0 657 0 103 0 580 Confidence Interval Lower Upper 0 828 0 996 0 806 0 987 Oe TTS 0 974 0 743 0 959 OTZ 0 943 0 682 0 926 0 652 0 908 0 623 0 889 0 594 0 869 0 566 0 849 07539 0 828 286 Chapter 8 27 000 1 000 441 000 0 684 0 075 26 000 1 000 465 000 0 658 0 077 25 000 1 000 495 000 0 632 0 078 23 000 1 000 58
193. e PROMPT What is the first factor TOKEN amp factor2 TYPE variable PROMPT What is the second factor TOKEN amp dep TYPE variable PROMPT What is the dependent variable NOTE Two way Analysis of Variance of NOTE edep using amp factorl and amp factor2 as factors DENSITY dep amp factorl amp factor2 BOX ANOVA CATEGORY amp factorl amp factor2 REPLACE DEPEND dep SAVE amp o0utfile RESID DATA ESTIMATE HYPOTHESIS POST amp factorl SCHEFFE TEST HYPOTHESIS POST EE SICEOEZ SGHEFFE TEST HYPOTHESIS POST amp factorl amp factor2 SCHEFFE TEST USE amp outfile CATEGORY amp factorl amp factor2 REPLACE LINE ESTIMATE amp factorl OVERLAY GROUP amp factor2 TITLE Least Squares Means YLAB amp dep CATEGORY amp factorl amp factor2 OFF PLOT student estimate SYM 1 FILL 1 STEM student 183 Command Language To create the same output without a template requires the following dialogs Box Plot ANOVA Estimate Model Three uses of GLM Pairwise Comparisons invoked thrice Line Chart Scatterplot Stem and Leaf For every dialog variable selection must occur Creating a command file does automate these analyses but command files do not generalize across data files By using this template we replace the eight dialogs and the necessary specifications for those dialogs with four simple prompts In addition the resulting template can generate results for any specified data file
194. e unfortunate event of a crash SY STAT attempts to recover the log output and data files of the session These files are saved to the Rescue sub folder within the SY STAT user folder Before closing the Rescue Report dialog pops up Rescue Report S S TAT has encountered a problem and needs to close We are sorry for the inconvenience We would like to know more about this event Click Send Report to generate an erail Message v Attempt to restore session Don t Send File information 153 Command Language Attempt to restore session Opens the recovered files if any on restarting SY STAT You will be prompted to save the recovered data files Details Displays the filename and location of the recovered files Send Report Generates an email message with the recovered files attached Don t Send Terminates the current session without generating the email message Working with DOS Commands Some of the tasks that SY STAT is capable of can be performed with minimum user intervention For instance there may be very large command files you want to execute or command files that require a long time to produce output or command files that produce a large number of graphs all of which you want to save It is indeed possible to do all this and much more in the Windows environment In fact you can work with SYSTAT command files even without having to open the SYSTAT application manually All you need to do is to invoke
195. e Model dialog box select COST as the dependent variable and BRAND as the factor variable m Click OK 83 SYSTAT Basics pA Analyze Analysis of Variance Estimate Model Mad Available variable z Dependents Repeated Measures BRANDS COST FOOD CALORIES Resampling FAT PROTEIN Factors ITA MIRA 3 BRAND Coding CALCIUM hai Effect IRON La COST eunn DIETS C Missing values Covariate s Options Effects coding used for categorical variables in model The categorical values encountered during processing are Variables Levels I sy i ye fa ya a a a a a BRANDS 7 levels gor he Sw Le Www st ty Dependent Variable COST N 28 Multiple R 0 861 Squared Multiple R 0 742 Estimates of Effects B X X 1X Y Factor Level COST A ee ee S 4 CONSTANT D505 BRANDS gor 0 695 BRANDS hc 0 505 BRANDS Sw 0 271 BRAND lc 0 149 BRANDS WW 0 165 BRANDS st 0 410 Analysis of Variance Source Type III SS af Mean Squares F Ratio p Value a a a ed pae D BRANDS 6 017 6 1 003 10 042 0 000 Error 2 097 21 0 100 84 Chapter 3 Least Squares Means Factor Level gor hce SW IG WW st ty i LS Mean Standard Error N An n ea e H o gt gt i 1 810 0 158 4 000 i 2 000 0 182 3 000 l 2 233 0 182 3 000 2 654 0 141 5 000 2 670 0 141 5 000 i 2 915 0 158 4 000 3 250 0 158 4 000
196. e by clicking on the auto hide pin H Customizing the Output Organizer You can customize the captioning of text nodes in the Output Organizer By default the caption is the title of the analysis that the node pertains to The associated command appears as a tooltip on mouse hover To see the tooltips themselves as node captions from the menus choose Edit Output Organizer Show Detailed Captions For a given analysis the associated command 1s the most significant command related to that analysis typically the HOT command For example for least squares regression the default node caption is OLS Regression whereas the detailed node caption is the MODEL command line Adding Examples The Examples tab in the Workspace contains a SYSTAT Examples tree that is organised by folders and nodes the folders corresponding to volumes or chapters of the SYSTAT User Manual and the nodes corresponding to the example command scripts therein Double clicking a node executes the underlying command script You can add your own examples to this tree organized according to the directory structure of your folder containing such examples To add examples from the menus choose Utilities Add Examples 207 Customization of the SYSTAT Environment Add Examples My Documents Y My Computer 2d My Network Places O 2 Internet Explorer In the dialog that opens the left hand side contains a box displaying all the drives folders
197. e ce oe Be See ee GS Hees e Bae Se ws d 120 Contour Plot of the Kriging Smoother 121 Advanced Statistics s 224 a a eh OSES 122 A eS 123 References for Groundwater Data 124 Commandsp ce s aose ea ae ii boat a 126 What Do Commands Look Like 127 Interactive Command Entry gt s o sce ana a ace A 127 Command yllax s 40 2 EA EE 24d 129 Command Syntax Rules 130 Autocomplete commands 2000005004 136 Command Coloring 44 dra a Be Ah bo e tee ge A 136 Online Help forCommands 004 137 Command FilES a aa ra a See 137 Working with Text cu dio ri aa oo 142 Submitting Command Files o 143 Alternative Command Editors 145 Comments in Command Files 146 Translating Legacy Commands 147 SYSTAT Command Translator 149 Command LOS nian s aea a A de A RS 150 Recording Seripis ia aa a pia ea 151 Rescuing Sessions 4 40444 Se A AAA A RS 152 Working with DOS Commands 4 153 Environment Variables 154 Example Computing Mean Using Environmental Variables 155 Command Templates 156 Automatic Token Substitution 158 Interactive Token Substitution 158 Viewing Tokens 0 imc om Sook Ge hate ce ead ob do dd 170 Predefined tokens x w cs ode a ee Be Bes 171
198. e color palette or create your own color by clicking More colors Clicking this opens the Color dialog Basic colors Mi ey eee A O A A HKH ee e iz E 4 Mm Hue Red o Sat jo Green o Define Custom Colors gt gt ColorfSolid Lum el Blue o Basic colors Click one of the basic colors and press OK to use that color ll amon See E Custom colors Click a basic color to begin with It shows up in the Color Solid area with the cross hair at the corresponding point in the full color spectrum above it and 243 Customization of the SYSTAT Environment an arrow at the corresponding point in the color bar beside the spectrum You can move the cross hair to any point in the full spectrum and slide the arrow to any height in the color bar You can also enter hue saturation luminosity and RGB values Press Add to Custom Colors to add the color to the Custom color palette You can create any number of colors in this way Finally click on a color and press OK to use that color File Locations Use the File Locations tab to specify the folder containing the files used in the Graph Gallery to designate file paths to append to filenames used in SYSTAT commands and define paths to store command graph and output files Set project directory Resets file paths for all file types to the appropriate sub folders within the designated folder Check Use common directory if you want all subsequent file ope
199. e default menu theme from the menus choose Utilities Themes Apply Default Theme 234 Chapter 7 Global Options SYSTAT has a host of global settings that you can customize according to your preferences These settings are automatically saved at the end of a session and remain in effect for subsequent sessions Most of them can also be accessed through the Global Options toolbar or the status bar To open the Global Options dialog box from the menus choose Edit Options The six tabs in the Options dialog box control different settings in SYSTAT General Specify general appearance and behavior options Data Specify Data Editor display options Output Specify the general appearance of output Output Scheme Specify font and color for individual components of the output as well as the background image or color for all of the output Graph Specify graph scaling line thickness character size and measurement units for all subsequent graphs File Locations Set folders in which SYSTAT should look for files of different types The General Output Output Scheme and File Locations tabs are described here For information about Data options see SYSTAT Data For information about Graph options see SYSTAT Graphics General Options The General tab of the Global Options dialog controls the ordering of variables in dialog boxes token processing and command recall 200 Customization of the SYSTAT Environment
200. e different fats F4T with each of three different surfactants SURF on the specific volume of bread loaves SPVOL baked from doughs mixed from each of the nine treatment combinations Four flours FLOUR of the same type but from different sources were used as blocking factors That is loaves were made using all nine treatment combinations for each of the four flours MJ173 Milliken and Johnson 1984 This is a hypothetical data set from a two way treatment structure in a completely randomized design with treatment T and treatment B each having three levels MJ202 Milliken and Johnson 1984 These data are from a home economics survey experiment DIFF is the change in test scores between pre test and post test on a nutritional knowledge questionnaire GROUP classifies whether or not a subject received food stamps AGE designates four age groups and RACE designates whites blacks and Hispanics 340 Chapter 9 MJ332 Milliken and Johnson 1984 An experiment involved 3 drugs to study the effect of each drug on heart rate of eight persons in four time periods The variables are PERSON HR DRUG TIME MJ338 Milliken and Johnson 1984 An engineer had three environments in which to test three types of clothing Four people two males and two females were put into an environmental chamber each one was assigned one of the three environments One male and one female wore clothing type 1 and the other male and female wore cloth
201. e multiple variables for a single token specify one of the following TOKEN amp var TYPE MULTIVAR TOKEN amp var TYPE CMULTIVAR TOKEN amp var TYPE NMULTIVAR When SYSTAT encounters the token amp var in the command file a dialog prompting the user to select multiple variables appears If no data file is currently open SY STAT prompts the user to open a file before proceeding to variable selection 165 Command Language Replace Variable Replace MAR with Available variable Selected variables COUNTRY COUNTRY POR 1983 POP_ 1983 FOF 1956 POP_1986 PORP_1990 POP 2020 URBAN BIRTH 82 DEATH_82 DEATH_RT BABYMT 82 BABYMORT Select one or more variables and click Add to include the variable s in the token replacement set To select multiple consecutive variables hold down the Shift key and click the first and last variables in the desired set To select multiple nonconsecutive variables hold down the Ctrl key and click each variable before clicking Add Click Continue to continue command processing The list of available variables corresponds to the dialog type To list all variables use TYPE MULTIVAR The variable list contains only string variables if the token type equals CMULTIVAR The NMULTIVAR type lists numeric variables for token substitution By default during multiple variable substitution SYSTAT inserts a space between the selected variables To specify an alternative character use the SEPARA
202. e name The name should 228 Chapter 7 be unique Click on the row and press the Delete key if you want to clear a name Press the Enter key or click outside the row to assign the name to the new list m Delete Row Deletes the selected list Alternatively right click on the list and select Delete Row For the set of command files in a list the two buttons have the following functions m New Adds a file to the selected list When adding a file to a list press the ellipsis El button at the right of the new entry to browse for a particular file Alternatively type the path and filename into the list of command files SYSTAT automatically appends the currently defined path for command files to any typed filenames without a path m Delete Deletes the selected command file from the list The command file is deleted from the list only the file is not deleted from the user s system Submission From File Lists In addition to offering a mechanism for organizing files command file lists also allow submission of the files contained in the lists As a result you can create templates for custom graphs assign them to a file list and apply them to the current data via a mouse click Use the Submit from File List button E on the Standard toolbar to submit files from previously defined command file lists Alternatively from the menus choose File Submit From Command File List This presents the names of all files in the com
203. e the translated commands to a SYSTAT command file 149 Command Language Open in Commandspace You can request that the translated commands be opened in an untitled tab in the Commandspace SYSTAT Command Translator In addition to the Translate Legacy Command Files dialog box SYSTAT provides a SYSTAT Command Translator application To access it click Command Translator in the SYSTAT 13 program group of the Windows Start Menu Select Variable s Replace amp war with Avallable yariable s Selected yarnables LOG_GDP EDUC EDUC 84 HEALTH EDUC ul LEADERS HEALTHS4 HEALTH MIL_o4 MIL GOVERN LEADERS Continue Cancel Add Files Add any number of files that you need to translate to SYSTAT 13 syntax In the File Open dialog box you will be able to click and drag the mouse or use the Ctrl Shift keys to select multiple files simultaneously When you press Open the files will get listed in the box beneath Click on any command file to view its content in the box beneath Translate from The following options are available Check one of the following m Version 12 to Version 13 m Version 11 to Version 13 E Version 11 to Version 12 150 Chapter 5 Translate When you press this button all the selected files will be translated so as to be suitable for execution in the specified version of SYSTAT Save Specify the folder to save the translated command files to They will be saved with an Trans suffixe
204. e through its tabs using the following keyboard shortcuts m CTRL ALT TAB Shifts focus one tab to the right E CTRL ALT SHIFT TAB Shifts focus one tab to the left Although each tab provides a unique function you can save the contents of any Commandspace tab to a command file for subsequent submission to SYSTAT 127 Command Language What Do Commands Look Like Here are some examples of SYSTAT commands XTAB l USE food 2 PLENGTH NONE LIST 3 TAB food brands diets 4 CSTAT 5 BY diets 6 CSTAT MEDIAN MIN MAX MEAN CI 7 BY 8 CORR 9 PEARSON calories fat protein cost BONF 10 SPLOM calories fat protein cost 11 PLOT calories protein LABEL brand 12 The CSTAT command on line 5 produces a set of descriptive statistics for all seven numeric variables in the FOOD data file Line 7 asks for the median minimum maximum means and confidence intervals for all of the variables SYSTAT commands are made up of keywords meaningful to the function that they perform on execution As far as possible all meaningful words associated with a given function are applicable For example CSTATISTICS CSTATS and STATISTICS will all give you descriptive statistics Likewise PLENGTH or DISPLAY will both allow you to specify the length of output produced by a given command A keyword will typically be made of letters of the alphabet and sometimes numbers All other characters like the hyphen and underscore are avoided a space and some other
205. e you use invoke and execute a dialog from the Data Graph Analyze Advanced or Quick Access menus or even from the corresponding DIALOG command 1t 1s added to the list of recently used dialog boxes This list persists across SY STAT sessions so if you consistently use the same set of dialog boxes they are always just a click away Simply click the Recent Dialogs button EN on the Standard toolbar or from the menus choose Utilities Recent Dialogs Selecting an item from the list presents the corresponding dialog box All options and variable lists in the recalled dialog box reflect your specifications from the last use of that dialog However opening a different data file changes the variables available for an analysis and consequently resets all dialog boxes to their default settings SYSTAT automatically updates the list of dialog boxes during your sessions The list contains up to fifteen dialog boxes ordered according to recency of use Each use of a dialog box results in a corresponding entry at the top of the Recent Dialogs list Any other instance of that dialog in the list is removed As a result no dialog box appears in the list more than once If your list contains fifteen entries and you use a dialog box not appearing in the list SYSTAT adds the new dialog to the top of the list and removes the oldest entry Some main dialog boxes require preliminary results before they can be used For instance the Hypothesis Test dialog can
206. ed by a check mark Windows XP Style Grids SYSTAT s Data Variable Editor grid now adopts the current Windows XP theme that is applied to the Windows Desktop Certain grid controls in dialog boxes like Data Transform If Then Let and Data Select Cases also have the same look and feel 7 11 What s New and Different in SYSTAT 13 Trim Leading and Trailing Spaces in String Data You may now control the trimming of leading and trailing spaces in string data as you type modify strings in the Data Editor Check uncheck this option in the Data tab of the Edit Options dialog box Modified Features 12 13 14 15 16 Autocomplete Commands Command arguments options and option values will be autocompleted as they are typed in the Interactive or batch Untitled tab of the Commandspace Arguments may be filenames variable names built in function names or specific key words If filenames or their paths involving spaces are selected then they are automatically enclosed in quotes Function names are automatically suffixed by parentheses Command Coloring Coloring of command keywords is now an optional feature though set by default You may set suppress this option in the General tab of the Edit Options dialog box Also variable names are now colored black and option values are colored green Dialog Boxes The tabbed dialog boxes of SYSTAT now have the tabs arranged vertically This allows more tabs to be eas
207. ed by mammals may be due to the environment they are in or their biological and physical characteristics These studies are used to assess whether physical and biological attributes in mammals play a significant role in determining the predatory danger faced by mammals Potential analyses include regression trees multiple regression and discriminant analysis Regression Tree with DIT Plots The input is USE SLEEPDM TREES MODEL DANGER BODY BRAIN SLO SLP DREAM SLP GESTATE ESTIMATE DENSITY DIT The output is 18 Cases Deleted due to Missing Data Split Variable PRE Improvement DREAM SLP 0 404 0 404 2 BODY 0 479 0 074 3 SLO SLP 0 547 0 068 Fitting Method Least Squares Predicted Variable DANGER Minimum Split Index Value 0 050 Minimum Improvement in PRE 0 050 Maximum Number of Nodes Allowed San Minimum Count Allowed in Each Node s D Number of Terminal Nodes in Final Tree 4 Proportional Reduction in Error PRE 0 547 Node From Count Mean SD Split Variable Cut Value Fit 44 659 380 DREAM SLP 1 200 0 404 14 929 072 BODY 4 190 0 408 30 067 081 SLO SLP 12 800 0 164 257 Applications Decision Tree CREAM BLP lt 120 Chemistry Enzyme Reaction Velocity ENZYMDM data consists of measurements of an enzymatic reaction measuring the effects of an inhibitor on the reaction velocity of an enzyme and substrate Variable Description VELOCITY Reaction velocity SUB CONC Substrate concentration INH_CONC
208. ed directly in the Data editor 4 R sii Value Labels MEE 9 x E 4 SYSTAT OL a Categorical Variables EE USE al Edit 2 201 a Transform ORIZON HORIZON URANIUM ARSENIC BORON BARIUM MOLY SELENIUM VANADIUM SULFATE PO 1 000 6 900 12 000 600 000 200 000 40 000 0 400 100 000 435 000 PO 1 000 21 730 12 200 5 000 000 50 000 150 000 0 400 500 000 795 000 PO 1 000 26 790 11 400 2 000 000 40 000 980 000 0 200 150 000 1 080 000 PO 1 000 56 200 12 700 00 000 100 000 10 000 0 200 100 000 450 000 PO 1 000 25 300 3 000 2 000 000 50 000 25 000 0 100 50 000 1 950 000 PO 1 000 4 420 10 300 300 000 200 000 2 000 0 400 10 000 40 000 Merge Files PO 1 000 29 750 21 400 564 000 49 000 7 000 0 200 111 000 975 000 iD 1D Variable PO 1 000 22 320 19 400 1155000 66 000 40 000 0 200 396 000 490 000 Order of Display PO 1 000 9 480 9 000 300 000 50 000 5 000 0 500 100 000 190 000 By Groups PO 1 000 13 460 6 500 990 000 132 000 26 000 1 900 132 000 95 000 PO 1 000 29 560 10 100 1 500 000 200 000 60 000 0 200 200 000 350 000 Le PO 1 000 13390 8 700 660 000 40 000 13 000 0 200 99 000 270 000 Edat PO 1 000 20 960 9 700 2 000 000 60 000 20 000 0 100 150 000 440 000 PO 1 000 26 670 6 400 990 000 99 000 7 000 0 100 66 000 1 220 000 PO 1 000 52 470 9 700 2 000 000 50 000 75 000 0 400 100 000 250 000 Case Weighting a mee PO 1 000 6 490 63 000 1 500 000 150 000 200 000 2 200 6
209. ed three treatments TREAT Subjects judges are stratified within blocks so the interaction of blocks and treatments cannot be analyzed and the outcome of the analysis is JUDGMENT BLOCKCCD Myers amp Montgomery 2002 This data set contains observations on the yield of a chemical process YIELD at different level combinations of two factors viz time TIME and temperatute TEMP on 14 experimental units However two different batches of raw materials were used The variable BLOCK defines the different batches BOARDS Montgomery 2005 It is an aggregated data set on the number of nonconformities found in 26 successive samples of 100 circuit boards For convenience the sample unit or inspection unit is defined as 100 boards That is although each sample contains 100 boards each sample is considered a sample of size 1 from a Poisson distribution The variables are SAMPLE Identifier DEFECTS A total count of the number of defects in each group of 100 Boards BOD Bates and Watts 1988 Marske created these data from stream samples in 1967 Each sample bottle is inoculated with a mixed culture of microorganisms sealed incubated and opened periodically for analysis of dissolved oxygen concentration The variables are DAYS and BOD BOOKPREF Conover 1999 The data set consists of the number of books sold in a week in 12 bookstores of four booksellers The variables are BOOKS STORE BOOKSELLER 322 Chapter 9 BOSTON
210. eger so we restrict the corresponding token to accept values of this type However t values can be decimal numbers so we only restrict our t value token to be a number instead of a character 176 Chapter 5 The template uses the two tokens to compute the desired statistics In addition the df token is used to generate a function plot and to title the plot The other token amp tval appears as a reference line in the function plot and in the output messages The output using a value of 1 88 for a t distribution having 3 degrees of freedom follows t Distribution with 3 DF 0 4 0 3 Density 0 1 0 0 Area to the left of 1 88 0 922 Two tailed p value 0 157 Example 4 Normal Random Deviates Using Tokens No other distribution has received more attention or been used more often than the normal In keeping with this trend we use tokens to generate random deviates from a normal distribution with a user specified mean and standard deviation The user also indicates the number of deviates to create The final command plots the normal distribution TOKEN amp num TYPE INTEGER PROMPT How many standard normal random observations should be generated 177 Command Language TOKEN mean TYPE NUMBER PROMPT What is the mean for the normal distribution TOKEN amp stdev TYPE NUMBER PROMPT What is the standard deviation for the normal distribution NEW REPEAT amp num LET nrd ZRN amp mean a
211. el files 30 exponential distribution 177 exporting graphics 197 198 F F10 key 220 F9 key 130 File menu 30 importing 30 file paths 243 filenames long names 132 spaces in 132 substituting for tokens 161 173 fonts FORMAT 244 Format 30 Align 30 Bulleted List 30 Collapse Tree 31 insert page breaks 30 Numbered List 30 Format Bar 23 217 formatting toolbar see Format Bar 217 FPATH 245 frequency tables 62 Full screen Viewspace 31 Index G GIF 30 197 global options 234 Glossary 42 GPRINT 201 GRAPH 245 graph panning 32 preview 34 realign frames 32 templates for graph options 180 viewing 29 zooming 32 graph editing Graph Editing toolbar 217 Graph Editor 25 close 34 context menu 34 properties 34 Graph menu 31 annotation 32 Edit 26 Lasso 32 Overlay 32 Realign 32 Zoom 26 Graph Properties dialog 34 graph toolbar 217 graphs 21 animate 27 exporting 197 198 printing 201 saving 193 196 197 grouping variables in scatterplots 59 GSAVE 197 H help 38 examples 40 navigating 38 online glossary 42 Help menu 33 Contents 38 Search 39 Help system 38 Contents 38 Favorites 39 Hide 39 Index 38 Refresh 39 toolbar 39 hot commands 130 HTML format 30 194 I IMMEDIATE 170 insert 30 case 31 image 30 page break 30 insertion 209 integers substituting for tokens 167 175 176 177 interactive tab recalling comm
212. elp related to the dialog box Ifa dialog box has more than one tab you will get help related to the active tab a Resets the selections in the dialog box or active tab to the defaults E 44 Resets the selections for all tabs in the dialog box Source variable list A list of variables in the working data file Only variable types numeric and or string allowed by the selected command are displayed in the source list Target variable list s One or more lists such as dependent and independent variable lists indicating the variables you have chosen for the analysis If an analysis compulsorily requires you to choose variables here you will see lt Required gt in the list If a list is empty all variables in the source list will be used for the analysis Special lists Some dialog boxes display lists with multiple columns where you can input as many rows of input as you desire Such lists can be customized using the two buttons m Insert a new row by pressing the 4 icon m Delete a row by pressing the x icon Pushbuttons Dialog boxes contain pushbuttons for performing the following tasks m Add one or more variables to the desired target list by selecting them and then pressing the corresponding Add gt button Alternatively right click on a variable or selection and select the Add to target list corresponding to the desired target list m Remove one or more variables from a target list by selecting them and then p
213. enus choose File Open Data m Select GDWTRDM and click Open My Recent Documents My Documents My anne My Network B Earnbill syz Pe Eclipse svz Jeducatn syz Jeggs syz B Egyptdra Syz B Ekman Syz F e Ee JEmf syz gt JEneray syz enzymdm syz Jenzyme syz Fe Estim Syz a Euclid sz 27 Euronew sz File name Files of type je SWZ BJEx2 syz BEExd svz BPJExda syvz BEJEx4b syz BFExer syz E Exports syz E a ractory syz Fidell syz Eh rlea syz E gt Pleabeetle syz B Food syz E Forbes syz E Forbesnew syz Forearmi yz qT BjForearm syz BFjFossils S 2 Be Fraction syz Fried SYZ Be Freflydm syz Gauge oye FE Gauge svz 5 6DP syz 5 Gdwtrdrn Sve Grades SY2 E arbOta syz 5Jarbd1b syz Fe GrbOic syz Bjarbo1d syz BE larbOte syz SYSTAT spa spa syd sys ka O amp El B Greoz sy Bhjare syz Be Growth Harddia Head sy 2 Headdim Be Heart2 s B Heart sv Helm syz HilRace hilo syz E gt Histamirn 2 Hobbsm Hosen 5 Hoslemm Cancel 102 Chapter 4 Data files that are opened or imported can be viewed and edited in the Data editor You can also see the results of transform variables select cases and so forth in the Data editor In this example measurements were taken of the levels of urantum and various other elements in the groundwater at each producing horizon The measurements for each variable can be viewed and manipulat
214. er 7 The left side of the status bar will show the status of some output related options QGRAPH Displayed when statistical Quick Graphs are set to appear in the Output Editor Toggle this mode on or off by clicking on it HTM Displayed when HTML based output is set to appear in the Output Editor Click on this to toggle between HTML formatting and plain text formatting of the output PLENGTH NONE SHORT MEDIUM LONG NONE SHORT MEDIUM or LONG is displayed when the corresponding output length is set using the Global Options dialog or the PLENGTH command ECHO Displayed when the commands issued by the user are set to appear in the output Click on it once if you do not want the commands to be echoed VDISP LABEL NAMES BOTH LABEL NAMES BOTH is displayed depending on the global setting for display of variable labels or the VDISPLAY command LDISP LABEL DATA BOTH LABEL DATA BOTH is displayed depending on the global setting for display of value labels or the LDISPLAY command NODE Displayed when detailed node captions are to be shown for the Output Organizer Click on it once to display brief captions PAGE NONE NARROW WIDE NONE NARROW WIDE is displayed depending on the global setting for page width or the PAGE command The middle portion of the status bar will show information about existing processing conditions on the data and also allow you to edit them SEL Displayed when case selection 1s in effect Pause the mouse on
215. er output will appear If an output file is already open it is closed with the option of saving it m Save As Save the file in the active tab as a separate file You will be prompted to specify the name and location of your choice Options Set SYSTAT s global options according to your preferences Note Cut Copy and Delete are available only when a selection has been made Output Organizer The Output Organizer serves primarily as a table of contents for the Output editor Use it to jump to any location in the Output editor without having to scroll through long statistical or graphical results SYSIAI Untitled syo 0 031109 0 041202 0 411447 0 966284 0 14335 0 442653 0 070238 0749086 0 151985 0454701 0 002305 0 111197 0 010796 0 069443 0 148996 KSCORE Karnofsky Score TMRTYPE 1 TMRTYPE TMRTYPE 3 TRIMNT trimntXther apy Treatment KSCORE Karnofsky Score 0 000027 TMRTYPE 1 0 000113 0080137 TMRIYPE Z 0 000296 0 038144 0 068478 TMRTYPE 3 0000187 0041489 0043837 0090811 TRIMNT Treatment 0 000037 0 012387 000837 0012587 0 054846 THERAPY Prior Therapy Status 0 000004 0 000667 0 000713 0 000145 10 007218 0 003074 0 000004 0 000473 0 000150 0 000616 0 004820 0 002304 0 001640 KSCORF Karnofsky Score TIARTYPE 1 TMRTYPE 2 TMATYPE 3 irtrmntXtheragy T KSCORE Karmofsky Score 1000000 TMRTYPE 1 0 076863 1 000000 gt Tootsie Josiagos
