Home

SIMSTAT - Provalis Research

1. 00 000000 116 206 MOSESHESE a adetinin ure An ups APS 117 207 NPAR matrix 2 6 0 0 0 00 127 210 One sample chi square test 129 186 Runs test 2 0 0 eee 140 222 SIME ponerse Lek eve oe ERE Va vets 143 226 Wilcoxon test 0 0 cee ee 152 231 Normal probability plot 49 87 120 134 dire Ne X eser apre 195 197 208 220 Notebook 0 000 000 eae 4 60 64 225 NPAR matrix 0 0 0 0 eee 127 210 185 Numeric recoding 0 000 c ee eee eee 59 o One sample chi square test 129 186 Oneway ANOVA 110 130 210 251 Opening a script file sleesl esses 169 Output management using Tabs 63 P PACE plot 2 2 2 cece pene ener EY AAA 146 228 Pak file 25 neo route vies nie 31 Pairwise deletion 75 127 190 201 210 Paradox files 00 000 c ee eee 38 40 199 Pareto chatter orire ee eee ee ors wee eS 86 87 195 Partial autocorrelation 146 228 Password protection 0 0 0 0 0 0 e ee ee eee 43 Pearson St betee bane be bet pad doasdaaws 75 78 191 Percentile table 69 86 182 184 195 217 PID S5 hof orden Loro A Sree arate tn Pa 78 190 191 Pointinarker 5 s ree reU 162 Power estimate llle 69 72 182 Principal components analysis 212 Principal coordinates analysis Lun 212 Printing el n hey eset PIE 19
2. Apply 8 Ignore x Cancel Help To restore previously used filtering conditions click on the down arrow button located to the right of the Filter edit box 34 SIMSTAT for WINDOWS Once a filtering condition has been entered you can apply the filter and leave this dialog box by clicking on the APPLY button If the filter expression is invalid a message is displayed and the exit is not performed To temporarily deactivate the current filter expression click on the IGNORE button The filter string will be kept in memory and may be reactivated by choosing the FILTER RECORDS command again To exit from the dialog box and restore the previous active filtering condition click on the CANCEL button Building a valid filtering expression The upper part of the Filter dialog box contains various elements to help you build a valid filtering expression Variable name list box Double clicking a variable name from the list box located to the left of the dialog box inserts that name in the edit box at the current caret position Function list box A list of valid xBase expression is displayed to the right of the dialog box Double clicking on a xBase function in the list box inserts that function at the current caret position When a function requires one or several arguments the argument section remains highlighted To replace the highlighted text with a value an expression or a variable name simply type the proper text on t
3. ETO BOXPLOT CUMUL PPLOT HISTOGRAM NBAR integer NORMAL AXNOMINAL PANEL FRIEDMAN Syntax FRIEDMAN varlist Description Display frequency table Sort table by Ascending frequency Descending frequency Ascending value Descending value Descriptive statistics Confidence interval width Percentile table number of categories Bar chart Pie chart Pareto chart Box amp whisker plot Cumulative frequency plot Normal probability plot Histogram Number of bars intervals Normal curve aximum number of values for table and nominal level charts Display the dialog box The FRIEDMAN test is a procedure for testing whether two or more related samples have been drawn from the same population The output displays the mean rank of each variable the number of cases chi square degree of freedom and probability value 196 SIMSTAT for WINDOWS FULL Syntax FULL options command options Description The FULL command allows you to perform various statistical analyses such as frequency analysis crosstabulation or multiple regression on successive bootstrap samples The first set of options allows you set the sample size and the number of samples to be drawn from the original sample A second set allows you to specify which analysis to perform and set the options normally available when calling the chosen statistical procedure This second set of options allows to control how the analysis is to be p
4. 0 0 0 eee ee eee 4 Status bar ea e e E EAA eee 11 225 226 235 Stepwise regression 118 119 121 208 Symphony files 000 38 40 199 Syntax Convention eee 173 System requirements 000000000 08 3 T T test dut comm E one dE DO RU TETUER 149 227 Tab delimited files 38 41 165 199 yh MANC A MAREC COINS 39 60 63 Tad lk ee 69 78 127 183 191 210 218 Technical support llle 2 3 253 Time series analysis sellers 147 228 TOO DAE 252253 dua fo cdot stro rok iof 9 11 225 235 Tools menu Pe iee E ERNA EAN ANAR eee 233 Transformation 05 50 57 59 188 Rank transformation 0 58 219 TUKEY 255 aes arde ah sae he e he de bey 130 211 Tutorial oss celLeeseseteseiemesiemseiena 13 19 V Valuelabels 22 271 30 38 41 45 230 Variable assignment 0000 eee 18 65 Variable name 000 00 eee ee 24 27 28 Variable Sets merre p wes a an a yes es 45 46 Variable statistics 0 0 eee 49 Varimax rotation 0000005 81 85 194 Ww WAY files zoe Des e aid Mie kes 180 213 Milcoxomdest co otros ee ve M 152 231 Windows bitmap 0005 165 214 Windows metafile 00000 165 214 WMF files 0 cee ee 165 214 x xBase syntax and functions 240 248 Z Zooming in and out 2 ee eee 163
5. es Turn on off the 3D option for the current chart A Display the 3D rotation dialog box 4 Zoom in the currently displayed graph Script window buttons b Run the entire script Eb Run the selected commands Tp Run the script from the current line GETTING STARTED 11 At the right end of the toolbar you will see four buttons representing SIMSTAT s four different windows Script Chart Data Notebook Click on one of these icons switch to another SIMSTAT window Status bar The status bar at the bottom of the screen displays information about the current session including the number of selected variables and the amount of free Windows resources and memory It also displays useful information about the current output page date and time of creation data file name or about the current chart data and time of creation The left most panel contains a gauge that is sometime used to indicate the progression of a task SAMPLE DBF 07 02 96 09 08 40 Aca w Record 53 15848K Progress bar Filename date amp time Selected variables Record status L Available memory Getting help No matter where you are in SIMSTAT you can get more information about the task you re working on by pressing F1 This function key accesses SIMSTAT s context sensitive help You may also click on the HELP button from any dialog box to obtain help on the various options available in the dialog box Alterna
6. 0 2 eee eee 81 193 Import data 0 eee eee 38 199 Independent t test 20 000 149 227 Installation 2 22 me cement casa das d oacocgalees 3 Interaction GLM ANOVA ANCOVA 90 91 196 197 Logistic regression 00 112 204 Internal consistency 100 136 137 221 251 Inter item correlation 5 137 221 Inter raters agreement 97 200 250 Inverse regression 133 220 Ttem analysis e pes acn p a a 0 eee eee 104 201 251 Item characteristic curves 0 104 201 K Kendall s tau 69 78 127 183 210 218 Kolmogorov Smirnov test 108 109 202 203 Krippendorff agreement measures 97 200 250 Kruskal Wallis command 110 202 Kurtosis 49 69 74 86 100 125 182 193 L Labels Axis labels cti RE 160 162 Value labels 22 27 30 38 41 45 230 Legend sete checa ee RE Ae d we aaa Se 160 162 Likelihood ratio 78 112 191 204 Limitations 0 000000 0 ce eee 38 40 253 Linear regression 000000005 19 220 Listing Cases 7 014 Let ce doen a se 111 204 Listwise deletion 75 104 127 190 201 210 Logistic regression eese 112 204 Lotus fes 22 3 52i teure suce Ex 38 40 199 M Mann Whitney test 114 152 205 Manual conventions 0000 eee eee 2 McNemartest llle 115 205
7. 05 49 72 87 Data files Backup of data files 0 0 0 45 Creating a new datafile 0 24 Defining variables 0 0000 eee 27 Entering and Editing Data 31 Exporting data to other applications 40 Filtering records 0 0 0 0 0 00000 33 Importing data from other applications 38 Limiting access to data files 43 Merging data files 0 2 0 eee ee 42 Opening an existing data file 21 Sorting records 0 0 00 ce eee eee 36 Data labels in a chart 200005 160 dBase files DBF 22 41 191 211 Decimal places Variables 25 27 28 50 226 238 Output e oe xt e es cient OR PS se 64 Charts xs ceseuhbestbere ie ietistis 159 186 Defining variables 00000005 27 230 Delete record 0 0 0 0 0 eee 31 238 Descriptive analysis 72 80 86 130 193 Deviation chart 0 00000 e eee 131 211 Directories 236 Discrimination index 0 100 201 Displaying data point values 160 Diversity indices 0 000000 eee eee 193 Dual histogram 0 00000 150 227 Dummy recoding 0 000002 eee 58 Durbin Watson GLM ANOVA ANCOVA 92 197 Multiple regression 120 208 Regression analysis 00 134 221 E Editing axis scaling and
8. ALIAS Returns the Alias name of the current work area as a string ALLTRIM String Trims both leading and trailing spaces from a string The string may be derived from any valid xBase expression ALLTRIM Provalis returns Provalis AT SearchString TargetString Determine whether a search string is contained within a target If found the function returns the position of the search string within the target string relative to 1 If not found the function returns 0 zero AT gh defghij returns 4 CHR Val Converts a decimal value to its ASCII equivalent CHR 83 returns S CTOD String Converts a character string into an xBase date The string must be formatted according to the Windows date format settings CTOD 12 31 94 DATE Returns the system date today Use DTOC DATE to retrieve today s date formatted according to the Windows settings 244 SIMSTAT for WINDOWS DAY DateField Returns the day portion of an xBase date as an integer DELETED Returns True if the record is deleted and False if not deleted DESCEND String An xBase function that inverts a key value using 2 s complement arithmetic The result of the operation is the arithmetic inverse of the key value When inverted keys are sorted in ascending sequence the result is in descending order A filter expression could be DESCEND DTOS billdate CUSTNO DTOC DateField Converts an xBase dat