216. ercentage average gain touchdown percentage and interception percentage The variables are NAMES Last name and first name of Quaterback ATTEMPTS Passing attempts COMPLETIONS Percentage of completions per attempt 342 Chapter 9 YARDS TDS INTS RATING Average yards gained per attempt Percentage of touchdown passes per attempt Percentage of interceptions per attempt NFL Ratings rounded to the nearest 0 1 NLSe The data used here have been extracted from the National Longitudinal Survey of Young Men 1979 containing information on 200 individuals on school enrollment NOTENR BLACK SOUTH EDUC AGE FED MED CULTURE NSIBS LW IQ FOMY School Enrollment Status 1 if not enrolled 0 otherwise A race dummy 0 for white A region dummy 0 for non South Highest completed grade Age Father s education Mother s education An index of reading material available in the home 1 for least 3 for most Number of siblings Log10 of wage An IQ measure Mean income of persons in father s occupation in 1960 OPERA The following data are from an editorial in The New York Times December 3 1987 They represent the duration HOURS of various plays films and operas 7ITLE ORDEREDOUTPUT Hollander and Wolfe 1999 18 male workers are divided into three groups as receiving no information about output Control receiving a rough estimate Group B and receiving accurate information Group C OURWORLD Variab
217. ered during processing are Variables i Levels A A A A A a ee POINTER 2 levels 1 000 1 000 Dependent Variable READING N 16 Multiple R i 0 000 Squared Multiple R 0 000 264 Chapter 8 Estimates of Effects B ano txry Factor Level READING O A A ery CONSTANT 10 500 POINTER 1 0 000 Analysis of Variance Source Type III SS df Mean Squares F ratio p value Aass a Dn A A Gaa S 4 ee ee ee eee eer POINTER 0 000 1 0 000 0 000 1 000 Error l 168 000 14 12 000 Least Squares Means READING 1 1 POINTER WARNING Case 11 is an Outlier Studentized Residual 2 839 Durbin Watson D Statistic 1 512 First Order Autocorrelation 0 201 Effects coding used for categorical variables in model The categorical values encountered during processing are Variables i Levels A E A E S Ho gt gt VENDOR 2 levels 1 000 1 000 Dependent Variable READING N 16 Multiple R 0 270 Squared Multiple R 0 073 265 Estimates of Effects B coo lx ry Factor Level READING E O A ae Ho gt CONSTANT 10 500 VENDOR 1 0 875 Analysis of Variance Source Type III SS df Mean Squares F ratio p value a 4 ee ee ee ee eee VENDOR 12 250 1 12 250 1 101 0 312 EXFOE i 155790 14 114125 Least Squares Means 15 13 O 11 Z Q mn aa 9 7 7 5 1 1 V
218. erent in SYSTAT 13 1 GENERAL FEATURES 56 000 da daa owed GS l Graphical User terfac s i lt a taa e a Si WAC eos Ge aes aaa dee eee Bee gh oa A a a a 2 Command edo cd e Seok Be dente eae he Syme he ed heeds G08 3 AAA bar doe We a ee ee tk ee oe Be dee 3 MBPS A a tose Ge Aad ee se Se de ee eRe as eee Be 4 STATISTICAL FEATURES 4 CU ora e Aa a 5 DD ad aos od ub as der ed do dd lo e at 9 Command Line Interface 11 QUIDUL e a y dik e A A oe 14 TA ACS a Ge He ods ge ee ps Os We dni A 14 Statistical ESAUTES ueus a oe Bee IA OS Rees ee SS 16 New FCatures ws at Sw Ge eS a a Ad 16 2 Introducing SYSTAT 21 User Intetiace casa a 21 WiEWSDACC 9 in a or we Bem GOS da Hs 22 WOR Da Cera da nde ot Ge la Oh ee os de ek De SL eae SF 21 COMmMMANGSPACe s ekea ep SS Se a ee SE A e i 28 Reorganizing the User Interface 29 MENUS ar Ada e 30 Dialos BOXES e teca io it da Be AE 35 Gota He Pe aS EAS A 38 111 3 SYSTAT Basics 45 Stanton ds eee ee Be ee ee ae 46 Enterme Daid m cara da a Ee tt 47 Reading an ASCII Text File 51 Graphi A II E A 53 SCallerplols y Cada E AE 53 Usine Command spaces med taa Me a E ae EA da 60 Sorting and Listing the Cases 60 A Quick Description essa eat A we ee es 62 Frequency Counts and Percentages 62 Descriptive Statistics aa a ds 66 Statistics BY Gro pP s sacara d ani SESE SO
219. erform an ANOVA the variable used must produce a straight line in a probability plot Clearly the distribution of SURVATD is skewed and must be transformed You can use the Graph Window to reduce the X axis power from 1 through successive exponential power transformation 0 9 to 0 1 and finally to 0 which is same as the log transformation Normal 0 0 1 0 Quantile SURVATD The second plot should appear Since the probability plot is much closer to a straight line we see that a log transformation is appropriate 281 Applications Survival Rates of Melanoma Patients MELANMDM data contains reports on melanoma patients Variable Description TIME The survival time for melanoma patients in days CENSOR The censoring variable WEIGHT The weight variable ULCER Presence or absence of ulcers DEPTH Depth of ulceration NODES Number of lymph nodes that are affected SEX The sex of the patient SEX The stratification variable coded for the analysis Survival studies are used in the area of drug development Survival rates of the patients on an experimental drug are studied to determine the effectiveness of the drug in treating melanoma Sex may be used as a stratification variable to examine the difference in the survival patterns of male and female patients Potential analyses include survival analysis and logistic regression Stratified Cox Regression The input is USE MELNMADM SURVIVAL MODEL TIME ULCER DEPTH NODES C
220. eriod LWT race RACE 1 white RACE 2 black RACE 3 other smoking status during pregnancy SMOKE history of premature labor PTL hypertension HT uterine irritability UI and number of physician visits during first trimester FTV The dependent variable is coded 1 for birth weights less than 2500 grams and coded 0 otherwise These variables have previously been identified as associated with low birth weight in the obstetrical literature The first model considered is the simple regression of LOW on a constant and LWD a dummy variable coded 1 if LWT is less than 110 pounds and coded 0 otherwise See H amp L Table 3 17 LWD and LWT are similar variable names Be sure to note which is being used in the models that follow The input is USE HOSLEM BLOGIT MODEL LOW CONSTANT LWD REF 0 ESTIMATE 2 Binary Logit with Multiple Predictors 2 Binary Logit with Interactions 2 Predicting Bankruptcy in the Telecommunicatio 2 Deciles of Risk and Model Diagnostics 2 Quantiles 2 Multinomial Logit 12 Conditional Logistic Regression 2 Discrete Choice Models 2 By Choice Data Format 2 Stepwise Regression 2 Hypothesis Testing 21 Tacklina different data format in Loaistic Reares Y gt The output begins with a listing of the dependent variable and the sample split between 0 reference and 1 response for the dependent variable A brief iteration history follows showing the progress of the pro
221. es appearing in the token definition are not included in the token value To direct SYSTAT to the correct path we use quotes around the token in the USE command Without those quotes the program would look for a file named program and would return an error Repeated submissions of this template allow rapid creation of exploratory bar charts to study the distributions of variables in the SURVEY2 file Due to the automatic substitution we are not prompted for a data file on each submission To change data files replace the path and the file in the first TOKEN command in the template The note appearing in the output automatically updates to reflect the new file Example 2 Token Substitution for Variables and Strings Variable substitution allows templates to be used for any data file The resulting output has the same general structure but varies in its content String number and integer substitution allows customization giving output from different files unique features 174 Chapter 5 Here we create a three dimensional scatterplot The string tokens provide custom labels and a title to help differentiate the plot from other 3D plots generated from other submissions of this template TOKEN amp xvar TYPE NVARIABLE IMMEDIATE PROMPT Select a variable for the x axis TOKEN amp xXvarlab TYPE STRING IMMEDIATE PROMPT Enter a label for the x axis TOKEN amp yvar TYPE NVARIABLE IMMEDIATE PROMPT Select a variable for the y
222. for a command If a key combination you have typed in the new shortcut key area has already been assigned to some other command then that command will be displayed in the Assigned to area and the Assign button will be disabled Also the new shortcut key area will not register any external keyboard shortcuts since such shortcuts may also be useful while working with SYSTAT In fact pressing such shortcuts will perform the associated external task For instance Alt Tab is a Windows shortcut that lists all open windows allowing you to select one by holding Alt down and repeatedly pressing Tab This functionality offers quick navigation between the SYSTAT user interface and any other program you may be running concurrently Access Key Customization The access key for a menu item 1s indicated by typing an ampersand before the underlined letter in the Button text area of the Button Appearance dialog box You can change the access key to use by moving the ampersand to be just before the desired letter in the caption Take care to see that you do not create duplicate access keys 223 Customization of the SYSTAT Environment Menu Customization SYSTAT has several context menus that pop up on right click in various parts of its user interface Use the Menu tab of the Customize dialog box to customize these menus as well as set a few other options Customize Commands Toolbars Keyboard Menu Application frame menus Co
223. for output data and graph appearance specify file locations for navigational ease define and set themes to suit your needs set the output to appear either based on data files used or in the order of execution of analyses 205 Customization of the SYSTAT Environment Commandspace Customization Users who frequently use SYSTAT s command language may prefer a larger command area for viewing and editing of command files To change the size of the Commandspace hover the mouse on its upper boundary until the mouse cursor changes to a double sided arrow hold down the mouse and drag to a new location The output area 1s automatically resized to accommodate the resized Commandspace Alternatively you can undock the Commandspace from the bottom edge of the user interface to increase the space available for displaying output To do this m Click the upper boundary or sidebar of the Commandspace ensuring that the mouse pointer does not change appearance and drag the outline to a new location without releasing the mouse button Hold down the Ctrl key as you drag to prevent docking with the user interface Release the mouse button and Ctrl key when the outline indicates the desired position m Double click the upper boundary of a docked Commandspace to detach it into its last undocked position Similarly you can dock the Commandspace to its original position m Click the title bar of the undocked Commandspace and drag the outline to
224. fourth allows you to enter the expected frequencies using the keyboard For another illustration of choice tokens try the Simple Correspondence Analysis Plot in the Graph Graph Gallery dialog box Chapter Working with Output Lou Ross revised by Poornima Holla All of SYSTAT s output appears in the Output editor with corresponding entries appearing in the Output Organizer You can save and print your results using the File menu Using these options you can m Reorganize and reformat output Save data and output in text files Save charts in a number of graphics formats Print data output and charts Save output from statistical and graphical procedures in SYSTAT output SY O files Rich Text Format RTF files Rich Text Format Wordpad compatible RTF files HyperText Markup Language HTML files or MHT files You can open SYSTAT output in word processing and other applications by saving them in a format that other software recognize SYSTAT offers a number of output and graph formats that are compatible with most Windows applications Often the easiest way to transfer results to other applications is by copying and pasting using the Windows clipboard This works well for charts tables and text although the results vary depending on the type of data and the target application 185 186 Chapter 6 Output Editor Format The Output editor displays statistical output and graphics You can activate the Ou
225. g output 189 190 resizing 191 set as active data file 34 transformations 190 tree folder 191 viewing 191 208 Output pane P PAGE 244 page setup 200 pairwise comparisons 95 183 PCT 196 Pearson correlations 69 pixels 215 PLENGTH 210 PNG 30 197 Index Portable Network Graphics 197 portrait orientation 200 201 PostScript files 196 predefined tokens 171 file paths 171 printing 199 200 graphs 201 Processing Conditions 25 project directory 243 commom directory 243 PROMPT 168 proportional output 239 PS 30 196 pushbuttons commands 35 dialog boxes 36 Q Quick Access menu 33 Quick Graphs 30 72 240 R random deviates 176 177 recent dialogs 229 Record Script 151 231 regression linear 179 REM 146 reorganizing user interface 29 Reset All buttons 212 Reset button 219 Rich Text Format 194 S SAS files 30 saving filename substitution 161 graphs 193 196 197 output 193 194 results from statistical analyses 195 scatterplot matrices 72 scatterplots 53 3 D 76 grouping variables 59 shortcut keys 220 224 smoothers 55 sorting cases 60 SPLOMs 72 S PLUS files 30 SPSS files 30 Standard toolbar 217 starting SYSTAT 46 Startpage 22 customization 209 STATA files 30 Statistica files 30 statistics toolbar 217 status bar context menu 211 customization 211 hiding 209 viewing 209 stratification 68 strings subs
226. group Each whole plot is split into four subplots and the four fertilizers are applied randomly to these DAYCREDMe Wilkinson Blank and Gruber 1996 This data set consists of three measures of a child s social competence including a measure for behavior at dinner a measure for behavior in dealing with strangers and one involving social problem solving in a cognitive test In addition there 1s a categorical variable for the setting in which a child was raised either by parents by a babysitter or by a daycare center The variables are SETTINGS Daycare setting in which child 1s raised SETTING Coded setting DINNER Behavioral measure of skill during dinner STRANGER Measure of skill in dealing with a stranger PROBLEM Social problem solving skills in a cognitive test DELTIME Montgomery Peck and Vining 2001 The data set deals with 25 delivery times of vending machines The delivery time DELTIME of these machines is affected by the number of cases of product stocked CASES and the distance walked by the route driver DISTANCE DESIGNDM e Devor Chang and Sutherland 1992 The data set consists of the results of an experiment designed to improve the performance of a fuel gauge The variables are RUN The case ID SPRING Dummy variable for the type of spring used POINTER Dummy variable for the type of pointer used 327 VENDOR ANGLE READING Data Files Dummy variable for the vendor used Dummy variable for the ty
227. gt Jackson 1991 These data are a covariance matrix of measures performed on 40 Nike rockets The variables are INTEGRAI PLANMTR1 INTEGRA2 and PLANMTR2 MJ006 Milliken and Johnson 1984 This data set came from an experiment that was conducted to determine how six different kinds of work tasks TASK affect a worker s pulse rate In this experiment 78 male workers were assigned at random to six different groups so that there were 13 workers in each group Each group of workers was trained to perform its assigned task On a selected day after training the pulse rates PULSE of the workers were measured after the workers had performed their assigned tasks for one hour Unfortunately some individuals withdrew from the experiment during the training process so that some groups contained fewer than 13 individuals The recorded data represent the number of pulsations in 20 seconds MJ020 Milliken and Johnson 1984 The data set is from a paired association learning task experiment performed on subjects under the influence of two drugs Group is a control no drug Group2 was given drugl Group3 was given drug2 and Group4 was given both drugs The variables are LEARNING and GROUP MJ129 Milliken and Johnson 1984 The data set is from a small two way treatment structure experiment conducted in a completely randomized design structure MJ166 Milliken and Johnson 1984 A bakery scientist wanted to study the effects of combining thre
228. h that is in the Graph editor and invoke the Edit Options dialog box Output Organizer You can rename tree nodes and folders expand or collapse the entire tree including any tree folders or multilevel nodes insert tree folders create a new output file clear all or save the content in the Output editor and request detailed node captions When a data node is selected you can also set it as the active data file When a text node is selected you can also cut or copy it and the corresponding output in the Output editor to the clipboard paste one or more nodes after copying them to the clipboard or even delete it which will also delete the corresponding content in the Output editor When a graph node is selected you can also view the corresponding graph in the Graph editor Examples You can run the underlying example command file s expand or collapse the entire tree including any sub folders or multilevel nodes When an example node not folder is selected you can also open the underlying command file in the Batch tab of the Commandspace Commandspace Apart from the various options for editing and submitting commands you can right click on the Batch tab to create a new command file open an existing command file save the content of the tab or close the tab In addition to these context menus are available for cells columns and rows in the Data editor command files in the Batch interactive and log tabs of the Commandspace d
229. have occurred together in 50 diseases The variables DIM and DIM2 are the coordinates in two dimensions after performing the multidimentional scaling on the cooccurrences of symptoms for 50 diseases The other variables LYME MALARIA YELLOW RABIES and FLU 5 among the 50 diseases are the dichotomous variables which indicate weather a particular symptom is present or not TABLET Netmaster Statistics Courses An experiment was undertaken to compare two methods HPLC and NIR to ascertain the amount of active content in tablets The tests have been applied to the same set of ten tablets breaking each tablet into two halves and applying one method to each half The resulting data consists of the following variables TABLET HPLC and NIR TABLET2 The data set is the indexed form of data set TABLET 354 Chapter 9 TARGET The data set is hypothetical It describes the success of an arrow throwing machine to hit the target The variables in the data set are NOOFTRAILS Number of trails NOOFEVENTS Number of events HEIGHT Height cms at which the machine is placed FORCE Force newton applied to hit the target TEACH Mickey et al 2004 The data set contains the two teaching methods and three teachers Each teacher uses each teaching method with four different batches of students The performance of each batch is measured by the average score of the batch in a common examination The variables are SCORE TEACHER and METHOD TE
230. he font color style and background color of error messages The default is a crimson color in the regular font style with a white background Warning Specify the font color style and background color of warning messages The default is a shade of brown in the regular font style with a white background Header Specify the font color style and background color of text headings The default 1s a shade of blue in the bold font style with a white background Sub header Specify the font color style and background color of text sub headings The default 1s a shade of blue in the bold font style with a white background 242 Chapter 7 Table caption Specify the font color style and background color of table captions The default is a shade of blue in the bold font style with a white background Table header footer Specify the font color style and background color of the text in table headers and footers The default 1s black color in the bold font style with an off white background Table body Specify the font color style and background color of the text in table body The default is black color in the bold font style with a white background Page background Specify the background color and or image for the entire page The image should be stored in the PNG BMP JPG GIF or EMF format and can be in any location Color Palette To change a color click the corresponding color button click on a pre defined color in th
231. he mercury concentration in a 3 year old fish from the lake AGEDATA Indicator of the availability of age data on fish sampled LNCHLORO Log of CHLORO MULTIRESP Myers amp Montgomery 2002 This data set contains observations on three responses at different level combinations of two factors time TIME and temperature TEMP of a chemical process The three responses are yield YIELD viscosity VISCOSITY and the number average molecular weight MOLWEIGHT The data set also contains coded versions of these variables X7 describes the TIME variables after being used coded and X2 describes TEMP after being coded NAFTA Two months before the North Atlantic Federal Trade Agreement approval and before the televised debate between Vice President Al Gore and businessman Ross Perot political pollsters queried a sample of 350 people asking Are you For Unsure or Against NAFTA After the debate the pollsters contacted the same people and asked the question a second time Variables include BEFORE AFTERS and COUNT NEWARK Collected by the U S Government and cited in Chambers et al 1983 These data are 64 average monthly temperatures TEMP in Newark New Jersey beginning with January 1964 NFL Johnson 1999 The data set is obtained from the NFL for the 1999 2000 season for those players with at least 1 500 passing attempts It is NFL Passer Rating Data RATING is based on performance standards established for completion p
232. hen one or more categorical variables are declared or exist in the data file Pause the mouse on this to see the currently defined categorical variable s in the tooltip that appears Click on CAT to invoke the Data Category dialog box and add delete categorical variables or turn off category declaration The right end of the status bar shows the current condition of the command autocompletion mode and four keyboard states AUTO Displayed when the Commandspace supports autocompletion of commands Click on it to toggle this mode See the Global Options section for details about this feature OVR Displayed when the keyboard is in overstrike mode In this state typed text replaces the text at the current location This gets grayed out when the Insert key on your keyboard is pressed to set it to the insert mode The insert mode allows insertion of new typed text at the current cursor location shifting any existing text to the right CAP Displayed when Caps Lock is active In this state every typed letter appears in upper case Use the Caps Lock key to toggle this state on and off NUM Displayed when Num Lock is active With Num Lock on the keyboard keypad enters numbers With Num Lock off the keypad moves the cursor in the current window The Num Lock key toggles this state on and off SCRL Displayed when Scroll Lock is active With Scroll Lock on if the Data Editor is active and you use the arrow keys on the keyboard the entire sheet
233. ht clicking in the Output Organizer provides some important features These are Rename Rename the selected tree folder Expand All Collapse All Expand Collapse the Output Organizer tree without affecting the output in the Output editor Insert Tree Folder Insert a new tree folder under the active Output Organizer data node You can drag and drop Output Organizer text and graph nodes and other tree folders into this tree folder Set as Active Data File Set the data file as active With more than one data file open in the Output Organizer this gives you the option to work with any previously opened data file as active View Data View the data file corresponding to the selected data file node New Output Open a new output file in the Output editor where further output will appear If an output file is already open it is closed with the option of saving it Clear Output Clear all the output generated in the Output editor so far View Graph View the graph corresponding to the selected graph node in the Graph tab Save As Save the file that is in focus as a separate file You will be prompted to specify the name and location of your choice Show Detailed Captions Show the underlying SYSTAT commands as Output Organizer node captions Saving Output and Graphs You can save the contents of the active tab or pane in a file SYSTAT saves combined statistical and graphical output in four file types In addition individual graph
234. iable Editor 26 Open Multiple Data Files 27 Recode Variables 28 Store and Retrieve Current Settings 3 Deleted Features 29 View Data Files Commands New Features 30 ACTIVE Command 31 Built in Functions 32 FOCUS Command 33 Macros Modified Features 34 FUNCTION Command 35 Multiple Option Values 36 PAGE NONE 37 Precedence Rules 38 String Subscripted Variables 39 Temporary Variables Deleted Features 40 Built In Variables Output New Features 41 Locales and Digit Grouping 42 Node and Link Captions What s New and Different in SYSTAT 13 4 Chapter 1 Graphics New Features 43 44 45 46 47 48 49 50 Sk Color using RGB Values Gradient Colors for Surfaces Label Dots in Dot Summary Charts Built In Colors Colors for overlaid graphs pie and stacked charts Stacked Bar Charts with Grouping Variable Individual Border Displays on Plots Multiple Slices in Pie Charts Numeric Case Labels STATISTICAL FEATURES New Features pe age ees ce et eee ARCH and GARCH Models in Time Series Best Subsets Regression Confirmatory Factor Analysis Environment Variables in Basic Statistics Hypothesis Testing for Multivariate Mean New Basic Statistics Bootstrap Analysis in Hypothesis Testing New Nonparametric Tests Polynomial Regression Modified Features 10 11 Analysis of Variance Crosstabulations 5 GUI What s New and Different in SYSTAT 13
235. ialog box elements status bar and the toolbar area These menus provide shortcuts to various data editing command submission dialog actions status bar content and menu actions respectively 35 Introducing SYSTAT Dialog Boxes Most menu selections in SYSTAT open dialog boxes which you use to select variables and options for analysis Each dialog box may have several basic components in separate tabs ES Regression Linear Least 5quares Model Available variables Dependent Estimation POP 1990 GDP_CAP GDP_CAP Options LIFEEPF Independent s MCDONALD LITERACY Resampling Tabs Since many SY STAT commands provide a great deal of flexibility not all of the possible choices can be contained in a single dialog box The main dialog box usually contains the minimum information required to run a command Additional specifications are made in tabs You can bring the content ofa tab into view by clicking it with the mouse Certain tabs require some input to be given in other tabs before they get enabled A tab may get disabled 1f its contents are irrelevant for the existing selections Command pushbuttons Buttons that instruct SY STAT to perform an action E Runs the procedure for the selections you have made This does not get enabled in some dialog boxes unless the minimum required input is given E Cancels the procedure Any selections you may have made will be discarded 36 Chapter 2 El Displays h
236. ics PNG format You can choose this or any one of the formats BMP JPG GIF and EMF Output Scheme The Output Scheme tab of the Global Options dialog allows you to customize the output format in terms of the font color style regular or bold and background color of various components of the output excluding graphs as well as the page background 241 Edit Options General Data Output Graph File Locations JKK Echo Color Background color Regular Text Color Background color 2 Regular Error Color Background color Regular Wi arming Color Background color Regular Header Color Background color Regular ll il O ca a a ll il O co a a ll Customization of the SYSTAT Environment Sub header Color Background color Regular Table caption Color Background color Regular Table header footer Color Background color Regular Table body Color Background color Regular Page background Page color Page Image C Documents and Settings a El Bold Echo Specify the font color style and background color of echoed commands The default is a shade of teal in the regular font style with a white background Text Specify the font color style and background color of all text The default 1s black color in the regular font style with a white background Error Specify t
237. ihood analysis MLE maximum likelihood estimate MML maximum marginal likelihood MS mean squares MSE mean square error MTW MINITAB v11 data files MU2 Guttman s mu2 monotonicity coefficients N NR Newton Raphson O OC operating characteristic ODBC open database capture and connectivity OLS ordinary least squares P PACE partial autocorrelation function PCA process capability analysis PCE iterated principal axis factoring pdf probability density function PLS partial least squares pmf probability mass function PNG Portable Network Graphics PVAF p v a f present value annuity factor p value probability value Q QC quality control R R amp R repeatability and reproducibility RAMONA Reticular Action Model or Near Approximation ROC receiver operating characteristic RSE robust standard errors RSM response surface methods RTF rich text format S SAV SPSS files Acronyms SBC Schwarz s Bayesian information criterion sc scale SC set correlation SD standard deviations sd2 sas7bdat SAS v9 files SE se S E standard error SETCOR Set and Canonical Correlations SQL structured query language SQRT SQR square root SRWR sum of rank weighted residuals SS sum of squares SSCP sum of squares and cross products SYC CMD SYSTAT command Files SYZ SYD SYS SYSTAT data files SYO SYSTAT output files T TLOSS Taguchi s Loss Function T
238. ilevel models in educational and social research London Griffin Greco W R Priore RL Sharma M Korytnyk W 1982 ROSFIT An enzyme kinetics nonlinear regression curve fitting package for a microcomputer Computers and Biomedical Research 15 39 45 Green P F and Carmone F J 1970 Multidimensional Scaling and related technique in marketing analysis Boston MA Allyn and Bacon Greenacre M J 1984 Theory and applications of correspondence analysis New Y ork Academic Press Gujarati D N 1995 Basic Econometrics 3th ed New York McGraw Hill Gujarati D N 2003 Basic Econometrics 4th ed New York McGraw Hill Hand D J Daly F Lunn A D McConway K J and Ostrowski E Editors 1996 A handbook of data sets London Chapman amp Hall Harman H H 1976 Modern factor analysis 3 d ed Chicago University of Chicago Press Hartigan J A 1975 Clustering algorithms New York John Wiley amp Sons Hedeker D and Gibbons R D 1996 MIXREG a computer program for mixed effects regression analysis with autocorrelated errors Computer Methods and Programs in Biomedicine 49 229 252 Helm C E 1959 A multidimensional ratio scaling analysis of color relations Technical Report Princeton University and Educational Testing Service June 1959 Hocking R R 1985 The analysis of linear models Monterrey CA Brooks Cole Hocking R R 2003 Methods and Applications of Linear
239. ily accessible with just a single click of the mouse Rescue Report SYSTAT now attempts to restore a session that has just crashed Also 1f you click Send Report the rescued files are automatically attached to the email message Shortcut Keys SYSTAT now has the following new shortcut keys provided by default Ctrl Q Quit SYSTAT Alt backspace Undo Ctrl Alt Enter Variable Properties Ctrl K View Workspace See the section on Keyboard Shortcuts in Chapter 7 Customization of the SYSTAT Environment for a complete list 8 Chapter 1 17 Status Bar The following enhancements have been made to the Status Bar m The page width can be set to Narrow Wide or None by clicking PAGE on the Status Bar m The states of the Insert Caps Lock Num Lock and Scroll Lock keys on the keyboard can be toggled through the Status Bar See Chapter 7 Customization of the SYSTAT Environment for a complete list of items on the Status Bar 18 Themes The following enhancements have been made to SYSTAT s Themes feature m Download Themes now has a dialog box interface wherein you may choose which themes to install m Theme files now have versions so that you will have the option to upgrade your theme file whenever a newer version is available on the SYSTAT server m When you apply a theme you will be prompted to save the current theme Deleted Features 19 Open Multiple Graphs View and Active Modes It is no longer possible