9. Effect coding Effect coding is similar to dummy coding except that the last group is assigned 1s in all vectors instead of Os When you click on the OK button the program reads all active records and asks you to confirm the computation of the required number of dichotomous variables If variables with similar names exist in the database the confirmation dialog box will display the number of existing variables that would be overwritten as well as the number of variables that need to be created by this command Click on the YES button to confirm the creation and overwriting of variables Click on NO to abort the procedure To exit from the dialog box without performing the recoding click on the CANCEL button Numeric recoding Most statistical analyses require variables that can be expressed as numeric values such as numeric date and logical data type The NUMERIC CODING command automatically creates a numeric variable to express the values of a string variable To create this numeric variable position the cursor on the string variable and select the TRANSFORM VARIABLE NUMERIC RECODING command sequence from the DATA menu When this command is evoked a dialog box appears asking you the name of the new numeric variable If you specify a non existing variable the program will ask you to confirm the creation of this new numeric variable If you click on the YES button this variable will be created and will contain integer values repre
10. LEFT String Length Returns the leftmost characters of the expression for the defined length LEFT xyzabc 3 returns XyZ LEN Expression Returns the length of the expression result as an integer LOWER String Converts the string expression into lower case MONTH DateField Returns the month portion of an xBase date as an integer ORDER Returns the current index order as an integer ORDKEY Returns the current index key as a string Same as INDEXKEY PADC String Length Character Centers the passed string between a number of the passed character to make the string the specified length PADC Scott 9 returns Scott 246 SIMSTAT for WINDOWS PADL String Length Character Pads the passed string to the specified length with the specified characters If the string is longer than the value specified by Length the string is truncated to this length PADL Scott 8 returns Scott PADL Loren Scott 8 returns Loren Sc PADR String Length Character Pads the passed string to the specified length using the specified character If the string is longer than the value specified by Length the string is truncated to this length PADR Scott 8 J returns Scott PADR Loren Scott 8 returns Loren Sc RAT SearchString TargetString Determine whether a searc
11. Options Type of cuttoff point MEAN Mean MEDIAN Median VALUE Value VALUES integer integer Values of X PANEL Display the dialog box SCRIPT LANGUAGE REFERENCE 223 SAVE Syntax SAVE windowtype Description This command can be used to store the contents of the notebook the chart or the script window on disk If the contents were created during the current session and have not been saved on disk before a dialog box will appear to allow you to specify the name and location of the new file Options Window types NOTEBOOK Notebook Statistical results window SCRIPT Script log window CHART Charts window Examples SAVE NOTEBOOK SAVE CHART SCED Syntax SCED varlist BY varlist options Description The SCED single case experimental design command provides some basic tools to study the effect of an intervention on the behavior of a single subject It involves the repeated objective measurement of the behavior of a single subject dependent variable over a long period of time interspersed with changes in the treatment condition independent variable The procedure will display a graph representing the evolution of the dependent variable Y at various phases defined by the independent X variable The various options allow you to obtain statistics for each phase of the analysis as well as graphic tools that can be used as judgemental aids to identify the experimental effect of
12. SEX 1 You may also use the filter building buttons located at the top of this dialog box to specify this condition To build this filtering condition using these buttons Click on the SEX variable in the list box located to the left of the dialog box e Click on the relational operator button to add an equal sign to the equation Click on the 1 button of the numeric keypad to insert this value after the equal sign To exit this dialog box and activate the filtering condition click on the APPLY button 18 SIMSTAT for WINDOWS Step 6 Performing regression analyses Choose the REGRESSION REGRESSION command from the STATISTICS menu The following dialog box appears Regression options x Analysis v o Type of analysis Linear z X Cancel Confidence interval joo Choose Significant test C One tail Two tails 3 Help Graphic Scatterplot Text M Residual analysis Residual caseplot Outliers 25 Durbin Watson test Residual scatterplot Residual normal plot Save residuals The first step necessary to perform our regression analysis is to change the currently selected variables In the previous analysis the variables were selected prior to the display of the analysis dialog box For this example we will alter the variable assignment while editing the regression analysis option To achieve this click on the CHOOSE button to activate the Variables
13. Smoothing This option allows the application of two methods of identifying trends in noisy or irregular time series The moving average procedure is obtained by averaging a selected number of points on either side of a target value while the running median procedure is computed by finding the median value of a specified number of points on either side of the original value Successive smoothing can be achieved by specifying more than one value For example the following option Moving average 424 instructs SIMSTAT to proceed with three successive moving average smoothings using 4 2 then 4 values to compute the mean Vertical line separators This option lets you specify whether vertical lines should be used in the time series chart to delineate the various phases of the experiment Control bars This option allows you to superimpose 3 horizontal bars that represent the mean and the upper and lower limits of a confidence interval Those bars can be used as a judgemental aid to identify a change in the series The interval option allows you to specify the width of the confidence interval Its value must be between 0 and 99 The minimum and maximum options allow you to specify which observation in the series will be used to calculate the mean and the confidence limits If those fields are left blank the first and the last observations will be used For example entering 1 and 10 as the minimum and maximum values tells SIMSTAT to compute the
14. Suspends the program execution for 5 seconds SCRIPT LANGUAGE REFERENCE 193 DESC Syntax DESC varlist Description The DESC command displays the mean standard deviation minimum and maximum values of each specified variable To obtain other descriptive statistics such as the variance skewness kurtosis mode median etc see the FREQUENCY command DIVERSITY Syntax DIVERSITY varlist options Description This procedure computes three diversity indices commonly used in ecology Simpson s Shannon s and Brillouin s Requires MVSP v2 2 Options TRANSPOSE Transpose data Select a transformation default LOG10 LOG10 Log base 10 LOGE Log base e LOG2 Log base 2 Select a coefficient default SIMPSON SIMPSON Simpson s diversity index SHANNON Shannon s diversity index BRILLOUIN Brillouin s diversity index FACTOR Syntax FACTOR varlist options Description The FACTOR command performs a principal component or an image covariance analysis with or without axis rotation Requires EASY FACTOR ANALYSIS v3 0 194 SIMSTAT for WINDOWS Options PCA IMAGE ER integer N real AX EL ACTION TATION WEIGHT F SCORE B I C R R FILTER string Description The FILTER command allows you to temporarily select cases according to some logical condition You can use this command to restrict your analysis to a subsample of cases or to t
15. NEW command from the FILE menu When this command is evoked a dialog box similar to this one appears reate new data file x AME Enter a unique variable name Nore Tiree Se pee pesn 5 5 5 5 7 CASENO Case number INTERVIEW Date of the interview SEX Sex of the respondent AGE Age of the respondent v OK X Cancel 7 Help The first step you need to take in order to create a new data file is to define the structure of the file The File Structure dialog box is a grid entry field type where each row represents a variable in the new data file This dialog box lets you define various attributes of the new variables such as their name whether they will contain numeric alphanumeric values or dates and their physical width You can also enter for each variable an alphanumeric description up to 60 characters Name The first column of the spreadsheet allows you to enter a name for each variable Each variable name must be unique within that data file Valid variable names begin with a letter and may contain letters numbers or underscore characters Punctuation marks blank spaces and other special characters are not permitted The maximum variable name length is 10 characters Type Each variable in the data file must have a type To specify a variable type move the cursor to the second column and enter the letter corresponding to the proper data type SIMSTAT for Windows supports the following types Key Type C Ch