240. indows clipboard or from a command file list You can save output in the SYSTAT syo or HTML mht formats You can also define page and printer settings preview and print the content of the Output editor or Data editor and Graph editor Graphs can be reviewed using the Page Mode under the View menu When the Graph Editor is active you can also export and print graphs You can export graphs in a variety of formats including WMF PS EPS BMP JPEG GIF TIFF PNG and PCT The File menu can also be used to open recent data commands and output files Edit Use the Edit menu to undo redo a few steps paste clipboard content to the active pane define output related settings like ID variables order of display of data values and display of variable as well as value labels change SYSTAT options including variable display order in dialog boxes the algorithm to be used for random number generation the behavior of the Enter key in the Data editor font characteristics for output data and graphs display of statistical Quick Graphs inclusion of command syntax in the output and measurement units for graphs reduction or enlargement of graphs and file locations Output editor In addition to the above options when the Output editor is active you can undo redo a few steps of output cut copy and paste statistical output and other text from and into the Output editor find and replace text strings clear text and output change font character
241. ing done to a graph highlight a point in a plot to view the corresponding case in the Data editor choose the region or lasso selection tools and show or hide any selection made using these tools in the plot Analyze Use the Analyze menu to run fundamental statistical analyses including crosstabulation column and row basic statistics and stem and leaf plots fitting distributions correspondence analysis loglinear models nonparametric and multinormal tests hypothesis testing univariate tests and Hotelling s T square tests simple as well as set and canonical correlations Cronbach s alpha linear and robust regression methods logistic regression probit analysis two stage least squares mixed as well as nonlinear regression methods nonparametric smoothing univariate and multivariate analysis of variance general linear models mixed models discriminant classical and robust cluster as well as factor analyses exploratory and confirmatory plotting transforming and smoothing time series autocorrelation and cross correlation functions seasonal adjustment ARIMA ARCH tests GARCH trend analysis and Fourier transformation Advanced Use the Advanced menu to perform advanced statistical analyses like missing value analysis quality analysis including Pareto Box and Whisker various control charts like Shewhart and X MR ARL and OCC computation and process capability analysis nonparametric Cox and parametric survival analysis
242. ing type 2 The comfort score of each person was recorded at the end of one hour SCORE 1 two hours SCORE 2 and three hours SCORE 3 MJ379 Milliken and Johnson 1984 An experimenter wanted to study the effects of three different herbicides HERB and four fertilizers FERT on the growth rate of corn Fifteen plots of land PLOT were available for the experiment and 5 plots were randomly assigned to each of the three herbicides Each of the 15 plots were further divided into 4 subplots and a different fertilizer treatment was randomly assigned to each At the beginning of the third week 10 plants were selected at random from each subplot And the height of each plant was measured The average of the 10 heights HEIGHT was recorded as the measurement from the subplot Unfortunately before any measurement could be taken 3 of the 15 whole plots were destroyed by excessive rainfall Herbicide 1 had been assigned to two of those subplots and herbicide 3 to the third MJ385 Milliken and Johnson 1984 These data form a small part of an experiment conducted to determine the effects of a drug on the scores obtained by depressed patients on a test to measure depression Two patients were in the placebo group and three in the drug group The variables are SCORE WEEK PATIENT TREATS MOTHERS Morrison 2004 These data are hypothetical profiles on three scales of mothers SCALE 1 to SCALE 3 in each of four socioeconomic classes CLASS
243. ingle dinner range from 160 to 550 with an average around 300 303 214 to be exact VITAMINA ranges from 0 to 100 with a mean of 18 9 Since the mean is not close to the middle of the range the distribution must be quite skewed or have a few extreme values Statistics By Group You can use By Groups on the Data menu to stratify the analysis E From the menus choose Data By Groups In the By Groups dialog box select DIET as the variable and click OK Return to the Basic Statistics dialog box Select the following measures N Minimum Maximum Arithmetic mean AM Cl of AM and Median m Click OK Results for DIET yes CALORIES FAT PROTEIN VITAMINA CALCIUM Ss a a A oe fe ee es Ss oe A ee ee E ee Ces es 4 ee ee eee ee N of Cases 13 000 T3000 13 000 13 000 13 000 Minimum 160 000 0 000 9 000 0 000 2 000 Maximum 280 000 8 000 24 000 30 000 30 000 Median 240 000 4 000 17 000 15 000 8 000 Arithmetic Mean 230 769 3 885 16 846 15 077 9 769 95 0 Lower Confidence Limit 209 769 2 544 14 225 7 921 4 629 95 0 Upper Confidence Limit 251 770 DL AD 19 467 224233 14 910 IRON COST WA a Se ET Ho gt gt N of Cases 13 000 13 000 Minimum 2 000 2 000 Maximum 15 000 2 990 Median 8 000 2 490 Arithmetic Mean 8 923 2 509 95 0 Lower Confidence Limit 5 999 2 265 95 0 Upper Confidence Limit 11 847 2 754 Results for DIETS no CALORIES FAT PROTEIN VITAMINA CALCIUM
244. inity of the lake mg L as Calcium Carbonate Measured PH of the lake Measured Calcium of the lake mg l Measured Chlorophyll of the lake mg l Average mercury concentration parts per million in the tissue of the fish sampled from the lake Number of fish sampled in the lake Minimum mercury concentration in sampled fish from lake Maximum mercury concentration in sampled fish from lake Regression estimate of the mercury concentration in a 3 year old fish from the lake Indicator of the availability of age data on fish sampled Log of CHLORO Mercury is a toxic element Its presence in the environment arises from pollution and it subsequently becomes part of the food chain creating potentially harmful effects for both animals and humans Understanding the level and causes of contamination of the environment by such pollutants is an important problem in environmental science Potential analyses include descriptive statistics variance and distribution transformations correlation and regression The input is Regression of Standard Mercury Level on Lake Alkalinity USE MRCURYDM PLOT STDMERC ALKLNTY ELL SMOOTH LINEAR BORDER DOX FILL 1 XLAB Alkalinity YLAB Mercury TITLE Measured Mercury Levels in Freshwater Fish vs Alkalinity COLOR 3 FCOLOR 2 270 Chapter 8 The output is Measured Mercury Levels in Freshwater Fish vs Alkalinity omy fro o oof o 50 100 150 Alkalinity The Graph Window can
245. irregularly distributed data points ACM Transactions on Mathematical software Allison and Cicchetti 1976 Sleep in mammals Ecological and constitutional correlates Science 194 732 734 Anderson E 1935 The irises of Gaspe peninsula Bulletin of the American Iris Society 59 2 5 Andrews D F and Herzberg A M 1985 Data A collection of problems from many fields for the student and research worker New York Springer Verlag Ansfield F Klotz J and the central Oncology Group 1977 A phase III study comparing the clinical utility of four regiments of 5 fluorouracil Cancer 39 34 40 Atkinson A C 1986 Aspects of diagnostic regression analysis Statistical Science 1 397 402 Automotive Industry Action Group 1995 Statistical process control SPC reference manual Chrysler Corporation Ford Motor Company General Motors Corporation Barnett V D and Lewis T 1967 A study of low temperature probabilities in the context of an industrial problem Journal of the Royal Statistical Society Series A 130 177 206 Bates D M and Watts D G 1988 Nonlinear regression analysis and its applications New York John Wiley amp Sons Beckman R J Nachtsheim C J and Cook D J 1987 Diagnostics for mixed model analysis of variance Technometrics 29 413 426 Belsley D A Kuh E and Welesh R E 1980 Regression diagnostics Identifying influential data and sources of collinearity New Yo
246. ist and the recorded set of commands when you open the User Menu Profile dialog subsequently For more information about this feature see Command Language To access a menu item created using the Add Delete Modify dialog or Record Script feature from the menus choose Utilities User Menu Menu List and under this the corresponding menu item name Clicking the name will execute the underlying set of commands 232 Chapter 7 Keyboard shortcuts Any user menu item can be accessed using the keyboard by pressing the underlined number preceding 1ts name the full sequence would be ALT U U L the underlined number Themes The themes feature of SYSTAT allows you to create store and apply any number of fully customized interface themes each with its own set of menu items and toolbars as well as the position and size of spaces content of the status bar and keyboard shortcuts These will be very useful if you do not need some of the menu items at all If you are comfortable with a different menu arrangement or terminology work with just a subset of all the data processing analyses and graphing techniques available in SYSTAT or work with one of several sets of features that you will need at various times For instance if you conduct various courses in Statistics starting from a basic course to an advanced one execute projects catering to various industries or do research in various application areas like Psychology Engineering o
247. istics including color and size create numbered and bulleted lists outdent indent text align text tables and graphs insert images and page breaks into your output and collapse expand links created by graphical and statistical procedures Data editor When the Data editor is active you can also undo redo up to 32 data editing operations cut copy and paste data from and into the Data editor add 31 Introducing SYSTAT empty rows in a new or existing data file insert delete cases and variables find a specific variable find replace occurrences of a string or number in any given column and go to a desired cell Graph editor When the Graph editor 1s active you can also copy graphs Output Organizer When the Output Organizer is active you can also cut copy paste and insert tree folders set the selected data file node as active rename nodes expand collapse trees and see detailed node captions View Use the View menu to view or hide the Workspace Commandspace Startpage processing conditions toolbars and status bar make tabs active and launch a full screen view of the Viewspace This menu also allows you to create and customize toolbars keyboard shortcuts and context menus When the Output editor 1s active you can also view graphs as frames only When the Graph editor is active use the View menu to switch between the Graph View and Page View and turn the display of rulers and graph tooltips on and off Data U
248. l they cannot be abbreviated Interpreting common commands Some commands like STANDARDIZE perform different functions within and outside modules Such commands will be interpreted based on a certain priority order BASIC commands commands related to the module currently loaded if any and then the rest of the commands If you want to use a global command a command that is globally available irrespective of the module loaded when a module is loaded then you have to issue EXIT to exit from the module Retrieving commands SYSTAT holds the most recently processed command lines in memory From the Interactive tab of the Commandspace use the Up arrow or F9 key to scroll through the commands Press Up arrow or F9 once to recall the previous command press it again to see the command before that and so on To define the number and source of commands to retain in memory set Command buffer options in the General tab of the Edit Options dialog box Continuing long commands onto a second line To continue a command onto another line type a comma at the end of the line For example typing CSTAT urban babymort pop 1990 MEAN SEM MEDIAN 1s the same as COTAr urban babymort pop 1990 MEAN SEM MEDIAN 132 Chapter 5 Do not use a comma at the end of the last line of a command this will cause SYSTAT to wait for the rest of the command Also one word cannot be typed into two lines for example USE OUR WORLD Or US E OURWOR
249. l C for copy apply to the tab and or pane that has the focus To bring a pane into focus click any of its constituent tabs To bring a tab into focus click it with the mouse or select its name from the View menu The user interface provides menus for running statistical analyses and producing graphs It also contains toolbars to provide quick access to many standard statistical techniques and graphs 21 22 Chapter 2 SYSTAT Startpage A A a aa a te ex Recent Command Files oF Pome Carlo i Recent Output Files To edit the graph format scales thickness etc D Quality Analysis from the menus choose Edit gt Options Or press F6 ay Exact Tests to open the Edit Options dialog box Then dick the Graph tab Themes Current Theme Default Classic Default gt Next Tip Introductory_Statistics Market_Research f Medical_Statistics Scratchpad MYSTAT Manuals A GettingStarted pdf F Data pdf Graphics pdf Statistics_1 11 111 1V pdf LanguageReference pdf MonteCarlo pdf QualityAnalysis pdf ExactTests pdf v Viewspace The Viewspace consists of four components Startpage Output editor untitled syo upon opening Data Variable editor untitled syz upon opening Graph editor graph1 when graph is in the Output editor Startpage The Startpage is typically the first tab in the Viewspace and it is divided into five panes Recent Files c
250. lay the first 30 You have to associate each menu item you define to either of the following File Displays the SYSTAT command filename if any associated with the currently selected menu item name To specify a different filename or when you are defining the menu item for the first time type the name of a command file including its path or press the l button and browse for it User input Displays the set of commands if any associated with the currently selected menu item name Edit existing commands or type a new set of commands just as you would in the Commandspace You may want to type one or more DIALOG commands here that would pop up frequently used dialog boxes or a command template that you could apply on various data files Status bar Displays the status bar help content currently associated with the selected menu item You can edit existing content or type new content Tooltip Displays the tooltip that will appear on mouse hover if the selected menu item is placed on a toolbar You can edit an existing tooltip or type a new one Bubble Help Displays the Bubble help content currently associated with the selected menu item You can edit existing content or type new content An alternative way of creating a user menu item is by using the Record Script feature This feature automatically creates a menu entry 1f you request it to do so and associates it with the command scripts it has just recorded You can see the menu item l
251. lectively Specify PLENGTH NONE and then individually specify the items you want to print To control Width select Narrow 80 77 82 characters wide in the HTML Classic format for a font size of 10 or Wide 132 106 113 characters wide in the HTML Classic format for a font size of 10 or None This applies to screen output how output is saved and printed The wide setting is useful for data listings and correlation matrices when there are more than five variables Selecting None prevents tables from splitting no matter how wide they are To control Width select Narrow 80 characters wide or Wide 132 characters wide This applies to screen output how output is saved and printed The wide setting is useful for data listings and correlation matrices when there are more than five variables Default font You can specify the font used in the output m Proportional output sets the font and font size for the HTML based output Monospaced output sets the font and font size for output appearing in the classic style and any output requiring fixed width font that facilitates automatic alignment of text like stem and leaf diagrams Wrap text in tables The text written in tables can be sometimes very long especially when variable and or value labels are defined In such cases by default in each cell the text will be wrapped into multiple lines if they extend beyond 15 characters Row headers will be wrapped if they extend
252. les recorded for each case country include COUNTRY URBAN Names of the 95 countries used in this data file Percentage of population living in urban areas LIFEEXPF LIFEEXPM _ Years of life expectancy for females and males GDP Group variable with codes Developed and Emerging GDP CAP Gross domestic product per capita in U S dollars BABYMORT BABYMTS2 a oe E oo mortality rate for 1990 BABYMT82 infant mor BIRTH RT Number of births per 1000 people in 1990 DEATH RT Number of deaths per 1000 people in 1990 BIRTH 8 amp 2 DEATH 82 Number of births and deaths per 1000 people in 1982 343 B TOD HEALTH EDUC MIL HEALTH84 EDUC 84 and MIL 84 POP 1983 POP_1986 POP 1990 POP 2020 GNP_82 GNP_86 RELIGIONS GOVS LEADERS LITERACY GROUPS URBANS MCDONALD LAT LON B TO D82 LOG GDP LIFE EXP Data Files Birth to death ratio in 1990 Expenditures in U S dollars per person for health education and the military in 1990 and in 1984 Populations in millions for the years 1983 1986 and 1990 POP 2020 1s the population projected by the United Nations for 2020 Gross national product in 1982 and 1986 Expenditures grouped by the religion or personal philosophy of those who govern the country Type of government Religion of the leaders of countries Percentage of the population that can read Europe Islamic or the New World Rural or urban Number of McDonald s restaurants per country Lati
253. lis Test Statistic 25 758829 p value is 0 000003 assuming Chi square Distribution with 2 df Y Nonparametric Kruskal Wallis test Kruskal Wallis One way Analysis of Variance for 57 Cases The categorical values encountered during processing are GROUP 3 levels NewWorld Dependent variable URBAN Grouping variable GROUPS Count Rank Sum 19 765 000000 16 198 000000 NewWorld 21 633 000000 Kruskal Wallis Test Statistic 25 758829 p value is 0 000003 assuming Chi square Distribution with 2 df CONOVER INMAN test Test for all pairwise comparisons GROUP Statistic Europe Islamic 6 787470 9 913351E 009 Europe NewWorld 2 639582 0 010878 J Islamic NewWorld 4 421699 0 000049 a Done QGRAPH HM EO E CE Ro fa CAT OR Car NUM Pe You can hide or view the entire Output Organizer without resizing it by selecting View Workspace Although the Output Organizer may be hidden the subsequent output still generates entries in the tree Consequently you can jump quickly to a specific output by reopening the Workspace and clicking on the entries Workspace settings persist across SY STAT sessions For example if you hide the Workspace and close SY STAT the next SY STAT session begins with the Workspace hidden To view the entire Viewspace in the full screen mode from the menus choose View Full Screen Viewspace 193 Working with Output Output Organizer Right Click Menu Rig
254. lso the peaks of the normal curves which represent the mean for a normal distribution are very close to the median values This indicates that the distributions are symmetric and have approximately the same spread variance This is not true for CALCIUM These distributions are right skewed and possibly should be transformed before analysis The mean values for PROTEIN are the same as those in the By Groups statistics 22 133 and 16 846 The standard deviations differ little 4 307 and 4 337 confirming what we observed in the box plots This means that we can use the results of the pooled variance t test printed below the means This test is usually the first one you see in introductory texts and assumes that the distributions have the same shape that is the variances do not differ For PROTEIN we conclude that the mean of 22 1 for the regular dinners does differ significantly from the mean of 16 8 for the diet dinners t 3 229 p value 0 0003 The separate variance test does not require the assumption of equal variances Considering the distributions for CALCIUM displayed in the box plots and that the standard deviations for the groups are 12 757 and 8 506 we use the separate variance t test results We are unable to report a difference in average CALCIUM values for the regular and diet dinners 0 501 p value 0 621 The discussion of SYSTAT s procedures is very exploratory at this stage so you should not conclude that CALCI
255. ly weighted moving average F G GARCH Generalized Autoregressive Conditional Heteroskedaticity GG Greenhouse Geisser GIF Graphics Interchange Format GLM generalized linear models GLS generalized least squares GMA geometric moving average GN Gauss Newton method H H amp L Hosmer and Lemeshow H L trace Holding Lawley trace HTML hyper text markup language I IIDMC independently and identically distributed Monte Carlo IMPSAMPI importance sampling integration IMPSAMPR importance sampling ratio IndMH Independent Metropolis Hastings INDSCAL individual differences scaling INITSAMP initial sample ITER iterations J JB Jarque Bera JMP JMP v3 2 data files JPEG JPG joint photographic experts group K K M Kaplan Meier K S test Kolmogorov Smirnov test KS1 one sample Kolmogorov Smirnov tests KS2 two sample Kolmogorov Smirnov tests L LAD least absolute deviations LCL lower control limit LMS least median of squares LM Test Lagrange Multiplier Tes LR likelihood ratio LRDEV likelihood ratio of deviate LW Lawless and Wang M MA moving average MAD mean absolute deviation MANCOVA multivariate analysis of covariance MANOVA multivariate analysis of variance MAX maximum MC Test McLeod Li Test MCMC Markov Chain Monte Carlo MDS multidimensional scaling MIN minimum M H Metropolis Hastings ML Maximum Likelihood MLA maximum likel
256. m is to use it You can ask for help in any of these ways m Click the button ina SYSTAT dialog box This takes you directly to a topic describing the use of the dialog box This is the fastest way to learn how to use a dialog box m Right click on any dialog box item and select What s this to get help on that particular item m Hover the mouse on a menu item that would have opened a dialog box and press F1 to get help on that particular dialog box Select Contents or Search from the Help menu For help on any term or phrase that is listed in the Help Index from the command prompt on the Interactive tab of the Commandspace type HELP phrase The quotes are required only if the phrase contains spaces This is very useful if you need help on SYSTAT commands Refer the Command Language chapter for details Alternatively type the term or phrase in any tab of the Commandspace right click on it and select HELP phrase You will need to select the whole phrase before you right click 1f it contains spaces Navigating the Help System The SYSTAT Help system has the following tabs Contents The Contents button takes you to the table of contents of the Help system Double click book icons in the Index listing to view the contents of that section Selecting a topic with a page icon opens the associated Help topic m Index Provides a searchable index of Help topics Enter the first few letters of the term you want to fin
257. m the menus choose View Full Screen Viewspace Alternatively right click in the toolbar area and select Full Screen Viewspace However some output may still require scrolling When resizing alone cannot create an area large enough to view your output consider hiding elements of the user interface such as the Commandspace or the Workspace Startpage Customization You can resize the partitions of the Startpage by positioning the mouse over any of the boundaries until the cursor changes to a double line 1 clicking and then dragging the boundary to the desired position You can close the Startpage for the remainder of the session by clicking the View menu and selecting Startpage by right clicking on its tab and selecting Close or by right clicking on the toolbar area and selecting Startpage You can even prevent the Startpage from appearing in subsequent sessions by unchecking the Show at startup check box in the Startpage Status Bar The status bar appears at the bottom of the user interface For Help press F1 HTM QGRAPH OVR NUM When the mouse pauses on a toolbar button or menu entry including right click menus the status bar displays a brief description of that item These descriptions help guide you to the most appropriate procedure for a desired task When the Graph Editor is active with a graph in it the status bar displays the name of the graph element on which the mouse pointer is currently positioned 210 Chapt
258. ma q 2q1 p q p 1 q Beta 2n oth Os Dl hye ti y qe p Beta 2ny Ngo tnig tP 2noo tgo Ngo y For generating random samples from p and q the generated value from the beta distribution 1s to be multiplied with 1 q and 1 p respectively Since 1t 1s not possible in our system to implement this let us consider p Beta 2n Pi Pig E ngog Taio Fig 7 q Betal2n tio tilg tpa MoeT izg thig 7 and whenever p and q appear in other full conditionals p is replaced by 1 q p and q is replaced by 1 p q Take a 2 B 2 and y 2 273 Applications Gene Frequency Estimation using Gibbs Sampling The input is FORMAT 10 5 MCMC TMP N1 182 TMP N2 60 TMP Pl 0 04762 TMP P2 0 31034 TMP B1 240 TMP B2 550 TMP D1 83 TMP D2 550 GVAR NAA 40 NBB 5 P 0 1 0 0 5 FUNCTION TMP FC1 TMP NAA NRN N1 P1 ENDFUNC FUNCTION TMP FC2 TMP NBB NRN N2 P2 ENDFUNC FUNCTION TMP FC3 TMP P BRN B1 B2 ENDFUNC FUNCTION TMP FC4 TMP Q BRN D1 D2 ENDFUNC SAVE GIBBSGENETIC GIBBS FCOND FC1 FC2 FC3 FC4 SIZE 10000 NSAMP 1 BURNIN 1000 GAP 1 RSEKED 1783 USE GIBBSGENETIC LET PP 1 Q01 P1 LET QQ 1 P1 Q1 LET RR 1 PP 00 LET RBEP 1 00 NAA14 1824 174 2 NAA14 182 17 2 2 176 182 60 NAA1 NBB1 2 LET RBEQ 1 PP NBB1 604 17 2 NBB1 60 17 2 2 176 182 60 NAA1 NBB1 2 LET RBER 1 RBEP RBEO STATISTICS PP QQ RR RBEP RBEQ RBER MAXIMUM MEAN MED
259. mand file list that is currently selected in the Command File List dialog The display contains only the filename not the path As a result some lists may contain multiple entries with the same name but which invoke different command files Using unique names for command files avoids this potentially confusing situation Selecting a file from the displayed list submits the corresponding file for processing The commands contained in the file do not appear on the middle tab of the Commandspace file submission does not affect this tab As a result you can have a command file open and submit a second file at the same time 229 Customization of the SYSTAT Environment Command file lists and the list of recent command files appearing on the File menu offer similar functionality but differ in several notable ways First command file lists allow you to group your files into categories whereas file lists based on recency of use do not Second you can create multiple command file lists each having an unlimited number of entries The recent command list allows only nine entries Third the structure of command file lists persists across sessions but lists of recent files change each time you open a file Finally command file lists submit the selected file for processing The recent file list merely opens the file on the middle tab of the Commandspace Recent Dialogs SYSTAT provides quick easy access to frequently used dialog boxes Every tim
260. mbinations but only 7 are shown here The brands for the diet dinners differ from those for the regular dinners 66 Chapter 3 You may want to display frequencies for two factors as a two way table Let us deselect the List layout feature and look at DIET by FOODS m From the menus choose Analyze Tables Two Way E Select DIETS as the row variable and FOOD as the column variable m Deselect List layout click the check box to deselect it if it is currently selected and select Frequencies from the table box Counts DIET rows byFOOD columns beef chicken pasta Total SA ee Ho gt gt no 6 6 3 15 yes i 0 8 5 13 A ree Total 6 14 8 28 We failed to get any beef dinners in the DIETS yes group Descriptive Statistics It is easy to request a panel of descriptive statistics However since we have not examined several of these distributions graphically we should avoid reporting means and standard deviations these statistics can be misleading when the shape of the distribution is highly skewed It is helpful to scan the sample size for each variable to determine whether values are missing The basic statistics are number of observations N minimum maximum arithmetic mean AM geometric mean harmonic mean sum standard deviation variance coefficient of variation CV range median standard error of AM etc E From the menus choose Analyze Basic Statistics m In
261. me pair of experimenters to reduce variability in the experiment Each pair of experimenters was treated as a block The variables are ORDER The overall performance order of the trial BLOCK The subject and experimenters block number HEIGHT 0 if step at the low 5 75 height 1 1f at the high 11 5 height FREQUENCY eas es 0 if slow 14 steps min 1 if medium 21 steps min 2 if high RESTHR The resting heart rate of the subject before a trial in beats per minute HR The final heart rate of the subject after a trial in beats per minute HELM Helm 1959 reprinted by Borg and Lingoes 1987 These data contain highly accurate estimates of distance between color pairs by one experimental subject CB Variables include A C E G I K M 0 O and S 334 Chapter 9 HILLRACE Atkinson 1986 The data set gives the record winning times TIME for 35 hill races RACES in Scotland The distance DISTANCE travelled and the height climbed CLIMB in each race are also given The variables are RACE DISTANCE CLIMB TIME Name of the Race Distance covered in miles Elevation climbed during race in feet Record time for race in minutes HILO These are hypothetical price data for a stock HIGH is the highest price for that month MONTH and MONTHS LOW is the low price and CLOSE is the closing price at the end of the month HISTAMINE Morrison and Zeppa 1963 It consists of data having a multivariate layout In
262. me will be treated as coming from the lowest numbered class possible For example consider the following commands used to draw a bar chart of the INCOME variable in the SURVEY data file USE SURVEY2 BAR INCOME COLOR BLUE In general the COLOR option accepts either a color name like RED BLUE YEL LOW and so on or a variable name as option value Incidentally BLUE 1s also a variable in the data file As color names belong to Class 0 in the above precedence rule whereas file variable names belong to Class 3 SYSTAT interprets BLUE as the color name If you need to set COLOR to the variable name BLUE rename the variable and then use it as the option value of COLOR The command script to do this is as follows 135 Command Language USE SURVEYZ LET BLUE2 BLUE BAR INCOME COLOR BLUE2 Shortcuts There are some shortcuts you can use when typing commands Listing consecutive variables When you want to specify more than two variables that are consecutive in the data file you can type the first and last variable and separate them with two periods instead of typing the entire list This shortcut will be referred to as the ellipsis For example instead of typing CSTAT babymort rte exp gnp 82 gnp 86 gdp Cap you can type CoTAT Dabymotk se gap cap You can type combinations of variable names and lists of consecutive variables using the ellipsis Multiple transformations the sign When you want to perfo