16. SIMSTAT Bootstrap computer simulation and statistical program for IBM personal computers Behavior Research Methods Instruments amp Computers 25 3 410 413 STINE R 1989 An introduction to bootstrap methods Examples and ideas Sociological Methods and Research 8 2 amp 3 243 290 WASSERMAN S amp BOCKENHOLT U 1989 Bootstrapping Applications to psychophysiology Psychophysiology 26 2 208 221 254 SIMSTAT for WINDOWS APPENDIX C LIMITATIONS Maximum number of variables 1022 Maximum number of cases 320 000 or more Maximum number of bootstrap resampling 32 000 Maximum number of notebook pages 16 320 Maximum number of charts 16 320 Maximum number of sections in a notebook 100 Maximum number of nominal values Frequency Crosstab 255 Breakdown Oneway Kruskall Wallis Maximum number of factors covariates GLM ANOVA ANCOVA 5 Maximum number of predictors in multiple regression 40 Maximum number of variables in reliability analysis 90 Maximum number of variables in item analysis 250 TECHNICAL SUPPORT 255 APPENDIX D TECHNICAL SUPPORT When contacting Provalis Research for technical support by fax or e mail be sure to include your product name and version Include a description of the problem including all the steps needed to replicate it and if applicable a complete program source code and data in the smallest form possible You can contact Provalis Research Phone 514 899 1672 collect call
17. gt Greater than lt Less than or equal to gt Greater than or equal to Is contained in 242 SIMSTAT for WINDOWS Logical Operators Notice the periods surrounding the operator AND both expressions are true OR either expression is true NOT either expression is false Evaluation Order When more than one type of operator appears in an xBase expression the order of evaluation is as follows Expressions containing more than one operator are evaluated from left to right Parentheses are used to change the evaluation order If parentheses are nested the innermost set is evaluated first Numeric operators are evaluated according to generally accepted arithmetic principles operators contained in parentheses exponentiation multiplication and division addition and subtraction Order of evaluation may be altered with parentheses 3 p4 516 29 3 4 5 6 41 3 4 5 6 77 Logical operators are evaluated as NOT first AND second and OR last Logical evaluation order may also be altered with parentheses In multiple conditional expressions that contain the NOT operator always use parentheses to enclose the NOT operator with the expression to which it applies APPENDIX A XBASE SYNTAX AND FUNCTIONS 243 Supported Xbase functions The following xBase functions are supported in the SORT RECORDS and FILTER RECORDS command NOTE Memo field names are not allowed in SIMSTAT xBase expressions
18. off point used to classify the subjects One problem is that increasing the cut off point in order to reduce the number of false negatives will usually generate an increase in false positives The sensitivity analysis allows one to assess the ability of a quantitative measure X to differentiate a dichotomous criterion condition Y and provide guidelines to choose a cut off point that will offer an appropriate trade off between false positive and false negative error rates The program provides the level of sensitivity 1 e the number of true cases detected by the test divided by the total number of true cases in the sample and the specificity i e the number of true negatives divided by the total number of cases without the problem or disease for each value of the quantitative measure The program also displays for each value of the test the percentage of false positives and false negatives If we plot on a cartesian scale and for each cutt off point the sensitivity at that point as a function of the proportion of false positives or 1 specificity and connect those points we obtain what is known as a ROC receiver operating characteristic curve This graph allows us to visualize the performance of the screening or diagnostic test used If a test has no discriminatory value the ROC curve will be a diagonal line with a 45 degree angle going from the lower left corner to the upper right corner of the graph The ROC curve of a perfect test will
19. 2 00 n morel EXACT Corrected for ties U W 1 tailed P Z l tailed P KORRO 1174 0 0000 c4 59125 0000 McNemar test STATISTICAL ANALYSIS 115 The McNEMAR TEST is a procedure applied to a pair of correlated dichotomous variables to test whether there is a significant difference in proportions of subjects that change from one category to another at two different points in time or in response to two different conditions A binomial test is used to compute the significance level when the total number of changes is less than 10 Otherwise a chi square statistic with the Yates correction for continuity is used OPTIONS Values This option allows you to specify the two values for both the independent and dependent variables For each variable the cases that match the first value are assigned to one condition and the cases that match the second value are assigned to the second condition 2 x 2 contingency table is then constructed and a significance test is computed for cases that are in different conditions Direction This option specifies whether the probability of the significance test is based on a one or two tailed test Sample output of McNemar test McNEMAR TEST VAR2 With VAR1 VAR2 VARI Chi 230 65 2 tailed probability 8026 116 SIMSTAT for WINDOWS Median test The MEDIAN TEST is a procedure for testing whether two or more independent groups differ in central
20. S 1956 Nonparametric statistics for the behavioral sciences New York McGraw Hill INTER RATER AGREEMENT MEASURE BRENNAN R L amp PREDIGER D J 1981 Coefficient kappa Some uses misuses and alternatives Educational and Psychological Measurement 41 687 699 COHEN J 1960 A coefficient of agreement for nominal scales Educational and Psychological Measurement 20 30 46 KRIPPENDORFF K 1970 Bivariate agreement coefficients for reliability of data In E F Borgatta and G W Bohrnstedt Eds Sociological methodology 1970 San Francisco Jossey Bass SCOTT W A 1955 Reliability of content analysis The case of nominal scale coding Public Opinion Quarterly 19 321 325 MULTIPLE COMPARISON PROCEDURES JACCARD J BECKER M A amp WOOD G 1984 Pairwise multiple comparison procedures A review Psychological Bulletin 96 589 596 TOOTHAKER L E 1993 Multiple comparison procedures Beverly Hill Sage Publication MULTIPLE REGRESSION DARLINGTON R B 1968 Multiple regression in psychological research and practice Psychological Bulletin 69 161 182 COHEN J amp COHEN P 1983 Applied Multiple Regression Correlation analysis for the Behavioral Sciences 2nd Ed Hillsdale N J Lawrence Earlbaum DRAPER N R amp SMITH H 1981 Applied Regression Analysis 2nd Edition New York John Wiley amp Sons PEDHAZUR E J 1982 Multiple Regression in Behavioral Research 2nd Editi
21. This chapter provides a description of every statistical analysis appearing under the STATISTICS menu To allow you to quickly find the information you need the statistical commands are presented in alphabetical order Working with charts This section describes of the steps involved in the creation modification printing and saving of charts Using scripts This section introduces you to various tasks that may be performed with script files such as how to open a existing script file how to execute it or how to use the recoding feature to easily create script Script language reference This section outlines the syntax conventions of the various commands and options and provides an alphabetical listing of all script commands Customizing the TOOLS menu This section provides instructions on how to add programs to delete programs from or edit programs using the TOOLS menu Setting the program preferences This section provides a description of all global options affecting the program s working environment file handling and printing Appendices The last part of this manual contains 4 appendices that deal with the following topics Appendix A Description of xBase syntax rules used in the FILTER SORTING and COMPUTE commands and of each xBase function Appendix B Various suggestions for further reading on statistics Appendix C Program limitations Appendix D Obtaining technical support Manual conventions The fo