263. ments 3 Desktop My Documents as My Computer My H etwork 92 Accident syz B Accidents sy2 a E Adjadaptor syz Admire sy z Admit sz 5 Aerosol syz Afifi syz Agel syz File name Files of type Bas Agesex syz 9 Agestat sy2 agri yz BE agrz syz agree syz BE iag syz gt aircraft syz 5 airline syz Fas Akima syz Be 4m syz 2 Anneal syz E ansfield syz 92 Amiety syz Ba Bank svz Bankruptcy s z EhjBarley syz Ebd yz ES Bigworld syz SE lBirths2 syz SE eirths svz Beit syz BE Block svz BE Blockecd svz Boards syz Plead S 2 Bookpref syz E JBoston syz B Boxcox AZ Boxes SWZ BE Boxesdm SYZ SYSTAT Data syz syzu sud sys bl cancer1 Cancer il lars Syz Fa Cement Be chardat gt Use the SAVE type for saving output data or graphs to files For example TOKEN amp gphfile TOKEN amp 0utfile PLOT YX GSAVE amp gphfile OSAVE soutfile TYPE SAVE TYPE SAVE BMP HTML 163 Save Specify a file for the token Egphfile E Accidents syz Fe adaptor Syz E aridstati Syz E addstat2 syz F My Recent Documents Mly Computer File name Save as lype My Netmork Afifi syz Aqel swz AQesex syz Agestab syz Aqrl swz Agre s z gree sz Aag sz Aircraft syz Airline syz Akima syz Am syz Anneal syz Command Language BE Bod Syz poole 2 ansfield syz Bes Anciety SVZ Pb Bank syvz
264. mized tree folders Use customized trees to place output from several procedures in one location To insert a new tree folder from the menus choose Edit Output Organizer Insert Tree Folder Alternatively you can right click on the Output Organizer and select Insert Tree Folder SYSTAT creates a folder named New Folder To rename it select the folder and go to Edit Output Organizer Rename Alternatively right click on the folder and select Rename Headings appear just below and at the same level as the selected Organizer entry Y ou can rename any Output Organizer entry collapse expand all trees from Output Organizer in the Edit menu or from the right click menu of Output Organizer You can also view a data from the right click menu of Output Organizer Configuring the Output Organizer Output Organizer headings are often truncated at the right edge of the pane To view the entire heading move the mouse over the heading Alternatively you can resize the Workspace by dragging the boundary between the Viewspace and Workspace to new locations Position the pointer of your mouse over the boundary until a double headed arrow appears Click your left mouse button and hold it down while you drag the pane edge to the desired location 192 Chapter 6 SYSTAT Untitled syo aca 44 SYSTAT Output ann amp nzu amp EZRA KRUSKAL URBAN GROUPS KRUSKAL URBAN GROUPS INMAN Kruskal Wal
265. mp Sstdev DENSITY nrd NORMAL This template writes the generated deviates to a new variable named NRD Alternatively you could use another token to prompt the user to specify a name for the new variable Example 5 Random Number Generation Using Tokens In this example we combine interactive and automatic token substitution to generate random deviates from one of four distributions Uniform Normal Exponential or Logistic TOKEN amp rndnum rndnum TOKEN RN RN TOKEN amp dist TYPE STRING IMMEDIATE PROMPT Select a distribution by entering a letter U Uniform Z Normal E Exponential L Logistic Default parameter values 0 1 TOKEN amp num TYPE INTEGER PROMPT How many random observations should be generated NEW REPEAT amp num LET amp dist amp rndnum amp dist amp RN DENSITY amp dist amp rndnum FILL 5 The amp dist token yields a dialog prompting for a single letter We use the IMMEDIATE option to prevent the prompt for the number of observations from appearing first The LET statement combines three tokens to yield one transformation statement A closer examination of this statement reveals some of the subtleties of token processing 178 Chapter 5 m First we need a replacement value for amp dist Due to the IMMEDIATE option this token already has a replacement value U Z E or L so processing continues Suppose the entered value equals U m Next we encounter the amp rndn
266. nd s intelligence BFINTGCE socioeconomic status BFSOCIEC parental aspiration BFPARASP occupational aspiration BFOCCASP and ambition BFAMBITN EX3 Mels and Koorts 1989 These data are taken from a job satisfaction survey of 213 nurses There are 10 manifest variables that serve as indicators of four latent variables job security JOBSEC attitude toward training TRAING opportunities for promotion PROMOT and relations with superiors RELSUP EX4A and EX4Be Lawley and Maxwell 1971 These data comprise a correlation matrix of nine ability tests administered to 72 children EXER The data consist of people who were randomly assigned to two different diets DIET low fat and not low fat and three different types of exercise EXERTYPE at rest walking leisurely and running A baseline pulse measurement PULSE was obtained at time 0 for every individual in the study However subsequent pulse measurements were taken at less regular 330 Chapter 9 time intervals The second pulse measurements were taken at approximately 2 minutes time 120 seconds the third pulse measurement was obtained at approximately 5 minutes time 300 seconds and the fourth and final pulse measurement was obtained at approximately 10 minutes time 600 seconds EXPORTS Hand Daly Lunn McConway and Ostrowski 1996 This data set consists of the value in millions of of British exports EXPORTS during the years 1820 to 185
267. nformation Ed Prompting text appears here i Cancel 161 Command Language Common information to include in the prompting text includes m the result of running the template file m changes to the data file if any m state of SYSTAT when template processing completes When command processing begins SYSTAT immediately displays the prompting text for a message token in a dialog Based on this text the user can elect to continue or cancel processing Pressing Cancel halts processing with no other commands in the template being executed If you exclude amp msg in the above command you will get a smaller message pop up A Prompting text appears here Filename Tokens Filename tokens represent any file that SYSTAT can open or save including data files command files and output files To substitute a filename for a token specify one of the following TOKEN file TYPE OPEN TOKEN file TYPE SAVE When SY STAT encounters the token amp file in the command file a dialog prompting the user for a filename appears SYSTAT substitutes the name of and path to the selected file for the corresponding token The OPEN type should be used when opening data files or for submitting command files For example TOKEN amp datafile TYPE OPEN TOKEN amp cmdfile TYPE OPEN USE amp datafile SUBMIT cmdfile 162 Chapter 5 Open Specify a file for the token Gdatafile Lack in 20 e m IN My A ecent Docu
268. ng any other command after a GSAVE command resets the internal index for the next GSAVE to the most recent graph To save all graphs in the Output Editor use GSAVE ROOT ALL FILETYPE When naming the resulting files the software appends consecutive integers beginning with 1 to ROOT To Export Results to Other Applications You can open your saved output and charts in word processing and other applications In SYSTAT save the file in a format that the other application can handle then open or import the file in that application SYSTAT offers a number of graph formats that are compatible with most Windows applications For example you can save a SYSTAT graph as a Windows Metafile WMF and then insert or import the metafile into most Windows word processing applications See the target application s documentation for specific information 198 Chapter 6 To Export Results Using the Clipboard Often the easiest way to transfer results to other applications is to copy and paste using the Windows clipboard This works for charts as well as text although results vary depending on the target application In SYSTAT select the output or chart From the menus choose Edit Copy m In the other application position the cursor where you want the output to appear From the menus choose Edit Paste Tips m If you have problems with Paste try using Paste Special on the Edit menu in the target application With Pas
269. ng homogeneity of variance and covariance across settings Potential analyses include ANOVA MANOVA regression and factor analysis MANOVA The input is USE DAYCREDM MANOVA PLENGTH LONG CATEGORY SETTING DEPEND DINNER STRANGER PROBLEM ESTIMATE 292 Chapter 8 The output is Effects coding used for categorical variables in model The categorical values encountered during processing are Variables Levels ano a S aa ea a ty set A a a aa ee Ho gt gt SETTING 3 levels 1 000 2 000 3 000 N of Cases Processed 48 Dependent Variable Means DINNER STRANGER PROBLEM 1288 188 714 250 54 083 Estimates of Effects B X X 1X Y Factor Level DINNER STRANGER PROBLEM A SR H gt gt CONSTANT 1308 795 690 589 5T T33 SETTING 1 166 479 62 116 2 207 SETTING 2 109 905 126 189 12 533 Standardized Estimates of Effects Factor Level DINNER STRANGER PROBLEM A is 4 CONSTANT 0 000 0 000 0 000 SETTING 1 0 278 0 176 0 069 SETTING 2 0 156 0 304 0 331 Total Sum of Product Matrix DINNER STRANGER PROBLEM A A ey ee ey DINNER 13624387 313 STRANGER 2382747 750 4713117 000 PROBLEM 241634 250 218044 000 39267 667 Residual Sum of Product Matrix E E Y Y Y XB i DINNER STRANGER PROBLEM E A a a ee 4 ee DINNER 1 12936578 626 STRANGER 2099145 095 3833722 926 PROBLEM 230259
270. ngth of life of the light bulbs so that standard deviation is less than 150 hours The data consists of LIFETIME of 20 bulbs BUSES Davis 1977 These data count the number of buses failing COUNT after driving 1 of 10 distances DISTANCE CANCER Morrison 1990 Bishop et al 1975 These studies examined breast cancer patients in three diagnostic centers CENTER three age groups AGE whether they survived after three years post diagnosis SURVIVE and the inflammation type minimum maximum and appearance of the tumor TUMOR malignant benign The variable NUMBER contains the number of women in each cell CANCERDM Cameron and Pauling 1978 The data set contains information from a study of the effects of supplemental vitamin C as part of routine cancer treatment for 100 patients and 1000 controls 10 controls for each patient CASE Case ID ORGAN Organ affected by cancer SEXS Sex of patient 323 AGE SURVATD CNTLATD SURVUNTR CNTLUNTR LOGSURVA LOGCNTLA LOGSURVU LOGCNTLU Data Files Age of patient Survival of patient measured from first hospital attendance Survival of control group from first hospital attendance Survival of patient from time cancer deemed untreatable Survival of control from time cancer deemed untreatable Logarithm of SURVATD Logarithm of CNTLATD Logarithm of SURVUNTR Logarithm of CVTLUNTR CARDOG Wilkinson 1975 This data set contains the INDSCAL configurations of the scalings
271. ning analytical procedures and graphs 125 126 Chapter 5 Commandspace Some of the functionality provided by SYSTAT s command language may not be available in the dialog box interface Moreover using the command language enables you to save sets of commands you use on a routine basis Commands are run in the Commandspace of the SYSTAT window The Commandspace has three tabs each of which allows you to access a different functionality of the command language Lo Untitled syc m Interactive tab Selecting the Interactive tab enables you to enter the commands in the interactive mode Type commands at the command prompt gt and issue them by pressing the Enter key You can save the contents of the tab SY STAT excludes the prompt and then use the file as a batch file Log tab Selecting the Log tab enables you to examine the read only log of the commands that you have run during your session You can save the command log or submit all or part of it Batch Untitled tab s Selecting a Untitled tab enables you to operate in batch mode You can open any number of existing command files and edit or submit any of these files You can also type an entire set of commands and submit the content of the tab or portions of it This tab is labeled Untitled until its content is saved The name that you specify while saving the content replaces the caption Untitled on the tab When the Commandspace is active you can cycl
272. ning and saving to occur directly within this folder Set custom directories As an alternative to specifying a project directory you can specify individual folders based on file type or file operation Graph Gallery Specify the folder containing the command files and graphics used to generate the Graph Gallery m Open data Sets the folder used for opening all SYSTAT data files SYZ and SYS When opening data files using the menus the Open dialog initially defaults to this folder This is set to the SYSTAT Data folder at the time of installation m Save data Defines the folder used for saving all SYSTAT data files SY Z When saving data files using the menus the Save As dialog initially defaults to this folder Ifa USE command is issued without a path SYSTAT also looks for the file in this folder This is set to the SYSTAT Data folder at the time of installation Work data Sets the folder used for saving all temporary data files SY Z Ifa USE command is issued without a path SYSTAT also looks for the file in this folder This is set to the Windows temporary folder at the time of installation Import data Identifies the folder used for all data file importing Export data Identifies the folder used for all data file exporting Command files Sets the folder used for opening and saving of SYSTAT command files When opening or saving command files using the menus the dialogs initially default to this folder This is set to
273. nover Inman 18 Chapter 1 9 Polynomial Regression SYSTAT offers polynomial regression on a single independent variable up to order 8 m In natural form or in orthogonal form Goodness of fit statistics R2 and adj R2 and ANOVA with p values for all models starting from the order specified by the user down to linear order 1 m Confidence and prediction interval plots along with estimates and a plot of residuals vs predicted values as quick graphs Modified Features 10 Analysis of Variance The Analysis of Variance feature now provides m Levene s test based on median for testing homogeneity of variances m A SUBCAT command that categorizes the desired factors just for the purpose of the analysis 11 Crosstabulation As part of 1ts Crosstabulation feature SYSTAT now offers m Relative Risk In a 2 x 2 table the relative risk is the ratio of the proportions of cases having a positive outcome in the two groups defined by row or column Relative Risk is a common measure of association for dichotomous variables m Mode SYSTAT gives an option to list only the first N categories in a one way table frequency distribution This is done by adding a MODE N option to the PLENGTH command within XTAB m Saved results with Mall requested columns in Multiway Standardize Evalue labels of the input variables for the corresponding columns of the saved results file Output categorized appropriately based on the type of table
274. ns Histograms Probability Plots Kriging Smoother Contour Plot of Kriging Smoother Graph Demonstration a y Data E I Graphics HW Statistics Command Templates Quality Analysis El vy Monte Carlo E I Exact Tests FPLOT z SIN x 2 y 2 SQR x 2 y 2 COLOR XMIN 2 XMAX 2 YMIN 2 YMAX 2 TITLE evieleyy Hat Function A For Help press F1 a 26 Chapter 2 You can perform many of the Graph editor related operations using the Graph Editing toolbar that is embedded in the Graph editor Use that and the menus to m Insert annotations and other text Change font color fill surface and line attributes Rescale axes Modify plot symbols Customize labels Edit legends Identify individual points in scatterplots Select a subset of cases using the Rectangular or Lasso tool Zoom and rotate graphs Change many other properties of a graph like changing its type drawing various smoothers specifying gradients for surfaces connecting and partitioning plot points slicing pie charts and setting attributes for each individual axis line You can view any number of graphs using the context menu of the Output Organizer See SYSTAT Graphics for more information about the Graph editor By default the tabs of the Viewspace are arranged in the following order Startpage Output editor Graph editor Active Data File Inactive Data Files When a new tab is opened it is in
275. nt variable and HORIZON as the factor Click on Options tab Check Shapiro Wilk option Click OK 112 Chapter 4 pa Analyze Analysis of Variance Estimate Model Model Available yariable s Dependent s CAMPLE UAANLOG LATITUDE aa LONGTUDE Resampling HORIZON HORIZON Factor z URANIUM HORIZON Coding ARSENIC 2 Effect BORON BARIUM Pacu MOLYBDEN Missing values SELENIUM Covarate s VANADILIM SULFATE TOT _ALE BICARBON Graph of Mean Uranium Levels Along with numeric output SYSTAT produces a Quick Graph a line connected plot of mean uranium levels and confidence intervals for the different producing horizons 113 Data Analysis Quick Tour Least Squares Means 4 3 O O JJ Z2 _ lt L X 1 0 1 2 3 4 5 HORIZON Most of SYSTAT s statistical procedures have associated Quick Graphs Quick Graphs speed up analysis by providing immediate visual feedback on results In this Quick Graph it is easily seen that the third group Quartermaster has a much higher level of uranium Output for ANOVA The numeric output of the ANOVA appears in the Output editor Analysis of Variance Source Type III SS df Mean Squares F ratio p value a a si a 4 gt gt HORIZON 14 978 4 3 744 3 252 0 014 Error 140 484 122 1 152 In the Analysis of Variance table the F test has a p value of 0 014 meaning that the
276. nterface The Workspace Viewspace and Commandspace can be resized 1f desired To do so m Drag the boundaries of the panes between Viewspace and Workspace Workspace and Commandspace and Viewspace and Commandspace in the desired direction You can also reposition the panes For this m Click the upper boundaries of the panes and drag the resulting outline to the new position As you drag the outline the border thins to indicate that the item will be docked to the main window at that location To prevent docking drag the item off the main window or hold down the Ctrl key as you drag Double clicking the upper boundary can undock docked items Undocking items enlarges the remaining panes but can result in a cluttered desktop You can collapse the Workspace and Commandspace so that they are only visible when you pause the mouse on the corresponding vertical bar at the edge To do this click the q at the top right corner of the pane The tabs of the Viewspace can be tiled so that you can view any two of the tabs simultaneously To do this m Click the Window menu or right click on the toolbar area and select Show Stacked or Show Side by Side All the panes in the Viewspace get laid out in a tiled fashion Double click one of the title bars to dock the panes to their default or previously docked positions Every toolbar can be repositioned by clicking and dragging the move handle ES Toolbars can also be dragged and docked to the boundary
277. ntext menus Show menus for Select context menu TE Y int Select a context menu click he Commands tab and drag items into the menu Font Popup menu fd Ue anos Reset The default menu structure of SYSTAT may be modified according to the user s preferences and needs as described earlier Use the Reset button to reset the menu structure to its default state Context menus are available for the Startpage Output Editor Data Editor columns rows and cells Graph Editor Output Organizer data view data graph other and main Examples folder and node Interactive Batch and Log tabs of the Commandspace and status bar To customize a context menu select it from the drop down list or right click in the associated pane so that it pops up Customize it as you would customize any other menu or toolbar If you drag and drop toolbar buttons the associated text is automatically displayed you cannot display only button images here Any changes are immediately applied Press the Reset button in the es menus group to reset the selected context menu to its default state Press the Close button at the top right corner or close the Customize dialog to close the popped ka menu 226 Chapter 7 Font Select the desired font and font size to be used for all the menu items Menu animation By default all SYSTAT menus pop up immediately on click You may choose to leave 1t that way or use one of the two available animation
278. nu item is virtually a button with text Simply right click on the desired button when the Customize dialog 1s open The following context menu pops up Reset to Default Copy Button Image Delete Button 4ppearance w Image Text Image and Text w Start Group Using this menu you can Reset to Default Resets the button appearance to its default state The default state for menu items without default images is the text displayed in the Commands list Copy Button Image Copies the button image to the clipboard You can then paste this in the Picture area while creating new images m Delete Deletes the button Alternatively you can simply drag a button out of the toolbar area to delete it Note that if you delete default buttons you can only retrieve them by pressing the Reset or Reset All buttons in the Toolbar and Menu tabs of the Customize dialog Button Appearance Pops up the Button Appearance dialog Use it as explained above to customize the selected button Image Text or Image and Text Sets the button appearance to show the specified image alone text alone or both image and text 217 Toolbars Customization of the SYSTAT Environment Start Group Inserts a separator before the selected button This is equivalent to dragging the button slightly to the right SYSTAT offers over 250 buttons categorized into 32 default toolbars to provide immediate access to most tasks Since showing all of these bu
279. nvironmental chambers Three different temperatures 65F 70F and 75F were assigned to three randomly selected chambers Two randomly selected men and two randomly selected women were assigned to each chamber The comfort of each person was measured after three hours in a scale of 1 to 15 where 1 cold 8 comfortable and 15 hot The variables are TEMP GENDER PERSON CHAMBER COMFORT COMPUTER Montgomery 2005 The following data represent the results of inspecting all units of a personal computer produced for 10 consecutive days DA Y UNITS are the number of computers inspected each day and NONCON is the number of nonconforming units found CONDENSE Messina 1987 The data file contains nonconformance data defects for 15 lots of condensers LOT is lot number TYPE is type of defect and TALLY is the frequency of a particular defect in a particular lot One thousand condensers were inspected in each lot CORK Rao 2002 Observations are obtained on 28 trees for thickness of cork borings in the NORTH N EAST E SOUTH S and WEST W directions The problem is to examine whether the bark deposit is same in all the directions We may consider the three characters contrast Ul N S E W U2 N S U3 E W CORN The data set gives the amount of inorganic phosphorous X1 organic phosphorous X2 present in the soil and the plant available phosphorous Y of corn grown in the soil COVAR s Winer 1971 Winer uses
280. nvolve yields of a chemical reaction YIELD under various combinations of four binary factors 4 B C and D Two reactions are observed under each combination of experimental factors so the number of cases per cell is two REGORTHO The data set consists of 25 random observations on X Y with X2 X Ke X4 X and X5 X Where X follows normal distribution with mean 5 and standard deviation 1 Y given X follows normal distribution with mean 1 X X and standard deviation 1 The data set is generated by using SYSTAT The variables in this data set are X Y X2 X3 X4 X5 REPEAT1 Winer 1971 These data contain two grouping factors ANXIETY and T ENSION and one trial factor TRIAL 1 to TRIAL 4 REPEAT 2 Winer 1971 This data set has one grouping factor NOISE and two trial factors period and dial The trial factors must be entered as dependent variables in a MODEL statement so the variables are named P D PID2 P3D3 For example P D2 means a score in the period1 dial2 cell RIESBY Reisby et al 1977 studied the relationship between desipramine and imipramine levels in plasma in 66 depressed patients classified as either endogenous or nonendogenous After receiving a placebo for one week the researchers administered a dose of imipramine each day for four weeks recording the imipramine and desipramine levels at the end of each week At the beginning of the placebo week and at the end of each week including the placebo
281. ny of these displays the corresponding menu items in the Commands list Now all you need to do is to drag and drop items from this list to the desired position If you are not sure what a particular item here corresponds to select it to view a description of the item in the Description area Items that have images preceding their names will be displayed as buttons with the images on them whereas the Button Appearance dialog pops up when you drop items that do not 214 Chapter 7 Button Appearance O Image only O Image and text 6 e i faevoo Description B x Start or stop recording the command script Button text Start Stop Recording Three choices are available Image only The image that you select from the Image area will be displayed Text only The button will only have a caption Use the default button text that is displayed in the Button text area or enter your own text Image and text Both the image that you select and the desired text will appear For the first and third options you can also create your own image or edit an existing one in the Image area Just press New or select an existing image and press Edit to invoke the Edit Button Image dialog box 215 Customization of the SYSTAT Environment Edit Button Image Picture Colors eo EEE Tools Tas Preview EN 010 a BAX Use any of the colors shown in the palette and any of the tools in the Tools area to crea
282. odel to help determine the amounts of three compounds present in samples from the Baltic Sea Lignin Sulfonate pulp industry pollution LS Humic Acids natural forest products HA and optical whitener from detergent D7 The data set consists of 16 samples of known concentrations of LS HA and DT with spectra based on 27 frequencies or equivalently wavelengths SPECTROMETERS Two mass spectrometers SPECTROMTR were compared for accuracy in measuring the ratio of LAN to N Three plots of land PLOT treated with SN were used and from every plot two soil samples SAMPLE were taken Each sample had two observations The response variable RATIO is the ratio of HN to PN multiplied by 1000 RATIO Ratio of two soil measurements SPECTROMTR ID of a spectrometer A B PLOT Plot number SAMPLE Sample number SPIRAL These data consist of a spiral in three dimensions with the variables X Y Z R and THETA SPLINE Brodlie 1980 These data are X and Y coordinates taken from a figure in Brodlie s discussion of cubic spline interpolation SPNDMONY Chatterjee Hadi and Price 2000 In this data set SPENDING is consumer expenditures and MONEY is money stock in billions of dollars in each quarter of the years 1952 1956 DATE STRESS Brown 2006 adapted from Folkman amp Lazarus 1970 Tobin Holroyd Reynolds amp Wigal 1989 The data set is a covariance matrix of 12 manifest variables which represents four distinctive
283. of input data to analyze matched sample case control studies with one case and any number of controls per set m Discrete choice model provides two data layout inputs Choice set and BY choice to model an individual s choices in response to the characteristics of the choices mIn the raw data layout choice set names for groups of variables can be defined and variables can be created edited or deleted mIn the by choice framework the choices sets already defined can be used in the data for the analyses 17 Mixed Models The Mixed Models feature performs significantly faster than in prior versions Chapter Introducing SYSTAT Keith Kroeger revised by Rajashree Kamath SYSTAT provides a powerful statistical and graphical analysis system in a graphical environment using descriptive menus and simple dialog boxes Most tasks can be accomplished simply by pointing and clicking the mouse This chapter provides an overview of the windows menus dialog boxes and Online Help available in SYSTAT For information on using SYSTAT s command language see Chapter 5 User Interface The SYSTAT window is made up of three panes which we term as m Workspace m Veiwspace Commandspace Each pane consists of various tabs or sets of tabs and allows you to accomplish specific tasks One pane and one tab within it will always be in focus At any given moment certain menu selections and their corresponding keyboard shortcuts like Ctr
284. of these files You could also type in an entire set of commands and then save or submit it The name that you specify while saving any content that you may have typed here replaces the caption Untitled on the tab Log Selecting the Log tab enables you to examine the read only log of the commands that you have run during your session You can save the command log or even submit one or more of the generated commands By default the tabs of the Commandspace are arranged in the following order m Interactive E Log Command Files When a new tab is opened it is inserted at the beginning of its group Batch You can click the arrow in the bottom right corner of the Commandspace and check Active Tab at the Beginning if you want a new tab to appear as the first tab of the Commandspace You can bring a tab into focus by clicking the arrow and checking the name of the desired tab If you have opened more than 9 command files the tab becomes the first tab in the Commandspace or in its group depending on whether Active Tab at the Beginning is checked or not This is especially useful when you have a lot of tabs open in the Commandspace You can close the tab in focus by right clicking and selecting Close or pressing the Close button in the bottom right corner of the Commandspace You can close all open 29 Introducing SYSTAT command files by right clicking in any tab of the Commandspace and selecting Close All Reorganizing the User I
285. om a study showing that inexperienced computer users prefer dialog menu interfaces while experienced users prefer command based interfaces SESSION is the session number and TASKS is the number of command based as opposed to dialog based tasks initiated by the user during that session 337 Data Files LEISURE Clausen 1998 These data show a cross classification between different leisure activities and different occupational status The following is a list of the different activities and occupational status The SYSTAT names are within parentheses Activities Occupational Status Sports Events Sports Manual MANUAL Cinema Cinema Low Non Manual LOWNM Dance Disco Dance High Non Manual HIGHNM Cafe Restaurant Cafe Farmer FRAMER Theatre Theatre Student STUDENT Art Exhibition Ar Retired RETIRED Library Library Church Service Church Classical Music Classical Pop Pop LIFE The data are lifetimes LIFE of 20 units of a certain equipment LONGLEY Longley 1967 These data are economic data selected by Longley to illustrate computational shortcomings of statistical software The variables are DEFLATOR GNP UNEMPLOY ARMFORCE POPULATN TIME and TOTAL LUNGDIS gt Hand Daly Lunn McConway and Ostrowski 1996 This data set consists of monthly MONTHS deaths DEATHS from lung diseases in the UK during the years YEAR 1974 to 1979 MACHINE These data are in the file MACHINE and