22. numeric or alphanumeric and a value To declare a memory variable you must provide its name and data type The explicit declaration of a variable is done with a DIM statement For example to declare an alphanumeric variable called NAME and a numeric variable called SAGE you must enter the following commands DIM SNAME AS STRING DIM SAGE AS NUMERIC SCRIPT LANGUAGE REFERENCE 175 You may also declare a new memory variable in any other command by explicitly stating its data type on its first appearance For example DIM SAGE AS NUMERIC LET SAGE 5 can also be expressed as LET SAGE AS NUMERIC 12 SIMSTAT gives each variable an initial value at the time it is declared A string variable is initialized to the empty string a string with no characters A numeric variable is initialized to zero You can also access any variable or field of the currently opened data file by putting a DB prefix to the variable s name For example to modify the value of a variable named AGE in the current data file you can use a LET command in the following way ET DB AGE 12 You can also read the value stored in this variable just like you would do with any other memory variable IF DB AGE lt 18 THEN LET CATEGORY 1 All reading and writing operations on data file variables are performed on the currently selected record In order to access the various rec
23. the percentage of the count over all cases and over valid cases only and the cumulative percentage over all valid cases Sort by Use this option to tell the program whether the frequency table should be sorted by the values of the variable or by order of frequency Type The Type option allows you to specify whether the frequency table should be sorted in ascending or descending order Descriptive statistics Use this option to obtain a table of various descriptive statistics This table includes the following statistics mean standard error of the mean sum mode median standard deviation variance skewness kurtosis minimum maximum and range Confidence interval This option prints a confidence interval around the mean The Width option allows you to set the interval width expressed as a percentage Its value should lie between 1 and 99 Percentile table This option displays a percentile table whereby the number of cases in the sample is divided into equal categories These categories indicate the percentage of cases that fall below the corresponding value of the variable The number of categories computed in this table is determined in the Number option and is displayed with the corresponding variable values For example if you set this option to 4 the program will rank all valid cases divide them into four equal groups and then display the values that delimit the 25th lower quartile 50th median and 75th upper quar
24. 0000 WORKING WITH CHARTS 153 7 WORKING WITH CHARTS Numerous statistical analyses within SIMSTAT include options to create high resolution charts This section describes the steps involved in the creation modification printing and saving of charts Creating specific charts Rather than providing separate commands and modules for numerical and graphical analysis SIMSTAT implements an integrated approach where charts are produced along with numerical analyses The table on the next page allows you to determine which statistical commands should be used to obtain a specific chart To create a chart select the variables and the proper statistical analysis command Set the check box corresponding to the chart type you want If some charting options are also displayed adjust them to suit your needs While the charting options available from statistical analysis dialog boxes are quite limited it is possible to modify almost any chart attribute afterward For more information on the various adjustments available see section Customizing Charts on page 158 154 SIMSTAT for WINDOWS Correspondence between chart types and statistical analysis commands TYPE OF VARIABLES One variable nominal One variable metric Two variables nominal Two variables metric x nominal Two variables metric x metric One metric variable over time Metric variable over time x nominal Other TYPE OF CHART Barchart Pie ch
25. 215 237 Normal probability plot 49 87 120 134 195 197 Program preferences 0 225 235 Q Quadratic regression 00 eee ee 136 220 Quattro Profiles 0 0 000 0005 38 40 199 Quit ng slideil gclseceifeis 207 217 235 236 R Rank Transforming data into ranks 58 219 Statistics 89 110 114 152 195 202 219 Recoding values 00 000 eee 57 59 219 Record Delete record reri oe ei RR 31 238 Filtering records 200000 33 194 Sorting records 0 0 00 e eee eee 36 227 Record script 1 2 0 aiii eee eee 64 171 236 REPETENCES 8 eo ter eet Ao cR A ee 249 252 Regression analysis Bootstrap 00000 70 183 196 217 GLM ANOVA ANCOVA 92 197 Logistic regression 000 112 204 Multiple regression 120 220 Regression analysis 00 133 220 Reliability analysis 100 103 136 139 221 251 Residual analysis 87 120 134 195 197 208 220 Rotated factor solution 0 81 194 Rounding numerical values 64 Running a script 000 171 185 Running median 145 146 224 228 R nstest te V LL 140 222 S Saving script files 0 0 0 00000 172 Scatterplot eere T Le ee leet 133 221 Residual scatterplot 92 120 134 197 208 220 Scheff 12s eod e ah ae es Le ada 130
26. 6 types of nonlinear regression The output includes the Pearson product moment correlations the intercept and slope of the regression line and an ANOVA table for the equation Various options allow you to obtain a bivariate scatterplot to select a one or two tailed test of probabilities and to request a standardized residuals caseplot a scatterplot of predicted values by standardized residuals or a normal probability plot of residuals Options INEAR UADRATIC UBIC TH TH NV LOG EXP Ho 50o0Z7 L Type of regression Linear regression Quadratic regression Cubic regression 4th degree polynomial 5th degree polynomial Inverse regression Logarithmic regression Exponential regression SCRIPT LANGUAGE REFERENCE 221 1TAIL 2TAIL Direction of test XYPLOT Bivariate scatterplot CI integer Confidence interval width CPLOT Caseplot of residuals OUTLIERS real Outliers criterion s d DURBIN Durbin Watson statistic RPLOT Residuals scatterplot PPLOT Probability plot of residuals SAVE Save predicted value and residuals PANEL Display the dialog box RELIABILITY Syntax RELIABILITY varlist BY varlist options Description The RELIABILITY command provides a means to assess the quality of multiple item additive scales through the computation of reliability statistics The options available offer the possibility to obtain various item statistics iter item variance covaria
27. AGGRESS Nb of agressive behavior Adjustment for unequal cells size Hierarchical KKKKKKKKKKKKKK kk kk Ck kk KK KK AG x x xx STEP a KKKKKKKKKKKKKKKKKKK KKK KKK KKK KKK KK xx Variable s entered on step 1 AGE Age of the child Multiple Regression Multiple R 00072 GbR Sieja Om R 03 09 Multiple R Square 0791 Adjusted R square 0630 Analysis of Variance Source of Variation Sum of Squares DF Mean Square F P Variables block 1 LAG iL L3G al LP GiL SAL SG 4 897 102A AGE MZ 6 GSH 1 UA GS SIRSIG 4 897 S0 Siil Explained TAERE 1 TAA AS 4 897 209 Residual 14679 508 57 25995 Total 15940 644 58 274 839 94 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 95 96 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 97 Inter raters analysis Inter rater agreement measures are used to assess the concordance in observed ratings of two judges at the same point in time Such measures can also be used to assess the reliability of the ratings of a single judge at different points in time The simplest measure of agreement for nominal level variables is the proportion of concordant ratings out of the total number of ratings made Unfortunately this measure often yields spuriously high values because it does not take into account chance agreements Several adjustment techniques have been proposed in the literature to correct for the chance factor three of which are available in the SIMSTAT program The following are the assumptions made by eac
28. Confidence interval This option allows you to set the width of the confidence interval for the unstandardized regression coefficients This interval width is expressed as a percentage and must lie between 1 and 99 92 SIMSTAT for WINDOWS Adjusted means This option prints the predicted mean for each cell of all the nominal categorical variables entered in the model controlling for every quantitative variable covariate and or interaction also in the model Test of change This option allows the printing of a summary report of the changes in R square at each step of the analysis Adjustment method This option allows you to choose between 3 types of adjustment for unbalanced designs In the Regression model all factors covariates and interactions are entered simultaneously in the Nonexperimental approach covariates are entered first followed by categorical factors and then by interactions the Hierarchical approach allows you to specify the order in which each variable will be entered in the model The following table shows the correspondence between those 3 methods and the terminology used by other sources SIMSTAT OVERALL amp SPIEGEL SPSS PC SAS Regression Method 1 Regression Type III Nonexperimental Method 2 Classic Experimental Type II Hierarchical Method 3 Hierarchical Type I DIAGNOSIS PAGE Residual caseplot This option allows the display of a casewise plot of standardized residuals including the predicted obt