286. ome of the variance using the covariate APTITUDE in an ANCOVA model there is a significant difference between instruction groups Potential analyses include ANOVA ANCOVA and regression Analysis of Covariance The input is USE INSTRDM GLM CATEGORY INSTRUCT EFFECT MODEL ACHIEVE CONSTANT INSTRUCT APTITUDE ESTIMATE 305 The output is Effects coding used for categorical variables in model The categorical values encountered during processing are Variables Levels a iais a peaa iaa fa a aaan as aSa iaa Aa Daa faata aa a a aG ia a Ho INSTRUCT 2 levels 1 000 2 000 Dependent Variable ACHIEVE N i 20 Multiple R 0 760 Squared Multiple R 0 578 Estimates of Effects B X X 1X Y Factor Level ACHIEVE det faa a a 4 CONSTANT 9 646 INSTRUCT 1 5 755 APTITUDE 0 502 Source i Type III SS df Mean Squares F ratio p value A a j a a y a 4 gt gt INSTRUCT 641 424 1 641 424 10 915 0 004 APTITUDE 961 017 1 961 017 16 354 0 001 Error 998 983 17 58 764 Factor Level LS Mean Standard Error N a a a a e n a H gt gt INSTRUCT 1 28 745 2 444 10 000 INSTRUCT 2 40 255 2 444 10 000 Least Squares Means 49 39 Lu gt Lu I O lt 29 19 1 2 INSTRUCT Applications 306 Chapter 8 Durbin Watson D Statistic wel Oy First Order Autocorrelation 0 171
287. on Repeat the steps for the previous plot but select VITAMINA CALCIUM IRON and COST as the row variables VITAMINA VITAMINA O a lt x O VITAMINA CALCIUM IRON COST COST is the Y variable for each plot on the bottom row There is no strong relationship between cost and nutritive value as measured by VITAMINA CALCIUM and IRON but there is a small cluster of low cost dinners with high calcium content Later we will find that these are pasta dinners 3 D Displays In this section we use 3 D displays for another look at calories protein and fat In the display on the left we label each dinner with its brand code in the display on the right we use the cost of the dinner to determine the size of the plot symbol ce SYSTAT Basics To produce 3 D displays m From the menus choose Graph Scatterplot m Inthe Scatterplot dialog box select FAT as the X variable PROTEIN as the Y variable and CALORIES as the Z variable Select Display grid lines in the X Axis Y Axis and Z Axis tabs Click the Options tab and select Vertical spikes to Y from the Connectors partitions group To produce the plot on the left click the Symbol and Label tab click Display case labels in the Case labels group and select BRANDY to label each plot point with the brand of the dinner To produce the plot on the right click the Symbol and Label tab click Select variable in the Symbol size group and select COST a
288. on Gallery The available applications are listed with icons and a brief description Clicking on any icon will open a page containing the detailed description and buttons for the main Application Gallery page Analyses page and Sources page Chapter SYSTAT Basics This chapter provides simple step by step instructions for performing basic analysis tasks in SYSTAT including Starting SYSTAT Entering data in the Data Editor Opening and saving data files Using menus and dialog boxes to create charts and run statistical analyses 45 46 Chapter 3 Starting SYSTAT To start SYSTAT for Windows XP 2000 ME and NT4 m Choose Start Programs SYSTAT 13 SYSTAT 13 SYSTAT Startpage Baa 4 SYSTAT Output Aus SYSTAT 13 To edit the graph format scales thickness etc from the menus choose Edit gt Options Or press F6 Themes Current Theme Default gt Next Tip Classic Default Introductory_Statistics Moarlent Nanosech Scratchpad Manuals GettingStarted pdf For Help press F1 47 Entering Data SYSTAT Basics This section discusses how to enter data If you prefer to start with data stored in a text file see Reading an ASCII Text File on p 51 In the frozen food section of the grocery store we recorded this information about seven dinners Brand Calories Lean Cuisine 240 Weight Watchers 220 Healthy Choice 2
289. on are best viewed in 16 bit or 32 bit true color on a high resolution monitor From the preceding statistical analysis we can conclude that there are differences in the uranium level between the producing horizons However we also have the latitude and longitude for each sample so we can perform a geographic analysis to better pinpoint the variations in uranium To accomplish this we will apply a smoothing technique called kriging pronounced kree ging to fit a 3 D scatterplot of uranium by latitude and longitude Kriging is a smoothing technique often used in geostatistics It uses local information around points to extrapolate complex and irregular geographic patterns 117 Data Analysis Quick Tour Kriging Smoother From the menus submit the file GDWTR2DM E From the menus choose File Submit File E Select the file GDWTR2DM from the Miscellaneous subfolder of the command directory and click Open The following graph is displayed in the Output editor Actual Uranium and Kriging Smoother by Geography Uranium as Q Vv S o S WY Y e This plot shows the level of uranium against latitude and longitude the data points and the kriging smoother the surface The plot provides us with a topography of the uranium level and we can see immediately that there is a pronounced peak near the center of the sampling area Rotation If you look at the Dynamic Explorer the rota
290. on drawing in Mathematical Methods in Computer Graphics and Design pp 1 37 Academic Press New York and London Brownlee K A 1960 Statistical theory and methodology in science and enginnering New York John Wiley amp Sons Cameron E and Pauling L 1978 Supplemental ascorbate in the supportive treatment of cancer Reevaluation of prolongation of survival times in terminal human cancer Proceedings of the National Academy of Sciences USA 75 4538 4542 Carey J R Liedo P Orozco D and Vaupel J W 1992 Slowing of Mortality Rates at Older Ages in Large Medfly Cohorts Science 258 457 461 Caroll J B Davies P and Richmond B 1971 The word frequency book Boston Mass Houghton Mifflin Chambers J M Cleveland W S Kleiner B Tukey P A 1983 Graphical methods for data analysis Duxbery Press Boston Chatterjee S Hadi A S and Price B 2000 Regression analysis by example 3rd ed New York John Wiley amp Sons Clarke C P Y 1987 Approximate confidence limits for a parameter function in nonlinear regression Journal of the American Statistical Association 85 544 551 Clausen S E 1998 Applied correspondence analysis An introduction University Paper 361 Data Files Series on Quantitative Application in Social Science 7 121 Thousand Oaks CA Sage Cleveland W S 1993 Visualizing Data Summit NJ Hobart Press Cochran W G and Cox G 1
291. only be used after estimating a model successfully These contingent dialogs do appear in the Recent Dialogs list but are removed each time a data file is opened 230 Chapter 7 Although the goal of Recent Dialogs is to present the most recently used dialogs some main dialogs do not appear in the list The Variable Properties and Add Empty Rows dialog boxes for example do not receive list entries Furthermore wizards that result in a sequence of dialogs only receive an entry for the first dialog of the sequence Note Because most dialog boxes require variable specifications Dialog Recall is disabled if there is no open data file User Menus SYSTAT s menus offer a dialog interface to most of the underlying command language You can also create an additional menu with entries designed to process sets of commands that you frequently run To add a user menu item from the menus choose Utilities User Menu Add Delete Modify iid Utilities User Menu Add Delete Modify 2 x Add name for menu tem Aun command fram File Menu Item Menul User input 231 Customization of the SYSTAT Environment Menu item Displays all the menu item names that are currently defined Use the E and buttons to insert new items and delete unwanted items respectively The names in this list will be displayed under the Menu List sub menu of User Menu You can define any number of menu items here but the Menu List will disp
292. ons To enter a module type its name after the prompt and press the Enter key For example type XTAB m Next identify which data to use For example type USE ourworld and press the Enter key m Now type a command line TABULATE leader group MEAN pop 1983 129 Command Language gt XTAB gt USE Ourworld syz gt TABULATE LEADER GROUP MEAN POP 1983 m Press the Enter key to obtain output To create graphs type the desired graph command followed by the variables to use Specify optional settings to customize the resulting display Valid graph commands include BAR CONE CYLINDER DENSITY DOUGHNUT DOT DRAW FOURIER FPLOT ICON LINE MAP PARALLEL PIE PLOT PPLOT PROFILE PYRAMID OPLOT SPLOM WRITE Note SYSTAT can use one of two modes for drawing graphs One is the DirectX mode and the other is the classic mode The options CONE CYLINDER DOUGHNUT are available only in the DirectX mode By default SYSTAT uses the classic mode You can run RENDER DIRECTX to switch to the DirectX mode Refer the Language Reference volume for details regarding general and data related commands Command Syntax Most SYSTAT commands have three parts a command an argument s and options command argument options Each module name or command must start on a new line A command must be separated from its argument by a space the equal sign is not allowed except in a few specific cases and options must be separated
293. ons as long as this dialog is open Commands Customization Any menu menu item within it or toolbar button can be moved from its default position to any other position either in the menu bar any menu or in any toolbar Keep the Customize dialog open or in the case of toolbar buttons and terminal menu items hold down the Alt key and drag and drop the item there will be a border around the item while it is being dragged to the desired position To copy an item instead of moving it hold down the Ctrl key as well To completely remove an item just drag it out of the menu and toolbar area Dragging an item slightly to the right creates a separator before it while dragging it slightly to the left removes the separator if any All changes can be reset using the Reset and Reset All buttons in the Toolbar and Menu 213 Customization of the SYSTAT Environment tabs of the Customize dialog or the Default Settings link in the SYSTAT program group of the Windows Start Menu You can also create new menus menu items or toolbar buttons by dragging and dropping items from the list of items in the Commands tab of Customize into the desired menu or toolbar position Customize Commands Toolbars Keyboard Menu Categories Commands Data Command Output Data Command al Database Capture The Categories list contains the names of all the menus and menu items Clicking a
294. ons it was desirable to partial those effects from the association Using SYSTAT the variables SEX times AGE and their squares were created The variables are D ANTISO_S MATER S CONVEN S ANTISO_O MATER O CONVEN O AGE SEX AGESO SEXAGE SEXAGESO ADMIT Graduate Record Examination Verbal GREV and Quantitative GREQ scores with a binary indicator of whether or not a student was awarded a Ph D PHD in a graduate psychology department The variables are YEAR GPA GREV GREQ GRE PHD GROUP N PHD AEROSOL Beckman Nachtsheim and Cook 1987 This is a study of high efficiency particulate air HEPA cartridges For this two aerosol types AEROSOL were used to test the three HEPA respirator filters FILTER from each of two different manufacturers MANUFACTURER 319 Data Files AFIFI Afifi and Azen 1974 The dependent variable SYSINCR is the increase in systolic blood pressure after administering one of four different drugs DRUG to patients with one of three different diseases DISEASE Patients were assigned randomly to one of the four possible drugs AGE1e The data set consists of two variables AGES and SEXY AGESEX U S Census 1980 These data show the distribution of MALES and FEMALES within age groups The variable AGE labels each age group by the upper age limit of its members AGESTAT The data set is randomly generated data consisting of two variables AGE and SEX AGRI and AGR2 The
295. ontaining a list of all the recently opened data command and output files you can reopen these files just by double clicking on their names Themes contain a list of menu themes double click any one to apply it to the SYSTAT window 23 Introducing SYSTAT Manuals containing a list of the user manual documents you can open the desired volume by double clicking on its name Tips providing useful tips about SYSTAT s features and how to achieve any given task clicking Next Tip will allow you to scroll through any number of tips Scratchpad for writing notes while you are working with SYSTAT Anything that you enter here remains across sessions You can click on the bar at the top of the Startpage to know about the new features in the current version of SYSTAT You can close the Startpage if you do not need it for the remainder of a session or even prevent it from appearing when SYSTAT restarts Output editor Graphs and statistical results appear in the Output editor Collapsible links are created for each analysis or graph that you request You can thus hide output that you do not need to see all the time Simply click on the link once to collapse the corresponding output click again to expand it You can perform some of the Output editor related operations using the Format Bar that is embedded in the Output editor For more information about the Output editor see Chapter 6 EJES SYSTAT Untitled
296. ontains two indicator variables X11 and X21 representing the cases obtained from the first and second model respectively X12 and X22 represent the market values of a certain product of two different companies with capital stocks X13 and X23 respectively The dependent variable Y represents the investment figures for the two companies The data set 1s fictitious JUICE Montgomery 2005 The number of defective orange juice cans DEFECTS found in each of 24 samples SAMPLE of 50 juice cans Data are collected on each of three shifts TIME with eight samples taken for each shift SHIFT SIZE is also a variable 336 Chapter 9 JUICE1 Montgomery 2005 The following fictitious variable has been added to JUICE DEFECTS The number of defective orange juice cans found in each of 24 samples SAMPLE of 50 juice cans KENTON Neter Kutner Nachtsheim and Wasserman 1996 These data comprise of unit sales of a product SALES under different types of package designs PACKAGE Each case represents a different store KOOIJMAN Kooijman 1979 reprinted in Upton and Fingleton 1990 The data consist of the locations of beadlet anemones Actinia equina on the surface of a boulder at Quiberon Island off the Brittany coast in May 1976 KUEHL Kuehl 2000 The original data source is Dr S Denise Department of Animal Sciences University of Arizona A genetic study with beef animals consisted of several sires each mated to a separa
297. orm Let Available variable Function type a CALORIES FAT PROTEIN VITAMINA CALCIO ri Add to Variable Expression Let CALCIUM SQRICALCIUM Click OK Now request the analysis of variance repeating the steps in the last example except that here we use CALCIUM as dependant variable and both DIET and FOOD as the factor variables Functions 93 Data for the following results were selected according to SELECT FOODS lt gt beef Effects coding used for categorical variables in model The categorical values encountered during processing are Variables DIETS 2 levels i Tn st eee vcd a a a 4 FOODS 2 levels Dependent Variable N Multiple R Squared Multiple R Estimates of Effects B X X 1X Y CONSTANT DIETS FOODS DIETS FOODS l i l l l l Levels no chicken 0 8 000 yes pasta CALCIUM Z2 04 47 no chicken no chicken Analysis of Variance Source FOODS DIETS FOODS Error l i l l Type LEL ss 7 908 22s La Least Squares Means Factor DIETS no DIETS yes Level LS Mean Least Squares Means Factor Least Square Factor DIET FOODS DIET FOODS DIETS FOODS DIETS FOODS The significant DIET by FOODS interaction suggests exercising caution when s i l l l l Level LS Mean WS SAR H gt
298. ors REGRESS MODEL amp resp ESTIMATE prompt Select the dependent varaible prompt prompt prompt prompt CONSTANT amp v1 amp v2 select select select select CONSTANT amp v1 amp v2 V3 amp v4 a a a a variable variable variable variable Unfortunately although these templates apply linear regression to user specified variables these templates only apply to models involving two and four predictors respectively To create templates allowing for a varying number of variables use the MULTIVAR NMULTIVAR and CMULTIVAR token types Here we create a linear regression template allowing any number of predictors and generate hypothesis tests to determine whether coefficients equal zero TOKEN resp TYPE NVARIABLE PROMPT Select the response variable TOKEN amp predictors TYPE NMULTIVAR SEPARATOR PROMPT Select the predictor variables for the multiple regression model 180 Chapter 5 TOKEN amp hypeff TYPE NMULTIVAR SEPARATOR PROMPT Select predictors whose coefficients you wish to test for differences from 0 REGRESS MODEL amp resp CONSTANT amp predictors ESTIMATE HYPOTHESIS ALL TEST HYPOTHESIS EFFECT hypeff TEST TOKEN OFF The amp predictors token represents all predictors in the model The user selects the variables to include and SYSTAT generates the token value by inserting a
299. p To optimize printed output you may need to adjust various page settings The available options vary for different printers To open the Page Setup dialog box choose Page Setup from the File menu Print Setup esmardcO2 HP Laserjer 1320 MGR 4th fl w hp LaserJet 1320 PCL 6 IP_10 0 0 222 Orientation a ME i rtra Sauce Capa If more than one printer is installed on your system or network you can choose which one to print to You can also specify paper size and orientation portrait tall or landscape wide 201 Working with Output Printing Graphs Using Commands You can print individual graphs by entering the following GPRINT LANDSCAPE or PORTRAIT SYSTAT automatically sends the most recently created graph to the default printer In the absence of an orientation specification the software uses the setting for the current printer Issuing multiple consecutive GPRINT commands results in multiple graphs being printed SYSTAT prints the most recent graph first the graph created before the most recent graph second and so on However issuing any other command after a GPRINT command resets the internal index for the next GPRINT to the most recent graph Chapter 7 Customization of the SYSTAT Environment Revised by Rajashree Kamath By default the user interface contains from top to bottom Toolbars m Workspace and Viewspace Commandspace Ed Status Bar However as you work with SYSTAT you
300. p Stops loading a page Refresh Refreshes the currently loaded page Home Loads the SYSTAT Help Copyright page Print Prints the current topic or all sub topics under the current heading when you click this with the Contents tab active When any other tab is active use this to print the current page Before printing the Print dialog pops up so that you can specify the desired print settings Options Enables you to do any of the above access the Windows Internet Options settings or specify whether you want search keywords to be highlighted in the listed pages or not Depending on the topic displayed the following buttons may appear in the current Help page How To Provides minimum specifications for performing the analysis Syntax Describes the associated SYSTAT command SYSTAT s command language offers some features not available in the dialog boxes 40 Chapter 2 m Examples Offers examples of analyses including SYSTAT command input and resulting output Copy and paste the example input to the Batch tab of the Commandspace to submit the example as 1s or modify the commands to your own analyses before submitting them Make sure the file paths match the file locations you have opted for More Lists analysis options and related tabs These topics are particularly useful for customizing your analyses m See Also Lists related procedures or graphs You can select cut copy paste and print the content of any Help p
301. p titlel the following dialog appears MM Information Enter the graph title O Continue Cancel Custom prompts can include carriage returns in the prompting text allowing you to define the text appearing on each line of a multi line prompt For example TOKEN amp varl TYPE VARIABLE PROMPT This is the first line this is the second and this is the third results in a three line prompt In the absence of carriage returns SYSTAT automatically wraps prompting text to fit the dialog Although the dialogs for string number and integer replacement have no practical limit on the number of lines that can be used as a prompt the dialogs for variable selection limit custom prompts to three lines of text 169 Command Language Choice Tokens In contrast to all other token types except message tokens choice tokens do not have a value Instead the choice token submits command files based on the choice given by the user To define a choice token specify TOKEN TYPE CHOICE choicel filenamel syc choice2 filename2 syc choiceN filenameN syc Select choice Select O choice O choice choice Choice tokens are executed immediately You may specify between 2 to 10 choices Dialog Sequences Processing of command files begins at the first line of the file and continues to the last line SYSTAT does not prompt for token replacement values until the token being defined 1s encountered in a command
302. p xyscale amp xyscale CSIZ charsize THICK amp linethickness SUBMIT amp cmdfile SCALE CSIZE THICK The final three commands return the global options to their default settings Example 8 Combining Analyses Two Way ANOVA Menus and dialogs offer a prescribed set of options resulting in a variety of statistics and graphs When performing a series of analyses or including graphs with statistical output using token substitution simplifies the process considerably For example multidimensional scaling requires a matrix input You could generate this matrix from a rectangular file using the CORR procedure before running MDS You could then save the final configuration for custom plotting Instead of running each procedure separately however we can automate the entire process using a template You can apply the template to any data to generate output customized to your needs In this example we focus on two way ANOVA Using four tokens we generate box plots displaying the distribution of the dependent variable for every level of each factor analysis of variance results post hoc tests for main and interaction effects 182 Chapter 5 an interaction plot displaying the dependent variable mean in each cross classification of the two factors a residual plot a stem and leaf plot of the residuals USE OURWORLD TOKEN ON TOKEN outfile TYPE SAVE PROMPT Save ANOVA Statistics TOKEN amp factorl TYPE Vvariabl
303. pe of angle bracket used The reading of the fuel gauge under the designed conditions DEVMER DEVEMER data file is derived from OURWORLD data file DIVORCE Wilkinson Blank and Gruber 1996 and originally from Long 1971 This data set includes grounds for divorce in the United States in 1971 DJONESs Brockwell and Davis 1991 The data set contains Dow Jones Index of stocks on the New York Stock Exchange at closing on 251 trading days ending 26 August 1994 The data set contains the following variables DJSTOCK Values of daily stocks of New York Stock Exchange DJPRC Percent relative price changes of the DJSTOCK series DOPTIMAL Myers and Montgomery 2002 The data set is from an experiment based on a D optimal design on adhesive bonding where the factors are amount of adhesive X7 and cure temperature X2 Here the response is the pull off force Y DOSE These data are from a toxicity study for a drug designed to combat tumors The data show the proportion of laboratory rats dying RESPONSE at each dose level DOSE of the drug LOGDOS dose in natural logarithm units ECLIPSE These data are from the National Aeronautics and Space Administration web site and represent the longitude and latitude for the paths of eight future solar eclipses Measurements occur attwo minute intervals The data are used courtesy of Fred Espenak NASA GSFC The variables are MAPNUM TIMES MAXLAT MAXLON MINLAT MINLON LABLA
304. ply right clicking on the variable name in the data file in the Comments box at the bottom you will see information on the variable This information on the variable is also seen as a tooltip by simply moving the mouse over the variable name For a data file you create you may construct this general file information by filling it in the File Comments dialog which can be opened by right clicking on the file name in the Data editor or on the top left cell Information on individual variables may be entered in the Comments box of the Variable Properties dialog The data file contains even more information which can be seen by clicking the Variable tab in the Data editor which opens the Variable editor This contains information on each variable as to its name label value labels type string or 317 318 Chapter 9 numeric categorical or not the number of characters number of decimals display type and comments It also contains information on which variables are involved in case selection has been chosen to be a frequency or a weight variable for BY groups analysis a category variable or an order variable The following data files are Read only ACCIDENT Jobson 1992 The data set relates to automobile accidents in Alberta Canada The variables are SEATBELT IMPACTS INJURY DRIVER FREQ ADAPTOR The adaptor body is one of the components of a machine Its outer diameter is denoted by DIA The
305. ponential Decline in Fruit Flies Over Time The input is USE FRTFLYDM NONLIN MODEL LIVING 1203646 exp A B DAY C DAY 2 DAY ESTIMATE ITER 50 253 The output is Iteration History No i Loss O 1 541E 013 1a on 0 Bae OSS 2 1 468E 013 3 1 416E 013 4 1 411E 013 5 1 411E 013 6 1 411E 013 7 1 410E 013 8 1 410E 013 9 1 410E 013 10 1 410E 013 Ti p LE 4108 OLS t2 1 41OBEFOLS 13 1 410E 013 A gt ELE IO Tr LITERO L2 16 4 213E 012 TE BA 18 1 621E 011 19 2 562E 010 20 2 282E 010 21 2 228E 010 22 2 164E 010 23 1 384E 010 24 1 309E 010 25 1 305E 010 26 1 305E 010 27 1 305E 010 28 1 305E 010 29 1 305E 010 30 1 305E 010 31 1 305E 010 Dependent Variable Sum of Squares and Source Regression Residual Total Mean corrected R squares Raw R square Mean Corrected R square 04 04 04 04 o0O0O0O0O0O0O00O0o00O0O0O0O0O000Oo0O0000O0o000o0Oo00o0Oo0O0Oo0OoO0Oo LIVING Mean Squares 363E 013 305E 010 364E 013 983E 013 1 Residual Total 1 Residual Corrected R square Observed vs Predicted Parameter Estimates Parameter Estimate ASE Parameter ASE A le a hc ae a fe A 0 013 0 001 14 165 B 0 002 0 000 21 259 C 0 000 0 000 4 773 o0O0O0O0O0Oo0O00O0o00O00O0O0O00Oo0O00o00O0000O0O0O0o0o0O0Oo0OoO0Oo Mean Squares 7 877E 012 76738341 153 0 999 07999 0 999 Wald 95 Confidence Lower Appli
306. press Quick Graphs ias 5 n Indicates whether to echo commands in the output or not E ue Controls the appearance of statistical results FPATH path PROJECT or Specifies a path prefix to append to filenames If path is filetype not specified all file locations are set to the program folder If no option is specified all directories are set to the specified path PROJECT will set path as the root directory under which sub folders Gallery Data Com mand and Output will be created For the filetype in the FPATH statement specify one of the following GALLERY USE SAVE WORK IMPORT EXPORT SUBMIT OSAVE OUTPUT GSAVE GET and PUT Applications SYSTAT offers applications in the following fields Anthropology Astronomy Biology Chemistry Engineering Environmental Sciences Genetics Manufacturing Medical Research Psychology Sociology Statistics Toxicology Chapter You can find these applications in the online Help Use the Contents tab of the Help system to access the Application Gallery In the gallery you will find sample analyses with their associated commands and menu selections All relevant data and command files are included 247 248 Chapter 8 Anthropology Egyptian Skulls Data EGYPTDM data consists of four measurements of male Egyptian skulls from five different time periods ranging from 4000 B C to 150 A D Variable Description MB BH BL NH Skull measurements YEAR Year of measurement
307. pter 3 Matrix of Bonferroni Probabilities CALORIES FAT PROTEIN cost A i Ho o gt CALORIES 0 000 EAT 0 000 0 000 PROTEIN 0 014 0 908 0 000 COST 1 000 1 000 eck SO 0 000 Scatter Plot Matrix PROTEIN FAT CALORIES COST CALORIES FAT PROTEIN COST In above output one Quick Graph is generated This is the Quick Graph that SY STAT automatically generates when you request correlations Quick Graphs are available for most statistical procedures If you want to turn off a Quick Graph use Options on the Edit menu The Quick Graph in this example is a scatterplot matrix SPLOM There is one bivariate scatterplot corresponding to each entry in the correlation matrix that follows Univariate histograms for each variable are displayed along the diagonal and 75 normal theory confidence ellipses are displayed within each plot The plot of FAT and CALORIES top left has the narrowest ellipse and thus the strongest correlation that is given that the configuration of the points is spread evenly is not nonlinear and has no anomalies In the Pearson correlation matrix displayed in above output the correlation between FAT and CALORIES is 0 758 The p value or Bonferroni adjusted probability associated with 0 758 is printed as 0 000 or less than 0 0005 As the scatterplot seemed to indicate the FAT and CALORIES Pearson correlation matrix is correlated a3 SYSTAT Basics PROTEIN also has
308. r Chemistry you may create one theme for each case and apply the appropriate theme as required You can save the changes you make to the default theme or any existing theme of SYSTAT in a theme file To do this from the menus choose Utilities Themes Save Current Theme In the dialog that pops up enter a suitable file name and press Save All menu items status bar content toolbar layout and location as well as those of the Workspace Viewspace and Commandspace will be saved in this file By default the file will be saved to the Themes folder of SYSTAT You may specify a different folder to save to the advantage of saving in the Themes folder is that the theme will be listed in the Themes section of the Startpage The name of the theme will be the same as the filename you simply have to double click the desired theme name to apply it In any case to apply any stored menu theme from the menus choose Utilities Themes Apply Theme Navigate to your themes folder select the desired file and press Open 233 Customization of the SYSTAT Environment New themes will be available on the SYSTAT server from time to time To download these from the menus choose Utilities Themes Download Themes Download Themes In the dialog box that opens check the themes that you want to install uncheck the ones that you do not need and press Download If you do not want to install themes at this time press Close To revert to th
309. r brand of instant coffee on sale in 15 different shops and amount in gm per pence in Milton Keynes on the same day in 1981 The variables are PRICE GM_ PER PENCE COLAS Schiffman Reynolds and Young 1981 These data consist of judgments by 10 subjects of the dissimilarity 0 100 between pairs of colas including DIETPEPS RC YUKON PEPPER SHASTA COKE DIETPEPR TAB PEPSI and DIETRITE COLOR These data provide the proportions of RED GREEN and BLUE that will produce the color specified in COLORS COLRPREF The data set contains color preferences RED ORANGE YELLOW GREEN BLUE among 15 people VAMES8S for five primary colors COMBAT Stouffer et al 1950 This data set contains reports of fear symptoms by selected U S soldiers after being withdrawn from World War II combat Nine symptoms are included for analysis and the number of soldiers in each profile of symptom is reported The variables are COUNT Number of soldiers in each profile of symptom POUNDING Violent pounding of the heart SINKING Sinking feeling in the stomach SHAKING Shaking or trembling all over NAUSEOUS Feeling sick to the stomach STIFF Cold sweat FAINT Feeling of weakness or feeling faint VOMIT Vomiting BOWELS Loss of bowel control URINE Loss of urinary control 325 Data Files COMFORT Milliken and Johnson 1992 In an experiment the effects of temperature on the comfort level of 18 men and 18 women was carried out using nine e
310. r toolbars can be displayed or closed using the Toolbars tab of the Customize dialog or the View gt Toolbars menu Positioning Toolbars Toolbars can be docked to pane borders or left floating in front of the user interface To move a toolbar click the handlebar at the left or top and drag the toolbar to the new location m Dragging a toolbar to the left or right side of a pane that is in the docked state attaches or docks the toolbar vertically to that side m Dragging a toolbar to the top or bottom of a pane that is in the docked state attaches or docks the toolbar horizontally m Dragging a toolbar anywhere other than window borders creates a detached floating toolbar Alternatively you can hold down the Ctrl key while dragging to prevent toolbar docking Clicking the in the upper right corner closes floating toolbars a Toolbar Customization The Toolbars tab of the Customize dialog enables you to close or display SYSTAT toolbars as well as create new toolbars 219 Customization of the SYSTAT Environment Customize Commands Toolbars Keyboard l Menu Show tooltips With shortcut keys Graphs Hypothesis Testing Linear Models Matrix lea ri The Toolbars list contains the names of the available toolbars prefixed by check boxes Notice that the Menu Bar Standard Graph and Statistics are checked by default and also that Menu Bar cannot be unchecked To close a toolbar except
311. re Coefficient Alpha Odd Items 0 613 Coefficient Alpha Even Items 0 661 Approximate Standard Error of Measurement of Total Score for 15 z score Intervals z score Total Score N Standard Error 3 750 4 458 0 3 250 3 298 0 2 O0 2 03 0 E RASO 0 860 0 OO 0 340 10 1 000 1250 1 539 16 1 000 0 750 Zit oo 6 1 000 0 250 3 936 29 T390 Oe 250 SaL 10 1 095 0 750 62557 8 1 000 1 250 TIO 8 0 000 1750 MS 6 1 000 2230 SS 0 2 790 11 134 0 3250 12 334 0 Item Reliability Statistics Item Standard Reliability Item Label Mean Deviation Item Total R Item Alpha 1 POUNDING 0 903 0 296 Gh 300 0 098 2 SINKING 0 785 0 411 0 499 04205 3 SHAKING 0 559 0 496 0 678 0 336 4 NAUSEOUS 0 613 0 487 0721 dal 5 STIFF OOO 0 499 0 693 0 346 6 FAINT 0 452 0 498 ie AES Ore 3 5 6 7 VOMIT 0 376 0 484 0 622 Ucar 8 BOWELS OZ LS 0 411 0625 00 201 9 URINE 0 097 0 296 04503 0 149 299 Applications Logistic Test Item Analysis The input is USE COMBATDM TESTAT MODEL POUNDING URINE FREQUENCY COUNT IDVAR COUNT ESTIMATE LOG1 The output is Case frequencies determined by value of variable COUNT 93 Cases were processed each containing 9 items 6 Cases were deleted by editing for missing data or for zero or perfect total scores after item editing O Items were deleted by editing for missing data or for zero or perfect total scores after item editing Data below are based on 87 Cases and 9 Items Total Score Mean 4 230 Standard Deviation
312. re is only a 1 4 chance that these data would be measured if the individual producing horizons have the same average level of urantum that is the uranium level differs significantly by producing horizon We saw this immediately in the Quick Graph In fact in the Quick Graph we also saw that producing horizon 3 the Quartermaster horizon differs the most 114 Chapter 4 Outliers and Diagnostics The Output editor also has warnings about outliers WARNING Case 30 is an Outlier Studentized Residual 4 732 Case 31 is an Outlier Studentized Residual 4 732 Test for Normality Durbin Watson D Statistic First Order Autocorrelation There are two outliers in the data cases 30 and 31 These are the same two that we lassoed earlier in the probability plot SYSTAT performs diagnostics to verify that the data meet the underlying assumptions for ANOVA Linear Regression and General Linear Models GLM Diagnostics speed up the analysis and help to produce more accurate results by alerting you to problems with the data Both the Durbin Watson D statistic and the first order autocorrelation appear by default and these are parts of such diagnostics The Options tab provided in the ANOVA dialog box performs diagnostics The Shapiro Wilk option performs the test for normality of residuals From the above output of Test for Normality the p value is an indication as in any hypothesis testing results of whether the hypothesis
313. re as a quantitative variable Danger index as a quantitative variable based on the above two indices SMOKE Greenacre 1984 The data comprise a hypothetical smoking survey in a company The variables are STAFF SMOKE FREQ SOCDES gt Strahan and Gerbasi 1972 The 20 item version of the Social Desirability Scale was administered as embedded items in another test to 359 undergraduate students in psychology The social desirability items were scored for the social desirability of the response and coded as 0 s and 1 s in this SYSTAT data set SOFTWARE 1 Musa 1979 The data set consists of failure times TIME in CPU seconds measured in terms of execution time of a real time command and control software system The variable INTER contains inter failure times SOIL Zinke and Stangenberger These data were taken from a compilation of worldwide carbon and nitrogen soil levels for more than 3500 scattered sites The full data set is available at the U S Carbon Dioxide Information Analysis Center CDIAC site on the World Wide Web The subset included in SYSTAT pertains to the continental U S Duplicate measurements at single sites are averaged LAT LON STATISTC CARBON Sample site latitude Sample site longitude Mean Carbon content in kg m 351 Data Files NITRO Nitrogen content in kg m ELEV Sample site elevation in meters SPECTRO Lindberg et al 1983 The data set was used to fit a spectrographic m
314. re contents of the active tab set the font of subsequently typed not generated or selected text in the Output Editor specify the printer paper size source and orienta tion to be considered while printing preview the data variable information before print ing 222 Chapter 7 Graph Editor Commandspace Ctrl P print data variable information Ctrl Z Alt Backspace undo step by step upto 32 steps of editing done Ctrl Y Ctrl F Ctrl H Ctrl R Ctrl A Alt Insert Ctrl Shift Insert Ctrl Shift Del Ctrl Shift P Shift Del Ctrl P Del F7 Ctrl L F8 Ctrl F7 Ctrl Shift V Ctrl Shift Alt P Ctrl Alt P Ctrl P Ctrl Z Alt Backspace Ctrl Y Ctrl F F3 Ctrl H Ctrl R Ctrl A Ctrl Shift F F9 Ctrl W redo step by step upto 32 steps of editing done locate a variable in the Data Editor replace instances of a string in a given column select entire contents of the active tab add empty rows in the Data Editor appends at the end of a file if one is already open insert variables in the Data Editor before or after a selected column delete the selected variables in the Data Editor open Variable Properties for the current column cut the selected variable or case print the graph that is in the Graph Editor delete any annotation that you may have created submit the contents of the active tab in the Command
315. re in the file GDWTRDM Measurements were recorded for the following variables Variable SAMPLE LATITUDE LONGTUDE HORIZONS HORIZON URANIUM ARSENIC BORON BARIUM MOLYBDEN SELENIUM VANADIUM SULFATE TOT ALK BICARBON CONDUCT PH URANLOG MOLYLOG Potential Analyses Description The ID of the groundwater sample Latitude at which the sample was taken Longitude at which the sample was taken Initials of producing horizon ID of producing horizon Uranium level in groundwater Arsenic level in groundwater Boron level in groundwater Barium level in groundwater Molybdenum level in groundwater Selenium level in groundwater Vanadium level in groundwater Sulfate level in groundwater Alkalinity of groundwater Bicarbonate level in groundwater Conductivity of groundwater pH of groundwater Log of uranium level in groundwater Log of molybdenum level in groundwater The following kinds of analyses may be useful in analyzing the groundwater data ANOVA Basic Statistics Transformations Nonparametric tests Regression Correlation Cluster analysis Discriminant analysis 101 Data Analysis Quick Tour m Spatial statistics Smoothing techniques such as kriging Contour plotting In these examples we will show you descriptive graphs ANOVA nonparametric tests smoothing and contour plotting The Groundwater Data File The data for this analysis are in the file GDWTRDM To open the file from the m
316. re reaction time in seconds RT versus angle of rotation in degrees ANGLE in a perception study The experiment measured the time it took subjects to make same judgments when comparing a picture of a three dimensional object to a picture of possible rotations of the object ROTHKOPF Rothkopf 1957 These data are adapted from an experiment by Rothkopf in which 598 subjects were asked to judge whether Morse code signals presented two in succession were the same All possible ordered pairs were tested For multidimensional scaling the data for letter signals 1s averaged across sequence and the diagonal pairs of the same signal is omitted The variables are A through Z RYAN Ryan 2002 Y and Y2 are the control variables and SAMPLE is the sample identifier SALARY These data compare the low and high salaries of executives in a particular firm The variables are SEX EARNINGS and COUNT SCHOOLS Neter Kutner Nachtsheim and Wasserman 2004 These data comprise a nested design where two teachers from each of three different schools are rated SCHOOL indicates the school that the case describes Each teacher variable TEACHER 1 3 represents a different school a value of 1 indicates teacher 1 for that school 2 indicates teacher 2 for that school and 0 indicates that the teacher does not teach at that school LEARNING measures the teacher s effectiveness the higher the better SCORES gt Hand at al 1996
317. represent the numbers N of conforming RESULT is 1 and nonconforming RESULT is 0 units produced by each of five machines MACHINE Ie Milliken and Johnson 1992 An experiment was conducted by a company to compare the performances of three different brands of machines when operated by the company s own personnel Six employees were selected at random and each of them had to operate each machine three different times The data set consists of overall scores that take into account both the quantity and quality of the output The variables are SCORE MACHINE OPERATOR and TIME MACHINE2e Milliken and Johnson 1992 It is an unbalanced data set where two machines were operated by six randomly selected operators Each operator was allowed to operate each machine at most three times 338 Chapter 9 MACKe Breslow and Day 1980 The data deals with the cases of eudiometrical cancer in a retirement community near Los Angeles The data are reproduced in their Appendix II The variables are CANCER AGE GALL Gallbladder disease HYP Hypertension OBESE Obesity EST Estrogen DOS Dose DUR Duration of conjugated estrogen exposure NON Other drugs The data are organized by sets with the case coming first followed by four controls and so on for a total of 315 observations 63 4 1 MANOVA Morrison 1990 These data are from a hypothetical experiment measuring weight loss in rats Each rat was assigned randomly to one of
318. required 1555520 squares to color either black with probability 0 29 or white with probability 0 71 From this a total of 1000 non overlapping samples each 344 Chapter 9 containing 16 of small squares were randomly selected and the number of black squares were counted in each case The data set consists of the frequency distribution of this count PATTISON Clarke 1987 In his 1987 JASA article C P Y Clarke discusses the data taken from an unpublished thesis by N B Pattinson for 13 grass samples collected in a pasture Pattinson recorded the weeks since grazing began in the pasture TIME and the weight of grass cut from 10 randomly sited quadrants then fit the Mitcherlitz equation 0 TIME GRASS 0 0 PDLEX1 Gujarati 1995 The data set relates to the SALES and INVENTORY of a product in 20 days PDLEX2 Gujarati 2003 The data set relates to the SALES and INVENTORY of a product for the United States for the period 1954 1999 PDLEX3 Gujarati 2003 The data set relates to income money supply model of USA for the period 1970 1999 The variables are as follows GDP Gross domestic product billions seasonally adjusted M2 Money supply billion seasonally adjusted GDPI Gross private domestic investment billion seasonally adjusted FEDEXP Federal government expenditure billion seasonally adjusted TBO Six month treasury bill rate PESTICIDE Milliken and Johnson 1992 Four chemical
319. ressing the corresponding Remove button Alternatively right click on a variable or selection and select Remove m Cross a variable in the source list with one in the target list by selecting them and then pressing the Cross button You can also add crossed terms of multiple variables directly by selecting these variables in the source list and pressing the Cross button m Usethe button when you want to include the variables as well as all their crossed terms You can also use this button with multiple variables m Use the Nest button to include nested terms in the target list 37 Introducing SYSTAT Selecting variables To add a single variable to the desired target list you simply highlight it in the source variable list and click the dd button Use the Remove button to undo your selection You can also double click individual variables to move them from the source list to the target list or vice versa When there is more than one target list this functionality will apply to one of them You can also select multiple variables To highlight multiple variables that are grouped together on the variable list click and drag the mouse cursor over the variables you want Alternatively you can click the first one and then Shift click the last one in the group To highlight multiple variables that are not grouped together on the variable list use the Ctrl click method Click the first variable and then
320. ria AIC Schwarz s BIC at at at at at Iterationl Iteration2 Iteration3 Iteration4 Iteration5 32 064 Dee Oe Parameter Estimates LDOSEB DO 032 z032 s032 Standard Error Applications 95 Confidence Interval Lower Upper 310 Chapter 8 Odds Ratio Estimates Parameter LDOSEB Odds Ratio Standard Error A a ee aN 4 ee ee ee eee eee 95 Confidence Interval Lower Case frequencies determined by value of variable COUNT The categorical values encountered during processing are Variables Levels Tai in A A A a a Ho gt RESPONSE 2 levels 0 000 1 000 Dependent Variable RESPONSE Analysis is Weighted by COUNT Sum of Weights 25 000 Input Records 9 Records for Analysis 9 Sample Split Category Count A oo ce R 0 RESPONSE 1 REFERENCE Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood Log Likelihood 15 000 10 000 Iteration History at at at at at Iterationl Iteration2 Iteration3 Iteration4 Iteration5 Information Criteria AIC Schwarz s BIC 3 3 2 064 AA a Parameter Estimates Parameter LDOSEB Estimate Odds Ratio Estimates Parameter LDOSEB I Odds Ratio A ee aaa es a D gt gt gt gt gt gt gt i Standard Error A SS ay ee i
321. riable We define the model with placeholders for these two variables Substituting empirical variables for these placeholders yields regression output for that model Either or both of these variables could be replaced to generate new output using the same general model for different data The ampersand character denotes tokens The text immediately following an amp corresponds to a token name Token names may contain any number of characters numbers underscores and dollar signs but the first character after the ampersand must be a letter or number Dollar signs do not denote strings and may appear anywhere in the token name As with variable names token names are not case sensitive The names amp tokn amp tOKn amp ToKn and amp TOKN are equivalent if all of these names appear in a template substituting a value for one of them also substitutes that value for the others In some instances ampersands should not be treated as token indicators For example the command USE JUNE amp JULY accesses the data file JUNE amp JULY However SYSTAT interprets the amp as a token indicator and prompts the user for replacement text for JUL Y Two methods exist for avoiding this problematic behavior m Ifthe command file does not involve any token substitution turn token processing off by including the line TOKEN OFF at the beginning of the command file or by using the General tab of the Global Options dialog Use TOKEN ON to reactiva
322. ribution 104 Chapter 4 Exploring the Groundwater Data Interactively The Graph Properties dialog box is a tool that allows you to explore data interactively increasing the efficiency of your analysis It can be used to modify features of a graph or frame or elements of the graph m To open the Graph Properties dialog box right click on the graph And click the Properties option to open the Graph Properties dialog box SYSTAT GraphGDI1 Elle Edt view Data Utiities Graph Analyze Advanced Quick Access Addons Window Help Ex p e ea aE nx a 4 SYSTAT Output HE USE A EEE USE mantis s ls DENSITY UR ls DENSITY UR E HE USE GDWTRDM ls DENSITY UR TE USE Gdwtrdm sy E HE USE mantis ls DENSITY UR Animate Spin with Mouse Realign Frames Copy Graph View Page View Y Show Toolbar Save As Options Ctrl P URANIUM Invoke the Graph Properties dialog box for editing a graph GRAPH HTM ECH SE E FRO ID T OR AP NUM 105 Data Analysis Quick Tour Axes Properties Mm fm Graph Frame Axes Element Display Transform w Font Uptions Limit lines Line Lower lo Display grid lines m Click the Axes tab in the Graph Properties dialog box and then select the Options tab Select Power in the Transform combo box This will enable the power combo box m Use the down arrow key in the keyboard to change
323. ribution E Select FOODS BRANDS and DIETS as the variables iS Analyze One Way Frequency Tables Man o Available variable s Selected variable s Cell Statistics Resampling m Click OK WITAMINA FOODS CALCIUM ate ee IRON COST DIETS Frequency distribution list layout Sample mode i End list after i LOWS Display rows with zero counts Counts and percents Percents Measures Pearson chi square Include missing values C Save table s 64 Chapter 3 Frequency Distribution for FOODS FOODS Frequency Cumulative Percent Cumulative i Frequency Percent Se E as Ss e beef i 6 6 21 429 21 429 chicken 14 20 50 000 71 429 pasta l 8 28 28 571 100 000 Frequency Distribution for BRANDS BRANDS Frequency Cumulative Percent Cumulative l Frequency Percent ee D gor i 4 4 14 286 14 286 hc 3 7 BOALO 25 000 Le i 5 12 LAO 42 897 st i 4 L6 14 286 AA E sw 3 19 Low LA 67 857 ty i 4 23 14 286 SZ LAS WW i 5 28 17 857 100 000 Frequency Distribution for DIETS DIETS Frequency Cumulative Percent Cumulative i Frequency Percent na a TA E H no 15 15 KZ SAL ako PSA yes l 13 28 46 429 100 000 In above output for FOODS the name appears at the top left in the first table 14 of the 28 dinners in the sample 50 in the Pct column are chicken 28 6 are pasta and 21 4 are beef The number of dinners per BRAND
324. rk John Wiley amp Sons 360 Chapter 9 Bennett R M and Desmarais R N 1975 Curve fitting of aeroelastic transient response data with exponential functions In Flutter Testing Techniques Report of a conference held at Dayton Flight Research Center Edwards CA October 9 10 1975 Washington DC NASA Pp 43 58 Birkes D and Dodge Y 1993 Alternative methods of regression New York John Wiley amp Sons pp 177 183 Bishop Y V V Fienberg S E and Holland F W 1975 Discrete multivariate analysis Cambridge MA MIT Press Bliss C I 1967 Statistics in biology New York McGraw Hill Borg I and Lingoes J 1987 Multidimensional similarity structure analysis New Y ork Springer Verlag Box G E P Jenkins G M and Reinsel G 1994 Time series analysis Forecasting amp control 3 ed Upper Saddle River NJ Prentice Hall Breiman L Friedman J H Olshen R A and Stone C I 1984 Classification and regression trees Belmont Calif Wadsworth Breslow N and Day N E 1980 Statistical methods in cancer research Vol II The design and analysis of cohort studies Lyon IARC Breyfogle F W III 2003 Implementing six sigma Smarter solution through statistical methods 2nd ed New York John Wiley amp Sons Brockwell P J and Davis R A 1991 Time series theory and methods Springer Verlag Brodlie K W 1980 A review of methods for curve and functi
325. rm the same transformation on several variables you can use the sign instead of typing a separate line for each transformation For example LET gdp cap L10 gdp cap LET mil ELO mil LET 9np 86 110 gnp 896 is the same as LET gdp cap mil gnp 86 110 136 Chapter 5 The sign acts as a placeholder for the variable names The variable names must be separated by commas and enclosed within parentheses Autocomplete commands As you begin typing commands in the Interactive or batch Untitled tab of the Commandspace you will be prompted with the possible command keywords available data files or available variables When a letter is typed all commands beginning with that letter will appear in a dropdown list Select the desired command or continue typing For a command involving file names on pressing space and then any letter the files of the relevant folder as specified in the File Locations tab of the Edit Options dialog box beginning with that letter will be listed For a command involving variable names if a data file is open all available variable names beginning with that letter will appear in a drop down list When you type expressions the relevant function names will be shown In general for any given letter that you type the relevant arguments options and option values will be listed If you do not know the exact syntax of a particular command press Ctrl Spacebar to get a list of
326. rom the Belgian Statistical survey describes the number of international phone calls from Belgium in years 1950 1973 The variables are X Years Y Number of phone calls PHOSPHOR Hocking 1985 The data set is about the concentration of phosphorus in the wash water The aim of the investigation is to determine how the concentration varies with the types of detergent and washing machines The experiment was carried out with four different types of detergents three different types of machines and seven laundromats The laundromats had different numbers of machines but each laundromat had only machines of a single type Thus laundromats are nested inside machine types The machines within each laundromat were divided into four groups of roughly equal sizes and the four types of detergent were allocated to them The response is the average amount of phosphorus in grams per liter from daily one hour samples over a seven day period The variables are Y N MACHINE LAUNDRY DETERG PHYSICAL Crowder and Hand 1990 The data set shows three groups of diabetic patients and one control group GROUP The response variable is observed at 12 time points and the corresponding variables are X X2 amp Y through Y 0 respectively PISTON Taguchi El Sayed Hslang 1989 This data set consists of diameter differences D A between the cylinder and the piston of a six cylinder engine The sample was selected from a month s MONTHS production of an
327. rovides a number of commands to save and print output as well as to control its appearance These commands may be particularly useful when creating command files OUTPUT command Enables you to route subsequent plain text output to a file or a printer PAGE command Enables you to specify a narrow 80 columns the default or wide format 132 columns for output You can also specify a title that appears at the top of each printed output page FORMAT command Enables you to specify the number of character spaces per field displayed in data listings and matrix layouts and the number of digits printed to the right of the decimal point You can also display very small numbers in exponential notation instead of being rounded to 0 147 Command Language NOTE command Enables you to add comments to your output For example NOTE THIS IS A COMMENT This is the second line of comments It s the third line here Each character string enclosed in either single or double quotation marks is printed on a separate line A note can span any number of lines and and can contain ASCII codes to display the corresponding ASCII characters Translating Legacy Commands SYSTAT provides a feature whereby you can translate legacy command files to the current command syntax supported by SY STAT 13 You can either translate commands that are in a file or directly type the commands to be translated To translate legacy command files from the
328. rst two characters represent the corresponding volume of the printed manual as follows da for Data called Data Volume in the Command folder gs for Getting Started gr for Graphics sl for Statistics I s2 for Statistics I s3 for Statistics II s4 for Statistics IV s5 for Quality Analysis if installed s6 for Monte Carlo if installed s7 for Exact Tests if installed The next two digits represent the chapter number within the volume and the last two digits represent the example number within the chapter These files are organized in the Command folder with nine subfolders seven of them corresponding to the seven volumes mentioned above a GraphDemo subfolder and a Miscellaneous one which contains commands of examples which are not numbered The names of files in the Miscellaneous folder are indicative of the examples they relate to For example to execute the commands given in Example 1 in Chapter 2 of Statistics IH submit the 42 Chapter 2 s30201 syc file Depending on your file location you may have to define paths for files and rename them appropriately Glossary The glossary offers an alphabetical listing of terms commonly encountered in statistical analyses The buttons at the top of the glossary scroll the window to the corresponding letter Clicking a glossary entry reveals the definition for that term E SYSTAT 13 DER e gt A Hide Back Forw Stop Refresh Home Print Op
329. rvals BOF EOF BOG EOG Character CHR SNUM CODE LEN Date Time FDAYM LDAYM FDAYW LDAYW ODD ROUND SINH NCAT MON 12 Chapter 1 32 33 34 Statistical BGCF GDCF P6CF BGDF GDDF P6DF BGIF GDIF P6IF BGRN GDRN PORN EMCF PSCF PECF EMDF PSDF PEDF EMIF PSIF PEIF EMRN PSRN PERN FOCUS Command SYSTAT now provides a FOCUS command for switching focus to the Data Editor Graph Editor or Output Editor Use 1t in command scripts to retain or force focus to be in a particular page of the Viewspace Macros SYSTAT now allows you to define and call macros in your command scripts A macro is a series of statements enclosed by the DEFMACRO and ENDMACRO commands Macros may be used to execute a set of commands in many different places in a program FUNCTION Command For user defined functions you now need to specify the type of the argument and the return type of the function as TMP The syntax of the FUNCTION command is now as follows FUNCTION TMP funcname TMP argl TMP arg2 statement statement2 RETURN expression ENDFUNC 13 What s New and Different in SYSTAT 13 35 Multiple Option Values SYSTAT now expects multiple option values to be enclosed in braces For example if you want to specify three colors for an overlaid graph type the option as COLOR MAGENTA BLUE YELLOW 36 PAGE NONE You can now set the page width to be unlimited using the PAGE NONE command 37 Preceden
330. s 1 00000E 5 in exponential notation A number that would otherwise violate the specified field width will also be converted to exponential notation while maintaining the number of decimal places Individual variable formats in the Data Editor override the default setting 239 Customization of the SYSTAT Environment Locale SYSTAT determines the initial default decimal and digit grouping symbols for numbers from the current settings in the Regional and Language Options dialog of the Windows Control Panel This is recognized as the System default You may change the setting to any of the locales provided in the dropdown list A sample number will be displayed alongside You may suppress digit grouping if you do not want digits to be grouped With this option you will be able to enter numbers in the Data Editor using the decimal and digit grouping symbols of your chosen locale The output displayed in the Output editor will also adhere to these locale specific settings You can thus create output suitable for any given locale Output results These settings control the display of the results of your analyses m Length specifies the amount of statistical output that is generated Short provides standard output the default Some statistical analyses provide additional results when you select Medium or Long Note that some procedures have no additional output Tip In command mode DISCRIM LOGLIN and XTAB allow you to add or delete items se
331. s between the diet and regular dinners Let us explore further by using both food type and dinner type to define cells that is we request a two way analysis of variance Using the Counts feature in Two Way Tables we found that although our sample has beef chicken and pasta dinners there were no beef dinners in the DIET yes group SYSTAT can analyze ANOVA designs with missing cells See SYSTAT Statistics II Chapter 3 for more information Let us use Select Cases on the Data menu to omit the beef dinners and then request an analysis of variance for a two by two design DIETS yes and no by chicken and pasta E From the menus choose Data Select Cases In the Select dialog box select FOODS as Expression1 Select lt gt not equal from the drop down list of operators For Expression2 type beef include the quotation marks while working with commands the dialog box takes care of this m Click OK 90 Chapter 3 ul Data Select Cases Avallable yvariable s Function type aan FOOD CALORIES FAT PROTEIN Mode of input VITAMINA CALCIUM Select O Type Functions Add to Expression Condition Expressionl Operator SELECT FOODS Complete Turn off a To get a bar chart of the cell means E From the menus choose Graph Bar Chart m Select CALCIUM as the Z variable DIETS as the Y variable and FOOD as the X variable Click the Error Bar tab and select none from the type
332. s can be saved in a number of graphic formats When you choose Save Active File from the File menu what is saved depends on which pane is active If either the Output Organizer or the Output editor is active the entire contents of both panes are saved When you choose Save All from the File menu the current output data file and the current file of the commandspace are all saved 194 Chapter 6 To Save Output SYSTAT displays statistical and graphical output in the Output editor Click the Output Organizer or Output editor and choose Save As from the File menu to save the contents ofthe pane You can save Data Command Output Graph or Log using Save from File menu Save As Soren TE BackGround C SysHtml Temp Themes 5 Tutorial SYSTAT Output syo Select a directory and specify a name and file type for the output Output can be saved in SYSTAT Output SYO Rich Text Format RTF Rich Text Format Wordpad compatible RTF Hyper Text Markup Language HTM or MHT format Note Unlike output saved in SYO or RTF format output saved in HTM or MHT format preserves some properties HTML or MHT outputs are not editable m As HTML or MHT underlies web page creation presenting the resulting output on the Internet involves simply creating a link from a web page to the filename htm or mht file In addition HTML or MHT output allows sharing your results with colleagues who do not yet have SYSTA
333. s in its Basic Statistics module These are variables that contain the computed values of various statistics for a given session a given data file and given variables These may be directly used in subsequent transformation statements for further processing of the computed statistics For details refer to Chapter 5 Command Language Hypothesis Testing for Multivariate Mean The Hypothesis Testing feature has been strengthened with tests for mean vectors of multivariate data m One sample Hotelling s T2 test for mean vector of multivariate data equal to a known vector Two sample Hotelling s T2 test for equality of two mean vectors of multivariate data New Basic Statistics SYSTAT now offers the following new basic statistics m Standard error and confidence interval for the trimmed mean m Winsorized mean its standard error and confidence interval Sample mode m Interquartile range Bootstrap Analysis in Hypothesis Testing The Hypothesis Testing feature now provides m Bootstrap based p values for all tests for mean one sample z one sample t two sample z two sample t paired t Poisson and variance single variance two variances and several variances New Nonparametric Tests The Nonparametric Tests feature has been updated to include m Jonckheere Terpstra test for ordered differences m Fligner Wolfe test for control vs treatments The following pairwise comparison tests m Dwass Steel Critchlow Fligner E Co