29. Size of the font between 6 and 100 Color of the font see below Displays the string in italic characters Displays the string in bold characters Hides the string SIZE 10 COLOR YELLOW BOLD Olive Navy Yellow Purple Teal Lime Blue Fuchsia Aqua White 210 SIMSTAT for WINDOWS NPAR Syntax NPAR varlist BY varlist options Description The NPAR command displays a matrix for various measures of association and concordance between two variables The options provide counts and exact probabilities of each coefficient or asterisks that show the probability level reached This probability test can be either one or two tailed Options Type of statistics TAU A Kendall s Tau a TAU B Kendall s Tau b TAU C Kendall Stuart s Tau C D SY Somers D symmetric D XDEP Somers D X dependent D YDEP Somers D Y dependent GAMMA Gamma RHO Spearman s Rho R Pearson s R EXACT Exact probabilities 1TAIL 2TAIL Direction of test Deletion of missing values PAIRWISE Pairwise LISTWISE Listwise PANEL Display the dialog box ONEWAY Syntax ONEWAY varlist BY varlist options Description The ONEWAY command performs a one way analysis of variance for all dependent variables on groups defined by each categorical numeric independent variable It allows testing whether the means of the groups 2 or more are not all equals to each other ONEWAY provides one way variance analysis and a standard tab
30. a shorter test in order to reduce the administration time or cost Second it often occurs that some items while positively correlated with the scale total may reduce the overall reliability of the scale or may contribute only marginally to this reliability The test developer needs to eliminate items without greatly reducing the scale s reliability To achieve this he she needs to know the specific contribution of each item to the scale reliability index The Alpha if item deleted represents such a measure As the name suggests it represents the Cronbach s alpha value we would obtain if the item was removed from the test By successively eliminating items that reduce the Cronbach s alpha or that contribute only slightly to the reliability index of the total score it becomes possible to significantly reduce the number of items while maintaining an acceptable level of reliability MEASURING ITEM DIFFICULTY AND DISCRIMINATION Many scales developed by psychologists and educational researchers are designed to differentiate among individuals at all levels of the measured construct such as when a scale is built to differentiate subjects according to their level of achievement or ability either for selection or classification purposes Several statistics can be used to assess the discriminative quality of an item in a questionnaire The most well known of these statistics is the difficulty index Any response that is answered correctly or missed by all
31. and percentile tables The program also computes bootstrap confidence intervals For estimators which can be tested for significance SIMSTAT also displays nonparametric power estimates for up to four alpha levels Power estimation with the bootstrap technique is straightforward while performing bootstrap on a given data set the proportion of redrawn samples that led to a statistically significant estimator at some given alpha level is computed and used as a power estimate In addition to simulation results the program displays the value of the seed used to initialize the random number generator This value may then be used to regenerate the same data at a later time or to compare various estimators using the same bootstrap samples 70 SIMSTAT for WINDOWS The FULL ANALYSIS bootstrap command also offers the possibility to compute almost any available statistical analysis on successive bootstrap samples and displays the entire results of those analyses EXTENSIONS TO BOOTSTRAP Toachieve an even greater range of potential application SIMSTAT implements two extensions to standard bootstrap simulation Variable sample size Typically a bootstrap simulation randomly draws samples the same size as the original one SIMSTAT offers the possibility to modify the dimension of the bootstrap samples thus allowing users to compare estimator distributions obtained from using different sample sizes You can set bootstrap simulations involving sample
32. as Cronbach s alpha internal consistency and Ferguson s discrimination index Descriptive statistics are also computed on percent correct and item total correlations STATITEM is closely integrated with SIMSTAT and will perform analyses on data stored in each file format currently supported by SIMSTAT The key responses used to specify correct responses may be stored in either the first record of the data file or in a key file NOTE STATITEM v1 0 is an addon module sold separately To get more information on this module or to order a copy contact Provalis Research INTRODUCTION TO ITEM ANALYSIS Usually when one is developing a scale not all items will work as expected Some may be confusing to the respondent some may not tell us what we thought they would others may be too easy or too hard to answer and so on Besides inspection and modification of wording to eliminate jargon terms ambiguous formulations double barrelled questions negatively stated items etc itis common practice to submit the items to a field test involving a sample of subjects and to apply several statistical techniques to the results in order to identify items that are not functioning as intended These techniques allow one to improve the reliability and validity of a test by identifying items that should be eliminated substituted or revised In doing so item analysis makes it possible to increase the overall quality of a scale while shortening it either by eliminat
33. autocorrelation value its variance and probability and a graphic text mode representation of those values from one to a specified number of lags PACF plot This option allows you to obtain a table and a graphic representation of partial autocorrelations with their variances and probabilities Number of lags This option allows you to specify the maximum number of lags to be displayed in the ACF and PACF plots CHART PAGE Time series This option allows the printing of a graphic representation of the transformed series in either text graphics mode or both The Number option allows you to restrict the number of observations to be plotted in text mode Smoothing This option allows the application of two methods of identifying trends in noisy or irregular time series data The Moving average procedure is obtained by averaging a selected number of points on either side of a target value while the running median procedure is computed by finding the median value in a specified number of STATISTICAL ANALYSIS 147 points on either side of the original value Successive smoothing can be achieved by specifying more than one value For example the following option Moving average 424 instructs SIMSTAT to proceed with three successive moving average smoothings using 4 2 then 4 values to compute the mean Control bars This option allows you to display 3 horizontal bars that represent the mean and the upper and lower limits of a con
34. cases with missing values by either PAIRWISE or LISTWISE deletion If you select pairwise deletion a case is excluded if it has a missing value on either of the two variables used to compute a given correlation coefficient However this case can be included in the computation of other coefficients Listwise deletion excludes cases containing missing data from the computation of all the correlations included in the matrix Cross product covariance matrix This option displays cross product deviation and covariance tables for the data 76 SIMSTAT for WINDOWS Sample output of a correlation matrix STATISTICAL ANALYSIS 77 Crosstabs The CROSSTAB command produces a standard contingency table for two variables where rows represent the dependent variable values and columns represent the independent variable values The dialog box allows you to include various statistics in the table count row column or total percentage etc and obtain various measures of association for nominal ordinal and interval levels of measurement OPTIONS TABLE PAGE Contingency table This option requests the output of a contingency table Sort by Use this option to tell the program whether rows and columns of the table should be sorted by the values of the variable or by frequency Type The type option allows you to specify whether rows and columns of the contingency table should be sorted in ascending or descending order Table content gro
35. conditions or with other behaviors of the same subject The SINGLE CASE command provides some basic tools for studying the effect of an intervention on the behavior of a single subject The procedure will display a graph representing the evolution of the dependent variable Y at various phases defined by the independent X variable The dialog box allows one to obtain various statistics for each phase of the analysis as well as various graphic tools that can be used as judgemental aids to identify the experimental effect of the intervention smoothed data trend lines control bars etc OPTIONS Statistics This option box allows you to display statistics about the different phases of the experiment When set to Brief the mean standard deviation minimum and maximum values and the number of cases are displayed on a single line To obtain additional statistics such as the skewness kurtosis mode median etc set this option to Detailed Cumulative frequency This option allows you to represent the values of the series as a cumulative record of frequency STATISTICAL ANALYSIS 145 Log transformation This option converts all values in the series into their natural logarithms Trend line This option allows you to draw for each condition a line representing either the mean the regression slope or a split middle trend line This last method draws a line through the median values of the first and second halves of each series
36. create listings of all data sets used in your analysis It may also be a good idea to keep a log of the analysis commands and options used with your statistical results To achieve this simply use the RECORD SCRIPT feature to automatically generate those commands and copy the content of the script window to an empty notebook page Rounding numerical values SIMSTAT provides a feature that allows you to reduce the number of decimal places of numeric values within a selected area or in the active page of the Notebook To round the values perform the following steps f necessary select the portion of text containing the numeric values you want to round Select the ROUND command from the EDIT menu The following dialog box will appear Round to how many decimals n B C Current page Selected text i x Cancel Help Specify the number of decimal places to which you want to round numbers Set the radio button to Current Page or Selected Text depending on whether you want to round values for the entire page or only in the selected portion of it e Select the OK button to exit the dialog box STATISTICAL ANALYSIS 65 6 STATISTICAL ANALYSIS This reference chapter provides a description of the Variable Selection dialog box followed by a description of every statistical analysis that appears in the STATISTICS menu To allow you to quickly find the information you need the statistical commands are
37. if we are notified of such possibility in advance Some states do not allow the exclusion or limitation of incidental or consequential damages so the foregoing limitation may not apply to you In no event shall Normand P ladeau s liability for any damages ever exceed the price paid for the license to use the software regardless of the form of claim This agreement shall be governed by the laws of the province of Quebec Canada and shall inure to the benefit of Normand P ladeau and any successors administrators heirs and assigns Any action or proceeding brought by either party against the other arising out of or related to this agreement shall be brought only in a PROVINCIAL or FEDERAL COURT of competent jurisdiction located in Montr al Qu bec The parties hereby consent to in personam jurisdiction of said courts COPYRIGHT Copyright 1996 Normand P ladeau All rights reserved No part of this publication may be reproduced or distributed without the prior written permission of Normand P ladeau 2414 Bennett Street Montreal QC CANADA HIV 384 Trademarks IBM PC and PC DOS are registered trademarks of International Business Machines Corporation Microsoft Windows and MS DOS are registered trademarks of Microsoft Corporation Excel is a product of Microsoft Corporation SPSS PC and SPSS for Windows are a registered trademark of SPSS Inc dBase and Paradox are registered trademarks of Borland International Quattro Pro is a regi
38. item Moves to the previous item To move directly to an item press the Alt key and the underlined letter in the associated text The action needed to edit the values of the various options depends on the type of input field Options panels can contain several types of input field Edit box The Edit box is a rectangular box in which you type a value using the keyboard They are used to enter a string of characters a number a filename etc When you are positioned on such a field a blinking text cursor appears The following table lists the keys that can be used while editing a data entry field Key Right arrow Left arrow lt Ctrl left gt lt Ctrl right gt lt Del gt lt Backspace gt lt Home gt lt End gt lt Ins gt Esc Effect Moves the cursor one character to the right Moves the cursor one character to the left Moves the cursor to the beginning of the previous word Moves the cursor to the beginning of the next word Deletes the character to the right of the cursor Deletes the character to the left of the cursor Moves the cursor to the beginning of the line Moves the cursor to end of the line Pressing this key toggles between the insert and overwrite modes In insert mode characters are inserted at the cursor position pushing text to the right of the cursor even further right When in overwrite mode characters at the cursor position are overwritten Aborts the editing process
39. kappa Scott s pi Free marginal nominal e Krippendorfs r bar Krippendorfs R Free marginal ordinal CHART PAGE Barchart If this option is activated the program displays a 2 dimensional barchart that provides a graphical presentation of the relationship between the dependent and the independent variable Type The Type option offers a choice between 4 different ways of displaying a bivariate barchart naclustered barchart the bars representing the frequency of each category of the independent variable are placed side by side When the overlayed barchart is chosen the bars art is displayed in perspective on a 3 axis plane where bars representing the frequency of each category of the independent variable are placed on different rows This chart type is only available in 3 D view While this kind of chart is very popular we strongly recommend STATISTICAL ANALYSIS 99 against its use since it is virtually impossible to determine the exact heights of the bars or compare the heights of bars located on different rows n a stacked barchart the bars representing the frequency of each category of the dependent variable are stacked on top of each other The 100 bars type is similar to the stacked barchart in that the bars for each category of the dependent variable are stacked on top of each other However just like a pie chart each bar represents the proportion of a category of the independent variable from
40. mean and the confidence interval on the first 10 observations Confidence interval This option allows you to set the confidence interval width for the control bars This interval width is expressed as a percentage and must lie between 1 and 99 146 SIMSTAT for WINDOWS Time series analysis The TIME SERIES command allows the examination of time series data The dialog box offers various transformations to remove trends or seasonal dependence in a series and provides a diagnostic of those transformations by displaying autocorrelation and partial autocorrelation function plots of the transformed series This procedure also allows the application of two smoothing methods 1 e moving average and running median to identify trends in noisy time series data OPTIONS ANALYSIS PAGE Log transformation When enabled this option converts all values in the series into their natural logarithms Remove mean When enabled this option subtracts the mean from each value in the series Difference This transformation subtracts the preceding value from each value in the series The Number option allows you to set the number of difference operation to be performed on the series Seasonality The Seasonality option allows you to remove the seasonality in a series by subtracting from every value the value that is a specified Number of lags behind it ACF plot The ACF plot produces an autocorrelation function plot This plot includes the
41. more of available RAM and 3 5 MB of hard disk space A mouse is optional but highly recommended The program does not need a numeric coprocessor but will use it if available However a coprocessor is strongly recommended for analysis of large samples or for extensive bootstrap resampling analysis Installing SIMSTAT This manual assumes that your hard disk is drive C and that the installation disk is on drive A You can change the drive and or directory by making the appropriate substitutions in these instructions To install the program on your hard disk Insert Program Disk 1 in the proper diskette drive If Windows is not currently running start Windows by typing WIN and press Enter From the Program Manager choose RUN from the FILE menu e Type A SETUP in the command line box and choose OK Follow the prompts on screen to complete the installation Making a backup copy Be sure to make a backup copy of the distribution disks in case you lose or damage the originals Keep the original disks in a safe place away from direct heat dust and magnetic sources 4 SIMSTAT for WINDOWS Basic program information Starting the program To run SIMSTAT for Windows start Windows if necessary If you run Windows 3 1 display the program manager icon double click the SIMSTAT for Windows group icon to open it and double click the SIMSTAT program icon If you run Windows 95 click on the START button then poi
42. name to this list box Using the same procedure move the variable SIBLING to the independent list box NOTE If you select the wrong variable you can remove the variable from the Independent list box by clicking on the variable name When a name in the independent list box becomes highlighted the arrow of the button beside this box changes direction indicating that pressing this button will put the variable back in the variable list box Click on the OK button to close the dialog box and activate the variable selection If you want to cancel the operation and restore the previous variable assignments click on the CANCEL button TUTORIAL 15 Step 3 Performing frequency analyses To perform a frequency analysis on the variables we have just chosen select DESCRIPTIVE FREQUENCIES from the STATISTICS menu The following dialog box appears on screen requency options x Table r E Iv Frequency table E Paes Be Sort by Type X Cancel Value Ascending B Choose C Frequency C Descending B cese Heb v Descriptive statistics Confidence interval Width e Percentiles table Number 20 Analysis Charts Options Using this dialog box we will instruct the program to print for every selected variable a frequency table sorted in ascending order of value detailed descriptive statistics and a histogram To achieve this Activate the Frequency Table check box a check wil