334. s the file to save results and data to but does not in itself trigger the saving of results the next HOT command does that Command Syntax Rules Upper or lower case Commands are not case sensitive You can type commands in upper or lower case or both CSTATTSTICS or cstatistics or CStatistics The only time SY STAT distinguishes between upper and lower case is in the values of string variables In other words for a variable named SEX SY STAT considers the text values male and MALE to be different 131 Command Language Abbreviating commands You can shorten commands and options to the first two to seven letters as long as the resulting abbreviation is unique and the largest expansion sounds nice commonly used For e g COV COVA and COVAR will all be permissible abbreviations of COVARIANCE For commands abbreviations till the full word even beyond 7 characters will be supported For example E CSTATISTICS can be shortened as CSTA or CST m DENSITY var can be shortened as DEN var m HELP phrase can be shortened as HE phrase In the case of commands within a module the abbreviation needs to be unique within the module For example STAR STAN STE and STO will be interpreted as START STANDARDIZE STEP and STOP respectively within the GLM module Outside GLM STAN will be treated as STANDARDIZE the command to standardize variables Note BASIC commands module and variable names must be typed in ful
335. s the price of one share of common stock divided by the earnings per share for the past year This ratio shows the dollar amount investors are willing to pay for the stock per dollar of current earnings of the company RORS Percent rate of return on total capital invested plus debt averaged over the past 5 years DE RATIO Bept to equity invested capital ratio for the past year This ratio indicates the extents to which management is using borrowed funds to operate the company 331 Data Files SALESGR5 percent annual compound growth rate of sales computed from the most recent five years compared with the previous five years EPS5 percent annual compound growth in earning per share computed from the most recent five years compared with the previous five years NPMI Percent net profit margin which is the net profits divided by the sales for the past year expressed as a percentage PAYOUTRI Annual dividend divided by the latest 12 month earnings per share This value represents the proportion of earnings paid out to shareholders rather than retained to operate and expand the company FOREARM1 Pearson and Lee 1903 The data set consists of ARMLENGH that is length of forearm in inches of 140 men FOSSILS The data give the incidence of fossil specimens of various flora found at various elevations of a site in British Columbia The variables are HEIGHT CHARA NITALLA JUNCUS RUMEX FRACTION These data are from a h
336. s the symbol size variable CALORIES CALORIES AO 20 eh Notice the back corner of the display on the left the tallest spike extends to sw indicating the dinner with the most calories On the floor of the display we read that its fat content is between 20 and 30 grams and that its protein is a little over 20 grams We see this same point in the display on the right the size of its circle is not extreme indicating a mid range price Notice the small circle toward the far right this dinner costs much less than the sw dinner and has a higher fat content and a similar protein value The most expensive dinners that is the larger circles do not concentrate in a particular region 78 Chapter 3 A Two Sample t Test One of the most common situations in statistical practice involves comparing the means for two groups For example does the average response for the treatment group differ from that for the control group Ideally the subjects should be randomly assigned to the groups For the food data we are interested in possible differences in PROTEIN and CALCIUM between the diet and regular dinners Thus the dinners are not randomly assigned to groups In a real observational study a researcher should carefully explore the data to ensure that other factors are not masking or enhancing a difference in means In the t test we test the hypothesis Hy Means of diet and regular dinners are equal The alternati
337. se the Data menu to define categorical variables transform including recode data values rank center or standardize data trim extreme values sort cases in the data file based on the values of one or more variables transpose cases rows and variables columns wrap unwrap or stack variables merge data files cases or variables define ID variables and order of display of data values specify grouping variables that split the data file into two or more groups for analysis select and extract subsets of cases list data in the Output editor define case frequencies and weight data for analysis based on the value of a weight variable When the Data editor is active you can also define variable properties and value labels as well as edit data Utilities Use the Utilities menu to access SYSTAT s MATRIX module perform probability calculations generate random samples from a variety of univariate discrete and continuous probability distributions generate a variety of experimental designs perform power analysis and calculations involving functions available in SYSTAT including probability calculations retrieve data file information and current SYSTAT settings record macros 1 e command scripts generated by actions of the user and play them create command file lists and customized user menus access recently invoked dialogs save apply and download SYSTAT menu themes as well as add examples to the Examples tab Graph Use the Graph menu to
338. serted at the beginning of its group You can click the arrow in the top right corner of the Viewspace and check Active Tab at the Beginning if you want a new tab to appear as the first tab of the Viewspace You can bring a tab into focus by clicking the arrow and checking the name of the desired tab If there are more tabs than are directly visible in the Viewspace the tab becomes the first tab in the Viewspace or in its group depending on whether Active Tab at the Beginning is checked or not This is especially useful when you have a lot of tabs open in the Viewspace 27 Introducing SYSTAT You can close an active or inactive data file by right clicking and selecting Close or by bringing the tab into focus and pressing the Close button in the top right corner of the Viewspace Workspace The Workspace consists of three tabs Output Organizer m Examples m Dynamic Explorer Output Organizer Use the Output Organizer primarily to navigate through the results of your statistical analysis Selecting a completed procedure from the outline displays the corresponding results in the Output editor You can also use the Output Organizer to select an item and then copy paste delete or move it allowing you to tailor SYSTAT s output to your preferences In addition you can quickly move to specific portions of the output without having to use the Output editor scrollbars For more information about the Output Organizer see Chapter 6
339. space submit the command line on which the cursor is currently positioned submit the selection in the active tab of the Commandspace submit a command file submit the contents of the clipboard specify the printer paper size source and orientation to be considered while printing preview the output before printing print data toggle between undoing and redoing the last step of editing redo the step that was last undone find text find the next instance of the text specified for the search replace text select entire contents of the active tab set the font to be used in the active tab recall commands from the command buffer one by one starting from the latest toggle visibility of Commandspace 223 Customization of the SYSTAT Environment Access keys Access keys provide an alternative to accelerator keys for accessing menu entries Access keys open menus using the Alt key and allow navigation to selected entries using designated letters m The name of each menu contains one underlined letter Pressing Alt and the underlined letter opens the corresponding menu After opening a menu you can execute any of the displayed entries m Like the menu titles each menu entry contains one underlined letter Pressing this letter runs the entry as if 1t had been selected using the mouse The list of access keys 1s too long to be displayed here To view the key required for a particular menu entry open the men
340. splayed in the Output editor of the Viewspace 107 Data Analysis Quick Tour Histogram for Uranium Probability Plot for Uranium 80 0 6 70 P 0 5 2 Cc a 3 1 50 vee O z E Da 3 40 0 33 S 0 O A 30 2 w E A 0 2 S E gt 0 1 2 10 0 0 0 3 0 00 150 0 5 00 150 50 1 0 1 URANIUM URANIUM In this plot we begin to glimpse SYSTAT s color and overlay capabilities This command file created a side by side overlay of a histogram and a probability plot of the URANIUM variable SYSTAT Windows and Commands SYSTAT gives you the flexibility to perform your analysis the way you want m Windows interface icons menus and dialog boxes m Typed commands typing commands at the Commandspace Batch Untitled command files submitting files directly or from the Commandspace Additionally all menu actions can be optionally echoed to the Output editor allowing you to perform initial analyses using the menus and then to cut and paste the commands into the Untitled tab of the Commandspace for repeated use THICK 2 AUSE GDWTRDM BEGIN DENSITY URANIUM FCOLOR BLUE COLOR 3 FILL TITLE Histogram for Uranium PPLOT URANIUM LOC 6IN OIN FCOLOR GRAY FILL COLOR 4 TITLE Probability Plot for Uranium END THICK 108 Chapter 4 Plotting Several Graphs Using Commands The commands in the file GDWTRIDM are THICK 2 USE GDWTRDM BEGIN DENS URANIUM HIST FCOLOR BLUE COL
341. string or membrane loosely attached to each data point then the higher the tension on the ends of the string the less influence any individual point has and the smoother averages across them all The lower the tension on the ends of the string the greater the influence of the individual data points and the smoother approaches a path that passes through each point 119 Data Analysis Quick Tour In addition to rotation with the help of Graph Properties dialog box you can also alter the tension of the kriging smoother To open the Graph Properties dialog box right click on the graph editor and select Properties Click the Graph tab in the Graph Properties dialog box Use the up arrow key in the keyboard to select the graph as Actual Uranium and Kriging Smoother by Geography Now click on the Element tab and select the Smoother tab Select Kriging from the Method combo box Use the down arrow key to change the tension value from 0 35 to 0 90 in Tension combo box Actual Uranium and Kriging Smoother by Geography Uranium Notice how the surface becomes flatter and lower recall from the histogram that most samples have a low value for the uranium level Decrease the tension from 0 90 to 0 10 120 Chapter 4 Actual Uranium and Kriging Smoother by Geography Uranium Notice how the surface reaches out to each individual point Page View If at this point you switch to the Page view by selecting from the
342. studies m Wichmann Hill This generates random numbers by a triple modulo method 236 Chapter 7 Mersenne Twister MT is the default option We recommend the MT option especially 1f the number of uniform random numbers to be generated for your Monte Carlo exercise 1s large say more than 10 000 If you would like to reproduce results involving random number generation from earlier SYSTAT versions with old command files or otherwise make sure that your random number generation option is Wichmann Hill and of course that your seed is the same as before For more details see Chapter 4 Data Transformations ofthe Data volume and user documentation on Monte Carlo if you have the Monte Carlo add on module Bubble Help Apart from the help provided on the status bar about each menu item a more detailed description 1s provided in a bubble that appears when you pause the mouse on the menu item for a few seconds You can specify the number of seconds to pause the mouse before the help appears or even turn off the help completely Default command file format SYSTAT provides two formats for saving command files For a given file you do have the option of saving in the ANSI format using the File type dropdown in the Save File dialog box The default choice may be set to one of the following m Unicode SYSTAT command files will be saved in the unicode format by default m ANSI SYSTAT command files will be saved in the ANSI format b
343. tability in panel models Sociological methodology D R Heise Ed 84 136 San Francisco Jossey Bass Wilkinson L 1975 The effect of involvement on similarity and preference structures Unpublished dissertation Yale University Wilkinson L 1988 SYSTAT The system for statistics Evanston IL Systat Inc Wilkinson L 2005 The grammer of graphics 2 ed New York Springer Verlag Wilkinson L Blank G and Gruber C 1996 Desktop data analysis with SYSTAT Upper Saddle River N J Prentice Hall Wilkinson L and Engelman L 1996 SYSTAT 7 0 New Statistics pp 235 SPSS Inc Williams D A 1986 Interval estimation of the median lethal dose Biometrics 42 641 645 Winer B J 1971 Statistical principles in experimental design 2d ed New York Mc Graw Hill Winer B J Brown D R and Michels K M 1991 Statistical principles in experimental design 3rd ed New York McGraw Hill Wludyka P S and Nelson P R 1997 An analysis of means type test for variances from normal populations Technometrics 39 3 274 285 Acronym Abbreviation A ABS absolute value ACF autocorrelation function ACT actuarial life table AD test Anderson Darling test AIC Akaike information criterion AID automatic interaction detection ALT alternative ANCOVA analysis of covariance ANOVA analysis of variance AR autoregressive ARCH Autoregressive Conditional Heteroskedasticity
344. te token processing for subsequent command submissions m If some ampersands denote tokens but others do not suppress token processing wherever needed by doubling the ampersand character For example replace JUNEGJULY with JUNE amp amp JULY SYSTAT interprets two consecutive ampersands as a single character rather than a token indicator As SYSTAT processes commands token substitution occurs either automatically or interactively In automatic substitution information supplied in the template replaces 158 Chapter 5 placeholders as they are encountered Interactive substitution on the other hand involves prompting the user for placeholder replacement information Command processing halts until valid information is supplied Automatic Token Substitution Define tokens for automatic substitution by specifying TOKEN amp tok value When SYSTAT encounters amp tok during command submission the defined value replaces the token automatically Quotes around token values are NOT included in the replacement value of the token For example TOKEN amp strl Depression LABEL dscore 1 amp strl BAR dscore XLAB amp strl TITLE Bar graph of amp strl defines the token amp str to have a value of Depression In the bar graph Depression appears entirely in capital letters for the tick label corresponding to 1 label but not for the title Because the token value does not include the quotes the value can be incorpor
345. te Special you can specify whether you want to paste the clipboard contents as text or a Windows Metafile graphic Note that Paste Special is not available in all applications m For columns to line up properly you must highlight text output after you paste it and apply a fixed pitch font for example Courier or Courier New Or use Paste Special on the Edit menu to paste the text as a metafile graphic 199 Working with Output Printing In any SYSTAT window choose Print from the File menu to open the Print dialog box 2 Print Select Printer E Add Printer Print to file Preferences Number of copies Enter either a single page number or a single oe pi page range For example 5 12 Select a printer and a print range You can choose to print the current selection the entire print range or a specific page range Use the Print Preview command in the File menu to preview the content before printing it 200 Chapter 6 Print Preview In any SYSTAT window choose Print Preview from the File menu to display the active document as 1t would appear when printed When you choose this command the main window will be replaced with a print preview window in which one or two pages will be displayed in their printed format The print preview toolbar offers you options to view elther one or two pages at a time move back and forth through the document zoom in and out of pages and initiate a print job Page Setu
346. te an image in the Picture area The Picture area is split into pixels arranged in 16 rows by 15 columns Clicking in the Picture area using any of the tools colors the pixels in various ways Pencil Fills any pixel that you click on with the color selected in the Colors area Fill Fills the enclosed area with an unbroken boundary made of a non default color in which you click with the selected color Color selection Reads the color of the pixel that you click on and automatically selects that color in the Color area Line Draws a line of the selected color along the pixels over which you press and drag the pointer Rectangle Draws a rectangle of the selected color the line over which you press and drag the pointer being the diagonal Ellipse Draws an ellipse of the selected color the line over which you press and drag the pointer being the diagonal Copy Copies the image in the Picture area to the clipboard 216 Chapter 7 m Paste Pastes the image in the clipboard to the Picture area m Delete Clears the image in the Picture area When you press OK the image will be displayed in the User defined image area Press OK to use it or press Edit to edit it further Button Customization The option to edit button appearance is also available for items in the Commands list that have default images In fact you can edit the button appearance and also do a lot more for any menu menu item or toolbar button A me
347. te group of dams The matings that resulted in male progeny calves were used for an inheritance study of birth weights The birth weights of eight male calves in each of five sire groups are given The variables are SIRE BIRTHW PROGENY and GR LAB Jackson 1991 The data set consists of four bivariate vector observations per laboratory Samples were tested in three different laboratories LAB using two different methods METHOD1 METHOD2 and each LAB received four samples LABOR U S Bureau of Labor Statistics These data show output productivity per labor hour in 1977 U S dollars for a 25 year period YEAR Other variables are US CANADA JAPAN and GERMANY and ENGLAND LATIN Neter Kutner Nachtsheim and Wasserman 1996 These data are from a Latin square design in which the response RESPONSE in each square SQUARE is from one of five days a week DAY for five weeks WEEK LAW Efron and Tibshirani 1993 The law school data A random sample of size 15 was taken from the universe of 82 USA law schools Two variables are average score on a national law test LSAT and average undergraduate grade point average GPA LEAD Ott and Longnecker 2001 The data set consists of lead concentrations mg kg dry weight of 37 stations in Kenya obtained from a geo chemical and oceanographic survey of inshore waters of Mombasa Kenya LEARN Gilfoil 1982 These data demonstrate a quadratic function with a ceiling They are fr
348. te or partial text SYSTAT searches the specified direction up or down from the current location A string search may consist of only letters or letters with numbers and punctuation For any search involving letters you can impose a case restriction For example selecting Match case prevents a search for median from finding Median Note SYSTAT operates in the active space Click the Output editor to make it active If the Commandspace is active SYSTAT searches in the active tab of the Commandspace Output Editor Right Click Menu Right clicking in the Output editor provides standard editing features These are Cut Cut the selection and place it in the clipboard for pasting at the desired location s m Copy Copy the selection and place it in the clipboard for pasting at the desired location s Paste Paste previously cut or copied output Delete Delete the selections in the active tab Copy All Copy all the content in the Output editor View Source View the HTML source code 189 Working with Output Refresh Refresh the content being viewed in the Output editor Print Preview Display the file in the active tab as 1t would appear when printed You can view multiple pages at a time scroll through and zoom in or out of pages Collapse All Expand All Collapse Expand all the links in the Output editor Show Toolbar Show or hide the Format Bar New Output Open a new output file in the Output editor where furth
349. ter 6 To Save Graphs SYSTAT displays graphs in the Output editor of the Viewspace You can save the graphs along with the output by using the Save on the File menu To save an individual graph double click the graph to activate the Graph tab and use Save As on the File menu Save in Data v Oo Fi Es ms My Documents Computer File name My Network Save as type By default the file is saved as a Windows Metafile WMF You can select a different file type from the drop down list Available formats include Windows Metafile WMF Windows Enhanced Metafile EMF Encapsulated Postscript EPS PostScript PS JPEG JPG Windows Bitmap BMP 197 Working with Output Computer Graphics Metafile binary or clear text CGM Tagged Image File Format TIFF Graphics Interchange Format GIF Portable Network Graphics PNG Depending on the graphic format you can select from a number of options when saving the file See the online help for details Using Commands To save an individual graph enter the following GSAVE FILENAME FILETYPE For FILETYPE enter one of the following WMF EMF EPS PS JPG BMP TIFF GIF or PNG SY STAT saves the most recently created graph as FILENAME Issuing multiple consecutive GSAVE commands results in multiple graphs being saved SYSTAT saves the most recent first the graph created before the most recent graph second and so on However issui
350. ter data we know exactly where the uranium is geographically concentrated both in terms of producing horizon and latitude and longitude We also have some very high quality graphics to communicate our findings in print or in a presentation SY STAT has taken us from data to discovery By the way this groundwater application has many other areas to explore other than the few that we have examined in this tour For example we have not even looked at the relationships between uranium and the other elements in the data set You are encouraged to explore the power of SYSTAT further through this application beginning with any of the other potential analyses mentioned earlier 124 Chapter 4 Alternatively examine any of the other 16 applications provided with SYSTAT You can access them either through the Application Gallery in the Help system Table of Contents or through the chapter Applications on p 247 in the Getting Started manual References for Groundwater Data The groundwater data used in these examples were obtained from the following sources Original Source Nichols C E Kane V E Browning M T and Cagle G W 1976 National Uranium Resource Evaluation Northwest Texas Pilot Geochemical Survey Union Carbide Corporation Nuclear Division Oak Ridge Gaseous Diffusion Plant Oak Ridge Tenn K UR 1 U S Department of Energy Grand Junction Colo GJBX 60 76 231 Data Reference Andrews D F and Herzberg
351. the Analyze Basic Statistics dialog box select all of the variables in the source list only numeric variables are available for this feature and click OK to calculate the default statistics 67 H Analyze Basic Statistics Han Available variable s H amp P Tiles CALORIES FAT PROTEIN VITAMINA CALCIUM IORI Options E All options N Mininumn M asimum Sum Normality Set Conditions Resampling Median Mode SYSTAT Basics Selected varable s CALORIES FAT PROTEIN VITAMINS CALCIUM Range Interguartile range Geometric mean GM C Skewness Harmonic mean HM SE of skewness Arithmetic mean AM SD SE of AM O Cl of AM O F Trimmed mean Th SE of TM Cl of TM Save statistics CALORIES N of Cases 28 000 Minimum 160 000 Maximum 550 000 Arithmetic Mean 303 214 Standard Deviation 871 815 IRON ja a Ni A A E GAR ae a 4 N of Cases 28 000 Minimum 2 000 Maximum 25 000 Arithmetic Mean 10 464 Standard Deviation 5 467 d Lv Vanance Winsorized mean i M F J PROTEIN Kurtosis SE of kurtosis EE Ed SE OF thy Cl ot iM er VITAMINA CALCIUM 28 000 28 000 0 000 0 000 100 000 40 000 18 929 10 857 LS 10 845 68 Chapter 3 For each variable SYSTAT gives the number of cases with nonmissing values the largest and smallest values and the mean and standard deviation CALORIES for a s
352. the MS DOS Prompt from the Windows Start Menu or the Windows Run dialog and type the following command line with appropriate command switches filepath1 App systat exe switch es filepath2 filename xxx where filepath is the SYSTAT installation folder path filepath2 is the location of the file on which SYSTAT will operate The quotes are required only if there are gaps in the file path or filename Depending on the switch es and xxx you give the tasks described below can be automated Switch xxx Description Example command IX syc or cmd Opens SYSTAT and submits filename syc Systat x c data namel syc Opens SYSTAT and loads filename xxx c ad onto the Untitled tab of the Command Systat e emy i data name2 cmd space Opens SYSTAT submits filename xxx lelx SYCON emg and exits the application 1f file not found pystal Je Ix c data name3 syc errors are encountered Opens SYSTAT executes any commands escem cgm the user may give and on exit automati Systat gscgm cally saves in CGM format all graphs in c graphs my graph cgm the Output Editor 154 Chapter 5 elog gexit IX m out mht q dat SyC XXX dat mht XXX Opens SYSTAT and stores all error mes sages encountered during command execution into filename xxx Opens SYSTAT submits filename xxx and exits the application if no graph is generated on running it Opens SYSTAT with its window mini mized you can
353. the SYSTAT Command folder at the time of installation 244 Chapter 7 Output files Associates the designated folder with all SYSTAT SYO as well as HTML MHT output files SYO When opening or saving output files using the menus the dialogs initially default to this folder ASCII output files Sets the folder used for saving ASCII output files DAT created using the OUTPUT command Export graphs Identifies the folder used for saving all graphic formats Basic GET Defines the folder used for reading ASCII files DAT using the GET command m Basic PUT Defines the folder used for writing ASCII files DAT using the PUT command Export HTML Identifies the folder used for saving all HTML files Export RTF Identifies the folder used for saving all RTF files Using Commands Among the general options use TOKEN ON or OFF to switch token substitution on or off The following commands specify global output display options FORMAT m n UNDERFLOW Indicates the format for numeric output DISPLAY SHORT MEDIUM Defines the length of statistical output LONG Bene MARRON Indicates the width of the output WIDE VDISPLAY LABEL NAME Defines the use of variable labels in the output BOTH LDISPLAY LABEL NAME Defines the use of value labels in the output BOTH 245 Customization of the SYSTAT Environment GRAPH Includes Quick Graphs generated by statistical procedures in the output Use GRAPH NONE to sup
354. the SYSTAT Environment Utilities User Menu Command File List Lists Displays all defined command file lists Select a list to view the names of all command files assigned to the list in the List Contents list You can define lists or remove defined lists as described below Once you do that select a list to assign it to the Submit From File List button and menu item SYSTAT automatically links the two You can change the list assigned to the toolbar button by selecting a different list at any time List Contents Displays the names of the command files assigned to the selected list You can assign files to or remove assigned files from the list For example suppose you have a file in C Folder1 that produces a plot of residuals against predicted values and another file in D Folder2 that produces a probability plot of residuals You can assign both files to a list called Regression Diagnostics The only condition is that the files should be text based Modify the index of command file lists or the contents of any list using the two customization tools For the index of command file lists these buttons have the following functions m Insert Row Creates a new command file list Alternatively right click in the Lists header and select Insert Row Once a row 1s created you can even press the Enter key to create more rows After inserting a row type a name for the new list The default name is set to List You can replace it by a suitabl
355. the power value of the X axis until the graph becomes a bell shaped curve As you do this SYSTAT is automatically calculating the power data transformation of the form URANIUM power A power of 0 5 is a square root transformation A power of 0 333 is a cube root transformation Transformed Graph At a power of 0 SYSTAT automatically performs a logarithmic transformation for example log URANIUM The log transformation appears to produce a very good bell shaped curve But this judgment is subjective and it is possible to use more formal and objective methods to examine the normality of the transformed data 106 Chapter 4 40 0 3 a U O 0 20 7 O 3 3 20 5 O O D om UJ 0 1 D 10 7 0 0 0 0 1 1 0 10 0 100 0 URANIUM Normally once the proper transformation has been identified using the Graph Properties dialog box you create the transformed variable using the Data editor We have already performed the transformation and included the variable URANLOG in the data file for further statistical analysis Histograms and Probability Plots Let us take another look at the URANIUM distribution We are going to plot two graphs a histogram and a probability plot by using commands From the menus submit the command file GDWTRIDM For this E From the menus choose File Submit File E Select GDWTRIDM from the Miscellaneous subfolder of the command directory and click Open m The following graphs are di
356. this study mongrel dogs were divided into four groups of four The groups received different drug treatments The dependent variable blood histamine in mg ml was measured at four times HISTAMINE HISTAMINE2 HISTAMINES and HISTAMINE4 after administration of the drug The data are incomplete since one of the dogs is missing in the last measurement HOSLEM Hosmer and Lemeshow 2000 The variables are ID LOW AGE LWT RACE SMOKE PTL AT UI FTV BWT Identification Code Low infant birth weight Mother s age Mother s weight during last menstrual period 1 white 2 black 3 other Smoking status during pregnancy History of premature labor Hypertension Uterine irritability Number of physician visits during first trimester Birth weight HOSLEMM Hosmer and Lemeshow 2000 It already exists in SY STAT as HOSLEM Four new variables are added to it which are fictitious The variables are SETSIZE GROUP REC DEPVAR The number of subjects in each strata which is AGE for this analysis Identity number of strata Case number The relative position of the case in a given matched set 335 Data Files HW It is a hypothetical data of height and weight of a group of people according to gender ILEA Goldstein 1987 It is a subset of data from the Inner London Education Authority ILEA The data consists of information about 2069 students within 96 schools The variables are ACH Measures of achievement The