43. of the unrotated factor solution will be included in the output You can turn this off if you re running multiple analyses on the same data and do not need to have this information repeated in the output Rotate factor solution When enabled the factors are rotated to the varimax criterion and the results of this analysis are included in the output You may wish to turn this option off during the initial analysis of a very large data set when you re most interested in determining the number of factors in your data as rotation of a large number of factors can be relatively time consuming Factor scoring weights When enabled EFA will calculate factor scores for both the rotated and unrotated solutions The factor scoring coefficients will be printed in the output By default factor scores are not calculated as they are not always needed are relatively time consuming to calculate and are usually not calculated until a final factor solution has been arrived at Display factor scores If you would like the factor scores for your subjects to be printed in the output toggle this option on Be warned that including the factor scores in the output can increase its size dramatically Sample output of a factor analysis FACTOR ANALYSIS Easy Factor Analysis 3 0 iem amena cS ALI TRIALS Reserved Analysis Parameters Path IN File SIM2EFA Analysis Type Principal Components Factors 10 Min Eigenvalue 1 00000 Max Iterations 30 Ro
44. or select it and click on the OK button If you want to open a script file that has been used previously click on the down arrow button at the right side of the File Name edit box and select the filename If the selected file is an encrypted file with a SCZ extension the text editor will be hidden and only the name of file will be displayed in the middle of the script window IS To prevent accidental modifications to the contents of a script file activate the Read Only check box in the Open File dialog box before clicking on the OK button While this procedure disables all editing features including the RECORD SCRIPT command it will still be possible to view or print the content of the script window and to cut text from it and paste this text to another SIMSTAT window or another application To prevent these operations create an encrypted version of the script file by saving the file under a new file name with a SCZ extension 170 SIMSTAT for WINDOWS Navigating in the script window To move the caret to a specific location click on that location with the mouse You can also use the following keys to navigate in the script window Key Left Right Up Down PgUp lt PgDn gt lt Home gt lt End gt lt Ctrl Home gt lt Ctrl End gt lt Ctrl Right gt lt Ctrl Left gt Action Move the caret one character to the left Move the caret one character to the right Move one line up Move one line do
45. polynomial Y a b Inverse Y a b Logarithmic Y a b Exponential Y ab x 4 bx rbaxrbax x 4 byx rbax rb x b x In x Confidence interval This option allows you to set the confidence interval for beta weight estimates This interval width is expressed as a percentage and must lie between 1 and 99 Significance test This option specifies whether the probability of the correlations probabilities is based on a one tailed directional or two tailed non directional test Scatterplot Option Scatterplot produces a bivariate scatterplot with the independent variable plotted along the horizontal axis axis and the dependent variable along the vertical Residual caseplot When checked this option produces a casewise plot of standardized residuals including the predicted obtained and residual values The Outliers value 134 SIMSTAT for WINDOWS allows you to restrict the residual caseplot to those cases for which the absolute standardized value is greater or equal to the specified value The Outliers value can be set to between 0 and 4 standard deviations Durbin Watson statistic This option tests for the presence of autocorrelation or serial correlation in the residuals The larger the autocorrelation the less reliable the results of the analysis Residual scatterplot This procedure produces a bivariate scatterplot where the predicted value is plotted along the horizontal axis and the standardized residua
46. prevent students from cheating by looking at your code Encrypted files are created quite simply by saving an opened script under a filename with a SCZ extension The resulting file will be about 50 to 80 smaller than the original file and may be run from within SIMSTAT just as any other script file However the file can no longer be viewed or edited either from within SIMSTAT or from an external editor USING SCRIPTS 169 Opening an existing script To open an existing script file select the SCRIPT OPEN command from the FILE menu This evokes an Open File dialog box When this dialog box is displayed the program points to the default data directory and displays all available data files in this directory in the File Name list box To open a file double click on its name or select it and click on the OK button If the name of the script file you want to open is not displayed type the filename in the File Name box including drive and path if necessary and select the OK button You may also use the following methods to locate the data file If the script file is on a different disk click on the down arrow of the Drives list box to display available drives and select the disk where the file is located Ifthe file is in a different directory double click on the directory names in the Directory or Folder to move through the directory tree If the file name is displayed in the File Name list box double click it to open the file
47. probabilities cases 1TAIL 2TAIL Direction of the test Deletion of missing values PAIRWISE Pairwise LISTWISE Listwise CESS Cross product deviation amp covariance CI integer Confidence interval width PANEL Display the dialog box CROSSTAB Syntax CROSSTAB varlist BY varlist options Description The CROSSTAB command computes a contingency table for two variables where rows represent all the independent variable x values while the columns represent all the dependent y variable values The options allow you to include various statistics in the table and obtain measures of association for nominal level chi square phi contingency coefficient and ordinal level gamma tau b tau c Somers d etc variables It also allows you to display a 3 d barchart of the two variables GEO Lu ROWPCT COLPCT TOTPCT EXPEC SID ESID wn De CHI2 LRATIO CONTINGENCY PHIPhi 2 x 2 TAU B TAU C GAMMA ERS RHOSpearman s Rho PEARSON or BARCHART OVERLAP STACKED 100 3D PANEL DATA ENDDATA Syntax DATA ENDDATA Description Cramer s V SCRIPT LANGUAGE REFERENCE 191 Sort table by Ascending frequency Descending frequency Ascending value Descending value Display table Row percentages Column percentages Total percentages Expected values Chi square residuals Standardized residuals Chi square statistics Lik
48. rank of each variable the number of cases chi square degrees of freedom and probability value Because the Friedman test is used to compare correlated samples it does not really make a distinction between dependent or independent variables The Variable Selection dialog box is used instead to specify which variables should be compared together For example assigning TIDEPRESS T2DEPRESS and T3DEPRESS to the Independent list box and TIAGGRESS T2AGGRESS and T3AGGRESS to the Dependent list box will result in two separate Friedman tests the first one comparing the first 3 variables and the second one comparing the other 3 Sample output of a Friedman two way anova for K related samples FRIEDMAN TWO WAY ANOVA Mean rank Variable doit TIDEPRES Doc AL Abts T2DEPRES d TIE T3DEPRES Cases Chi sqare DARS Significance 38 51997199 2 0422 90 SIMSTAT for WINDOWS GLM ANOVA ANCOVA GLM ANOVA ANCOVA is a procedure which permits the researcher to examine the effects of one or more variables on a single continuous dependent variable This procedure provides a way to test the equality of means in categories within a single variable or factor main effects as well as categories formed by combinations of two or more variables or factors interaction effects Analysis of covariance ANCOVA allows the comparison of the effect of categorical variables on the dependent variable while controlling for the effects of one or more quanti
49. recoding of JOBCAT Variable prefix JOBCAT Type of recoding D y C Effect Create variable for last value OK x Cancel Yd Help Variable prefix The variable prefix option allows you to enter a prefix that will be used to generate dummy variable names The maximum length of this prefix is 7 characters Numeric values going from 1 up to the number of variables required will be appended to this prefix Type of recoding This option provides a choice between two different methods for recoding nominal variables Dummy The dummy or binary coding involves the creation of dichotomous vectors to represent membership in the categories of nominal variables such that subjects in a given category are assigned 1 while non members are given a score of 0 The number of dummy variables needed to contain all the information of a nominal variable equals the number of different values in this variable minus one While most analyses involving dummy variables require you to enter only those g 1 dichotomous variables DATA TRANSFORMATION 59 you may need to perform separate analysis for each category When the dummy coding type is chosen a check box appears allowing you to create this last dichotomous variable For example if you recode a variable containing 3 different values setting this check box will create 3 dummy variables one for each group If this option is disabled the procedure will create only 2 variables