357. this to see the condition used for selection in the tooltip that appears Click on SEL to invoke the Data Select Cases dialog box and edit the condition or turn off selection BY Displayed when one or more grouping By Groups variables are declared Pause the mouse on this to see the currently defined grouping variable s in the tooltip that appears Click on BY to invoke the Data By Groups dialog box and add delete grouping variables or turn off the By Groups declaration WGT Displayed when a weight variable is declared or exists in the data file Pause the mouse on this to see the currently defined weight variable in the tooltip that appears Click on WGT to invoke the Data Case Weighting By Weight dialog box and change the weight variable or turn off case weighting FRQ Displayed when a frequency variable is declared or exists in the data file Pause the mouse on this to see the currently defined frequency variable in the tooltip that appears Click on FRQ to invoke the Data Case Weighting By 211 Customization of the SYSTAT Environment Frequency dialog box and change the frequency variable or turn off frequency declaration ID Displayed when an ID variable is declared or exists in the data file Pause the mouse on this to see the currently defined ID variable in the tooltip that appears Click on ID to invoke the Data ID Variable dialog box and change the ID variable or turn off ID variable declaration CAT Displayed w
358. tion arrows have been activated The rotation arrows can be used interactively to rotate the plot in three dimensions 118 Chapter 4 allowing you to examine your data from all angles Try pressing each of the four rotation keys to examine how the plot changes Notable features include m True graphical rotation with automatic recalculation of the graph upon each rotation SYSTAT does not just rotate a picture or bitmap it physically transforms the graph data and replots the graph and all of its elements in real time with each rotation Realistic 3 D lighting to increase the volume effect Notable 3 D fonts on each axis that rotate along with the graph The ability to view from all angles including above and below Closer data points look larger and more distant points look smaller Smoothers SYSTAT offers 126 nonparametric smoothers for exploratory analysis In addition nineteen smoothers can be directly added to graphical output The smoothing options available for scatterplots are None LOWESS Inverse Andrews Linear DWLS Mean Bisquare Quadratic Spline Median Huber Log Step Mode Trimmed Power NEXPO Midrange Kriging Smoothers help you view your data in unique and informative ways In this case we are using kriging because it is especially designed for examining spatial distributions such as mineral deposits Tension of Smoothers Each smoother has a tension associated with it If you consider the smoother to be a
359. tions Contents Index Search Favortes Glossary LA Applications 8 Introducing SYSTAT Q Working with Data Click a letter below to scroll the glossary to terms beginning with that letter Click on a glossary term to Command Language view the corresponding definition gt Working with Output pra IE J 18 El 13 161 1 1 6 E rapnics E Statistics IN SOY EJ 8 R E a 10 4 a E 5 EZ gt Language Reference Using the Keyboard AIC and Schwarz s BIC Data File References Andrews Fourier Plot Acronyms ARCH Q Glossary Area under ROC curve Axial Design Bandwidth Bartlett test Bayesian Analysis Biweight kernel Bonferroni Test Bonferroni Correction for Probability Box Whiskers Plot Box Cox power transformation Bootstrap Box Behnken Design Bray Curtis Distance Bubble Plot Central Composite Design Centroid Design CHAID Cochran s Test of Linear Trend Cohen s Kappa Column Percents Communality Component Loadings Confirmatory Factor Analysis CFA Contingency Coefficient Contour Plot Coordinate Exchange Algorithm 43 Introducing SYSTAT Application Gallery In addition to examples of each procedure SYSTAT includes examples drawn from several fields of research Chapter 8 provides a brief introduction to each application You can access the complete applications from the Contents tab of the Help system Double click the Applications book icon and select Applicati
360. tituting for tokens 166 173 submit 143 clipboard 144 current line 144 from current line to end 144 from file list 226 selection 144 window 150 Submit Window from Log tab 150 SYC 154 syntax see commands SYO 194 SYSTAT data files 243 T t test two sample 78 Tab key 37 templates 161 automatic token substitution 158 177 custom prompts 168 dialog sequences 169 examples 173 175 176 177 179 180 181 filename substitution 161 173 IMMEDIATE option 170 integer substitution 167 175 176 177 interactive substitution 158 messages 160 multiple instances of a token 158 number substitution 167 175 176 opening files 161 ordering tokens 169 PROMPT option 168 prompting for input 158 resetting tokens 158 saving files 161 string substitution 166 173 177 variable substitution 163 164 173 179 viewing tokens 170 themes 232 applying 232 default 233 downloading 233 saving 232 TIFF 197 TOKEN 237 tokens see templates toolbars 218 creating 218 default buttons 217 deleting 218 hiding 218 renaming 219 supplied with SYSTAT 217 tree folder 191 Tukey pairwise mean comparisons 87 two sample t test 78 two way analysis of variance 89 181 Index U uniform distribution 177 unit of measurement 133 untitled tab 126 user interface Analyze 32 commandspace 21 data editor 24 Data menu 31 dynamic explorer 27 Edit menu 30 File menu 30 graph editor
361. tly using the initialization files in the INI sub folder of the SYSTAT program folder Edit the SycSamples ini file while maintaining the formatting of the content described below This initialization file expects the related command files to be in the SYSTAT Command folder So you can add nodes for your own command files provided they are saved in the Command folder Alternatively you can save your command files in any desired location create a new initialization file in the INI folder and enter the file path of the location suffixed by vour cmdfiles ini in the SysMaster ini file that is in the INI folder Use the following guidelines while creating the content of your cmdfiles ini m Type the top level folder caption without indentation m Use a hash at the end of a caption to define tree folders or nodes m Indent with the appropriate number of tab stops to create sub folders or nodes within a given folder m Ifacaption relates to a node type the filename including the file extension after the hash You can even include a sub folder name with the filename You can also skip the caption in which case the filename will be used as the node caption Viewspace Customization By default the Data Editor and the Graph Editor tabs are in the Viewspace However users may want to view the Data Editor and the Graph Editor simultaneously To do this click the Window menu or right click in the toolbar area and select Show Stacked or Sho
362. to view multiple graphs in the Graph Editor The latest graph or a graph that you double click on will be displayed in the Graph Editor for editing 20 Print Content of Data Variable Editor SYSTAT no longer supports printing the content of the Data Variable Editor To print data list the variables in the output and print the output To print variable information click Utilities gt File Information gt Dictionary and print the resultant output 9 Data What s New and Different in SYSTAT 13 New Features 21 22 23 24 Close Data Files You may now close data files using the context menu of the Data Editor or the CLOSE command Run CLOSE filename to close a particular file or CLOSE ALL to close all but the active data file Default Variable Format You may now set a distinct default numeric variable format for new numeric variables in the Data Editor This format is now independent of the numeric output format Save View Mode Data Files You may now save data files that are in the view mode Simply bring the desired view mode tab into focus and click the Save button on the Standard toolbar or click File gt Save Import Business Objects SYSTAT now offers the option of using a Business Objects Universe as a data source similar to the other choices such as ODBC Excel etc Business Objects is business intelligence platform organization which supports pre defined reports ad hoc reporting
363. tput Editor by clicking on the tab or selecting View Output Editor Using the Output editor you can reorganize output and insert formatted text to achieve any desired appearance In addition paragraphs or table cells can be left center or right aligned Tables Several procedures produce tabular output You can format text in selected cells to have a particular font color or style To further customize the appearance of the table borders shading and so on copy and paste the table into a word processing program Collapsible links Output from statistical procedures appears in the form of collapsible links You can collapse expand these links to hide view certain parts of the output Graphs Double clicking on a graph opens the Graph in the Graph tab When the Output editor contains more than one graph the Graph tab contains the last graph Note The Output editor supports opening and editing output files of SSYO format created in previous versions of SYSTAT Such output files however cannot be saved SYSTAT displays different formatting tools To change the formats of the outputs go to Edit Format and then apply different formatting tools Common formatting tools also appear on the toolbar in Customize in the View menu and in the toolbar in the Output editor Fonts SYSTAT displays output in an Arial font by default Select Font dialog box from Edit Format Font 187 Working with Output Font
364. ttons or toolbars would greatly diminish the area available for output and commands only six default toolbars with functionality designed to appeal to most users are set up to show in the user interface during the installation of SY STAT The default buttons on each of the five default toolbars are Menu Bar File Edit View Data Utilities Graph Analyze Advanced Quick Access Addons Window and Help Standard New Open Save Save All Cut Copy Paste Undo Redo Print Print Preview Full Screen Viewspace View Hide Workspace View Hide Commandspace Customize Recent Dialogs Submit from File List Start Stop Recording Play Recording and Help Format Bar Font Font Size Block Format Bold Italic Underline Font Color Outdent Indent Align Left Align Center Align Right Insert Image and Font Dialog Data Edit Bar Variable name Row number and Value of the variable at that row Graph Bar Chart Line Chart Pie Chart Histogram Box Plot Scatterplot SPLOM Function Plot and Map Statistics Column Statistics Two Way Tables Two Sample t Test ANOVA Estimate Model Design of Experiments Wizard Correlations Least Squares Regression Classical Discriminant Analysis and Nonlinear Estimate Model The Format Bar and two more toolbars namely Data Edit Bar and Graph Editing are embedded in the Output editor Data editor and Graph editor tabs respectively The Data and Graph Editing toolbars have the following buttons
365. tude and latitude at the center of the state according to the World Almanac and Book of Facts 1992 Pharo Books New York Salaries for U S governors USVOTES This data file breaks down the votes for CLINTON BUSH and PEROT by DIVISIONS VOLTAGE Montgomery and Peck 2002 The data set contains observations on the battery voltage drop VOLTAGE of a guided missile motor over the time of the missile flight TIME WATERQUALITY Databook 2005 The data file contains measurements of several physio chemical properties of water in five different cities The variables used are CHLORIDES and SULPHATES WESTWOOD Neter Kutner Nachtsheim and Wasserman 1996 A spare part is manufactured by the Westwood Company once a month The lot sizes manufactured vary from month to month because of differences in demand These data show the number of man hours of labor for each of 10 lot sizes manufactured The variables are PROD RUN LOT SIZE and MAN HRS WILL Williams 1986 RESPONSE is the dependent variable LDOSE is the logarithm of the dose stimulus and COUNT is the number of subjects with that response 358 Chapter 9 WILLIAMS Cochran and Cox 1957 These data are from a crossover design for an experiment studying the effect of three different feed schedules FEED on milk production by cows MILK The design of the study has the form of two 3 x 3 Latin squares PERIOD represents the period RESIDUAL indicates the treatment of
366. tude and longitude measurements of the center of the country Birth to death ratio in 1982 Log of gross domestic product per capita Years of life expectancy PAINTS Milliken and Johnson 1992 The dataset consists of four different paints Yellow 1 Yellow 2 White 1 and White2 that are manufactured by two different companies where the 1 and 2 refer to the company Each of the paint is applied on three different paving surfaces Asphaltl Asphaltl and Concrete The response is the life time measured in weeks In original data only the cell means and error sum of squares have been reported so the following data set has been generated artificially to have the same cell means and error sum of squares as the original data The variables are Y PAINT PAVES PAROLE Maltz 1984 These data record the number of Illinois parolees COUNT who failed conditions of their parole after a certain number of months MONTH An additional 149 parolees failed after 22 months but these are not used PATMISS Hocking 2003 In an experiment a pharmaceutical company was trying to test a new medicine Three clinics were selected at random from a large number of clinics The drug was administered to ten randomly selected patients However some of the measurements from some of the clinics have not been reported The variables are CLINIC and Y PATTERN Laner Morris and Oldfield 1957 In a psychological experiment of visual perception there were
367. u DASL Datafiles MercuryinBass html Genetics Data Sources Data Source Rao C R 1973 Linear Statistical Inference and its Applications 2nd edition New York John Wiley amp Sons McLachlan G J and Krishnan T 1997 The EM algorithm and extensions New Y ork John Wiley amp Sons Manufacturing Data Sources 314 Chapter 8 Original Source Messina W S 1987 Statistical quality control for manufacturing managers New York Wiley Data Reference Stenson H and Wilkinson L 1996 SYSTAT 6 0 for Windows Graphics SPSS pp 291 369 Medicine Data Sources Original Source Cameron E and Pauling L 1978 Supplemental ascorbate in the supportive treatment of cancer Reevaluation of prolongation of survival times in terminal human cancer Proc Natl Acad Sci U S A 75 4538 4542 Data Reference Andrews D F and Herzberg A M 1985 Data pp 203 207 Springer Verlag 315 Applications Medical Research Data Reference Wilkinson L and Engelman L 1996 SYSTAT 7 0 New Statistics pp 235 SPSS Inc Psychology Data Reference Wilkinson L Blank G and Gruber C 1996 Desktop data analysis with SYSTAT Upper Saddle River NJ Prentice Hall p 454 Stroufer S A Guttmann L Suchman E A Lazarsfeld P F Staf S A and Clausen J A 1950 Measurement and prediction Princeton N J Princeton University Press Sociology Data Reference Wilkinson L Blank G and Gruber
368. u and scan through the underlined letters You will quickly become familiar with the procedures and graphs you use frequently Customize Commands Toolbars Keyboard Menu Category Set accelerator for ovat E Curent keys Press new shortcut key 224 Chapter 7 Keyboard Shortcut Customization The default keyboard shortcuts may be changed and new keyboard shortcuts can be defined using the Keyboard tab of the Customize dialog Category Lists all the menus in the Menu Bar and one entry for all commands put together Commands Lists all the menu items under the menu selected in Category Select a command to see 1ts description in the Description area Current keys Displays the keyboard shortcut s already assigned either by SYSTAT or by you to the command selected in Commands If you do not want to use an existing keyboard shortcut key select 1t and press the Remove button to remove the assignment To reset keyboard shortcuts for all commands to their default assignments press Reset All Press new shortcut key Press the desired shortcut key or key combination for the selected command The key name will be automatically displayed in this area as you press 1t Key combinations will have to begin with Shift Ctrl Alt or any combination of these and end with one other key When you are satisfied with the key combination you have typed press Assign You can define more than one keyboard shortcut
369. ubmit any of them To create a new command file m From the menus choose File New Command Or click in the Commandspace and press the New toolbar button on the Standard toolbar Or double click on the empty space beside the last tab in the Commandspace Or right click on a batch Untitled tab and select New m Type SYSTAT commands in the batch Untitled tab For more information on SYSTAT commands see SYSTAT Language reference 139 Command Language x USE Ourvwor ld CSTAT POP_1983 URBAN HEALTH BABYMORT USE OURWORLD CSTAT pop 1983 urban health babymort m To save the command file click the corresponding tab and from the menus choose File Save Active File Or Save As Or right click on the corresponding tab and click Save m In the Save in field select the appropriate drive or folder to save to m Type a suitable filename or select an existing file from the list if you want to overwrite m The default format is unicode If you want to save the command file in ANSI format select SYC Files ANSI syc in the Files of type field Select All files 1f you want to use a different file extension m Press Save 140 Chapter 5 Save in O 2 DE 5 Datavolume et ExactTests My Recent Getting_Started Documents 2 GraphDeme C Graphics q Miscellaneous 9 Montecarlo 3 Quality Analysis E Statistics_1 C Statistics TI 5 Statistics III Mu Documents j C St
370. um token The first TOKEN statement assigns this token a value of rndnum As a result the left side of the LET statement becomes LET Urndnum After the equals sign we again find the amp dist token which has a value of U The final token on this line amp RN has an assigned value of RN resulting in the following valid transformation statement after token substitution LET Urndnum URN The template creates a new variable with a seven character name The first character of the name denotes the distribution used to generate the values and the final six indicate that the entries correspond to random numbers The output after randomly generating 100 observations from a uniform distribution follows 20 0 2 15 y O z O 5 5 0 15 O jo D S uy S N URNDNUM 179 Example 6 Multiple Variable Substitution Command Language The number of variables analyzed often varies across applications of a particular technique For instance one regression model may include two variables but another may include four We can create a template for each model as follows TOKEN ON TOKEN amp file TYPE OPEN PROMPT Choose a file to run regression TOKEN resp TYPE variable of the model TOKEN amp v1l TYPE variable TOKEN amp v2 TYPE variable TOKEN amp v3 TYPE variable TOKEN amp v4 TYPE variable USE amp file REM Two predictors REGRESS MODEL amp resp ESTIMATE REM Four predict
371. ustomization oaoa a a a a 212 Button Customization 2 20 0 0 eee eee 216 Toolbars e ts ERE oe ee Se ee 217 Positioning Toolbars ooa aaa eee 218 Toolbar Customization oaoa oa a a a a 218 Keyboard Shortcuts eor a e A a A eS 220 Keyboard Shortcut Customization 224 Menu Customization ooo a 2 eee eee 223 Command File Lists o 226 Submission From File Lists 228 Recent Dialogs E a Se a a es 229 SAS oy e ede py crs HO Ge vos taste ey a Hoe a a A 230 Vil Themes 2d a de Se Se ey ds e aa 2352 sO Dal OP MONS 5 a2 ete ee ee ea a wee AR 234 General Options 45 a a Dew eae oe eee eee 234 Output OPUONS oe 66 he ob ee Sa PS ee EARS BS 238 Output SCHEME geaen i SOR eae segue We Sei a G 240 Pile Locations pue Be Da ee BOSE EES 243 Using Commands 0 00 eee ee eee 244 Applications 247 Anthropology y oud 6 oe da eS 248 Egyptian Skulls Data oaaae 248 ASTON al ds Bs a es ee we a 251 SUNSDO CYCLES aoa bom amp Gk ie Be OR ee Bod Some E 251 Bolol yraa e a E e e e ea A ia Ea 252 Mortality Rates of Mediterranean Fruit Flies 252 Animal Predatory Danger 255 CASAS de Se wh cay x ee Re ise tds Soc Se 257 Enzyme Reaction Velocity 0004 pe ERSIMESTAS as Stored ur ee Oe ake A he ee eee 262 Robust Design Design of Experiments 262 Environmental Science
372. utliers on the lower left of the graph by holding down the left mouse button and circling them m Click the Show Selection icon to highlight the selected cases 110 Chapter 4 Dynamically Highlighted Cases Cases selected by the Lasso tool are highlighted in the Data editor Click on the Data Editor to see these cases 30 and 31 directly SYSTAT mantis SYSTAT SYSTAT13 SYSTAT_13 Data Gdwtrdm syz Utilities Graph Analyze Advanced Quick Access Addons Window Help Jex Ele Edt View Data NA alo a BA acerca EE nx 4 SYSTAT Output HE USE a HE USE imantis Sy a ee ee ls DENSITY UR HORIZON HORIZON URANIUM ARSENIC BORON BARIUM MOLY SELENIUM VANADIUM SU Me DENSITYURI 49 2 7241 000 33 308 101 757 TPO 1 000 20960 9 700 2 000 000 60 000 20000 0100 150 000 4 G E USE GOWTROM 20 2243 000 33 314 101 809 TPO 1 000 26670 6400 990000 99000 7 000 0 100 66 000 1 2 eae 21 23245000 33 314 101 926 TPO 1 000 52 470 9 700 2 000 00 50 000 76 000 0400 100 000 2 e HE USE limantisisv 22 2 248 000 33 311 101 869 TPO 1 000 6490 63 000 1 500 000 150 000 200 000 2200 600 000 11 le DENSITYUR 23 2 250 000 33 312 101 978 TPO 1 000 16780 16500 1 500 000 75 000 20000 7900 200000 1 BE USEGDWTRDN 24 2253000 33 271 101 805 TPO 1 000 21 190 10 700 2 000 000 100 000 200 000 0 700 200 000 4 reel 25 2255 000 33167 101 450 TPO 1 000 13160 18200 559 000 33 000 9000 0500
373. ve the resulting file as ASCII text We recommend using the SYC extension when saving these files Although any text file containing commands can be processed using an SYC extension for these files allows maximal Windows functionality such as double clicking a file to automatically open 1t In addition you can use a text editor in conjunction with the Windows Clipboard to submit syntax for processing without creating command files or using the Commandspace After typing the commands in your editor select and copy them In the processing environment select Submit Clipboard from the File menu or the context menu of the Interactive batch Untitled tabs of the Commandspace The software processes the commands without changing any text in the Interactive or batch Untitled tabs of the Commandspace Using a text editor for command entry allows you to hide the Commandspace creating more area in which to display the output To hide unhide collapse or resize the Commandspace see Commandspace Customization cross refrerence in Chapter 7 Customization of the SYSTAT Environment As you change between the editing and processing environments the currently active application appears in front of the other Consequently you can maximize the area for both the input and the output switching between the two by toggling between the applications You can also have multiple command files open submitting commands from each of them using the Copy Submit Clipboard
374. ve to this hypothesis could be H Mean of Diet is greater than mean of regular or H Mean of Diet is not equal to mean of regular or H Mean of Diet is less than mean of regular Since we have no information let us choose the second alternative H Mean of diet is not equal to mean of regular In other words do diet and regular dinners differ in protein and calcium content In this example we use the t test procedure m From the menus choose Analyze Hypothesis Testing Mean Two Sample t Test m Inthe Two Sample t Test dialog box select PROTEIN and CALCIUM as the variables and select DIET as the grouping variable In the Alternative type choose not equal Click OK 79 SYSTAT Basics Hypothesis Testing Mean Two Sample t Test Man Available variable s Data layout Resampling BRAND Indexed data Data in columns FOODS Selected variables CALORIES PROTEIN de Add CALCIUM PROTEIN VITAMINA lt Remove CALCIUM IRON COST DIETS Required gt Grouping variable emale pe hot equa Ww Pibe Confidence Dunn Sidak HO Meanl Mean2 vs H1 Meanl lt gt Mean2 Grouping Variable Variable PROTEIN l I CALCIUM DIETS Standard N Mean Deviation 15 000 22 133 4 307 13 000 16 846 4 337 15 000 11 800 12 757 13 000 9 769 8 506 80 Chapter 3 Separate Variance i 95 00 Confidence Interval Variable
375. w Side by Side All the panes in the Viewspace get laid out in a tiled fashion Click the Minimize 2 or Close 3 Gf it is enabled button of the panes that you do not want to see and select Show Stacked or Show Side by Side again The pane that is active will be placed first in the tiled layout Using the Window menu or context menu of the toolbar area you can also Cascade windows or Arrange Icons that have been minimized Double click one of the title bars to dock the panes to their default or previously docked positions Maximizing the Viewspace Almost every command and dialog box creates output all of which appears in the Output Editor of the Viewspace Occasionally statistical output or graphs may be too large to be viewed in the Output Editor Even data files will typically contain more 209 Customization of the SYSTAT Environment number of rows than visible in one view Although scrollbars allow control over the contents of the viewable area displaying graphs or results in their entirety in a single pane simplifies interpretation The most obvious method for increasing the size of the Output Editor involves maximizing the user interface to fit the size of your monitor You can close toolbars that you do not use frequently You can resize or undock the Commandspace or Workspace to increase the viewable output region You can also work with the Viewspace in the full screen mode To set the Viewspace to the full screen mode Fro
376. ww chicken 190 0 12 10 4 2 49 yes st beef 390 24 20 2 4 15 2 99 no st beef 370 19 24 2 20 15 2 99 no st chicken 320 10 21 10 15 8 2 69 no st chicken 330 16 18 2 2 4 2 99 no gor beef 290 8 18 15 4 10 1 75 no gor pasta 370 16 20 30 40 4 1 99 no gor pasta 440 26 20 100 35 10 1 75 no gor beef 300 34 22 15 10 20 1 75 no ty beef 330 14 24 8 10 10 3 00 no ty chicken 400 8 27 25 0 10 3 50 no ty chicken 340 7 31 70 0 15 3 50 no ty chicken 430 24 20 45 4 6 3 00 no SW chicken 550 25 22 0 6 15 2 25 no SW beef 330 9 25 10 2 25 2 85 no SW pasta 300 1 14 0 25 10 1 60 no The first line contains names for the columns SYSTAT will count these names finding 10 and read 10 values for each case dinner We name this ASCII file FOOD DAT Let us read the FOOD DAT file and convert it to a SYSTAT file called FOOD SYZ m From the menus choose File Open Data In the Open dialog box select All Files from the drop down list of file types select FOOD DAT and click Open The contents of the data file are displayed in the Data editor E From the menus choose File Save As 53 SYSTAT Basics m Type FOOD for the filename in the Save dialog box and click OK The subsequent sections will show you how to create charts and run statistical analysis using SYSTAT menus and dialog boxes Graphics Scatterplots Scatterplots provide a visual impression of the relation between two quantitative variables Let us plot CALORIES versus FAT for this l
377. y default Command buffer The command buffer contains the most recently processed commands Use this buffer for quick recall modification and resubmission of commands using the F9 key The number of commands to keep defines the size of the buffer use the up and down arrows to adjust the number of retrievable command lines The software uses the buffer to store commands generated from any of the following sources Command prompt Commands submitted using the Interactive tab of the Commandspace m Files Commandspace and clipboard Commands submitted from the middle and Log tabs of the Commandspace This option also includes commands submitted directly from the Windows Clipboard and command files submitted via the SUBMIT command Dialogs Commands generated after clicking the OK button in any dialog Select this option to use the dialog interface to generate a command line that you expect to refine iteratively 231 Customization of the SYSTAT Environment Autocomplete commands As you type commands in any tab of the Commandspace you will be prompted with the possible command keywords arguments options option values available data files or available variables For instance the data files in the folder specified under Open data in the Global Options dialog will be listed 1f you type USE This feature is enabled by default You can turn it off if you do not want commands to be autocompleted Color command keywords By def
378. y whether you want the Startpage to show at startup clear recent data command and output files that are listed in the Recent Files quadrant refresh the content of the Startpage close 1t for the rest of the session and invoke the Edit Options dialog box Output editor You can cut or copy the selected content in the Output editor to the Windows clipboard paste content from the clipboard to the Output editor copy all the content in the Output editor to the clipboard view the HTML source refresh or preview the content for printing collapse expand links in the output show the Format Bar create a new output file clear all or save the content in the Output editor and invoke the Edit Options dialog box Data Variable editor You can copy all the content in the Data editor set one of the inactive data files in the Data editor as the active data file switch between the Data and Variable editors enter or view and edit comments for a data file show the Data Edit bar and Data toolbars create a new data file save data files invoke the Edit Options dialog box close a data file and show the processing conditions in effect if the Variable editor is active 34 Chapter 2 Graph Editor You can invoke the Graph Properties dialog box animate a 3 D graph realign any graph frames you may have moved from their original positions copy or preview for printing the graph in the Graph editor show the Graph Editing toolbar save the grap
379. yed a multiethnic sample of 256 community members for an epidemiological study of depression and help seeking behavior among adults Afifi and Clark 2004 The CESD depression index was used to measure depression The index is constructed by asking people to respond to 20 items I felt I could not shake off the blues My sleep was restless and so on For each item respondents answered less than time per day score 0 1 to 2 days per week score 1 3 to 4 days per week score 2 or 5 to 7 days score 3 Responses to the 20 items were summed to form a TOTAL score Persons with a CESD TOTAL greater than or equal to 16 are classified as depressed Variables include ID Subject identification number SEX 1 male 2 female AGE Age in years at last birthday MARITAL never married 2 married 3 divorced 4 separated 5 widowed 1 less than high school 2 some high school 3 finished high school EDUCATN 4 some college 5 finished bachelor s degree 6 finished master s degree 7 finished doctorate 1 full time 2 part time 3 unemployed 4 retired 5 houseperson 6 in school 7 other INCOME Thousands of dollars per year SORT INC Square root of income EMPLOY 353 Data Files RELIGION 1 Protestant 2 Catholic 3 Jewish 4 none 6 other BLUE to DISLIKE Depression items TOTAL Total CESD score CASECONT 0 normal 1 depressed CESD gt 16 DRIN

Getting Started Manual - Applied Biostatistics II

Contents

Download Pdf Manuals

Related Search

Related Contents