50. returns 20 Constants PI 3 1415926535897932385 Numeric and Trigonometric Functions Syntax FUNCTION value variable or expression ABS Absolute ACOS Arccosine ASIN Arcsine ATAN Arctangent CSC Cosecant COS Cosine EXP Exponential FACT Factorial LN Natural logarithm LOG Base 10 logarithm MOD10 Modulus RND Round SEC Secant SORT Square root SOR Square SIN Sine DATA TRANSFORMATION 53 TAN Tangent TRUNC Truncate Statistical Functions across variables The following statistical functions are computed on several variables for each record individually For example if you enter MEAN Q1 O2 Q3 Q4 Q5 or MEAN Q1 0Q5 SIMSTAT will compute the mean of values stored in the five specified variables of the current record missing values are excluded To obtain a statistic on all selected records see the next section Statistical Functions across records Syntax FUNCTION Var Var Var Var MEAN Mean COUNT Count missing value excluded SUM Sum STDEV Standard deviation VAR Variance MIN Minimum MAX Maximum WSUM Weighted sum adjusted for missing value Statistical Functions across records The following statistical functions are computed on a single variable for all currently selected records For example if you enter VMEAN AGE SIMSTAT will compute the mean of the values stored in the AGE variable f
51. right while all text entered in the CENTER will be printed in the center of the page Special codes can be inserted in any of these boxes to print specific information Code Effect p This code will be replaced by a page number d Putthis code in the title to print the date of the page creation t Print the time of the page creation f This code will be replaced by the original data file name Print a dollar sign Text printing mode You can choose among 3 different printing methods for printing the content of the notebook Oneanalysis When this option is enabled every analysis will start printing at the top of a new page e Smart When you select this option the program will try to fit several analyses on a single page If an analysis cannot be entirely printed in the remaining space SIMSTAT will start printing at the top of a new page Economy Selecting this option will print every analysis one after the other Even if an analysis cannot be entirely printed in the remaining space the first lines will be printed on the current page while the remaining lines will be printed on a new page Left margin This option allows you to increase the left margin of the printed analysis outputs Maximum width This option lets you specify the maximum width of analysis results By default this value is set to 90 characters You can increase this value up to 254 characters so that larger correlation matrix and contingency ta
52. sciences 2nd edition San Francisco Dellen Macmillan FERGUSON G A 1989 Statistical analysis in psychology amp education sixth edition New York McGraw Hill TABACHNICK B G amp FIDELL 1989 Using multivariate statistics 2nd edition New York Harper Collins WINER BJ 1971 Statistical principles in experimental design New York McGraw Hill COMMON PROBLEMS IN STATISTICAL ANALYSIS ALLISON D B GORMAN B S PRIMAVERA L H 1993 Some of the most common questions asked of statistical consultants Our favorite responses and recommended readings Genetic Social and General Psychology Monographs 119 2 153 185 COHEN J 1990 Things I have learned so far American Psychologist 45 1304 1312 OAKES M 1986 Statistical Inference New York Wiley MEASURES OF ASSOCIATION GIBBONS J D 1993 Nonparametric measures of association Beverly Hill Sage Publication HILDEBRAND D K LAING J D amp ROSENTHAL H 1977 Analysis of ordinal data Beverly Hill Sage Publication LIEBETRAU A M 1983 Measures of association Beverly Hill Sage Publication REYNOLDS H T 1977 Analysis of nominal data Beverly Hill Sage Publication NONPARAMETRIC STATISTICS CONOVER W J 1980 Practical nonparametric statistics 2nd Ed New York John Wiley amp Sons HOLLANDER M amp WOLFE D 1973 Nonparametric statistical methods New York John Wiley amp Son REFERENCES 251 SIEGEL
53. selection dialog box Using the previous instructions assign the AGE and HOURSTV variables to the Independent list box and move the AGGRESS variable to the Dependent list box Click OK to confirm this variable selection and return to the analysis dialog box Then using the proper keys or mouse actions Set the Type of Analysis option to Linear e Set the numeric field beside Confidence Interval to 90 in order to obtain a 90 confidence interval on beta weights Select a 2 tailed test by clicking on the proper radio button To obtain a scatterplot that will allow you to visualize the relationship between the dependent and independent variables set the SCATTERPLOT option to Graphic TUTORIAL 19 Disable the remaining options so that the dialog box looks similar to the one displayed above When you click on the OK button SIMSTAT calculates two separate linear regression equations with AGGRESS as the dependent variable and AGE and HOURSTV as predictors You may browse through the results of this analysis if you so desire Step 7 Keeping hard copies of results Once analyses are performed you will often want to keep copies of the numerical results and charts produced during a session To save the results on disk Choose the NOTEBOOK SAVE command from the FILE menu A file save dialog box will appear By default SIMSTAT use the SNB extension for notebook files If no extension is given the program automatically adds th
54. tendencies It tests the likelihood that those groups were drawn from populations with the same median The output displays the number of cases greater than the number of cases less than and the number of cases equal to the median for each category of the grouping variable Also displayed are the median chi square degrees of freedom and probability value OPTIONS Type The Type option allows you to choose between a design including 2 samples or an extended version that tests for more than 2 Values of x You must enter two values in the Values of X option If the design chosen is a 2 samples test then two groups are formed using the two values If the extended test is chosen then every value in the range defined by the two values forms a group A test is then performed on all groups Sample output of a median test MEDIAN TEST AGGRESS Nb of agressive behavior With SIBLING Number of siblings Gt Median Le Median SIBLING 00 J D SIBLING 1 00 TI sles SIBLING 2 00 4 2 Cases Median Chi square D F Signifiance DO 26 000 3900950 2 SOVEN STATISTICAL ANALYSIS 117 Moses test of extreme reactions The MOSES TEST of extreme reactions tests whether the range of an ordinal variable is the same in a control group as in a comparison group as defined by a grouping variable The output includes counts for both groups number of outliers removed the span of the control group before and after outliers are removed a
55. the location and setting of those numerous options multiple page dialog boxes have been used to functionally group those options on different pages For example many statistical analyses contain a separate page for all options related to the production of high resolution charts while the numerical analyses options are located on another page You can access a particular page in the dialog box by clicking on its associated tab located at the bottom of the dialog box After the values have been edited to suit your needs you have to click on the OK button in some dialog boxes this button is named APPLY to accept those values and proceed If you want to leave the dialog box restore the previous values and suspend the current operation just click on the CANCEL button A HELP button is often displayed to give you access to the context sensitive help file Some dialog boxes also provide additional command buttons which give you access to special functions GETTING STARTED 9 Toolbar The main toolbar is displayed across the top of the application window below the menu bar The toolbar provides quick mouse access to many commands used in SIMSTAT To hide or display the toolbar you can use the Preferences menu command and set the Show Toolbar check box accordingly When the mouse cursor rests over a toolbar button for more than one second a small help hint appears displaying a short description of this button To turn these help hints on
56. the beginning and end of the block ending keyword Comments can be inserted anywhere in the file by placing an asterisk as the first character Comments can also be included anywhere within a command by enclosing them between braces PANEL The PANEL keyword can be used with almost any statistical analysis command to display the dialog box associated with the command allowing the user to change options before performing the analysis The character is used as a command terminator Each command must end with the character A colon at the beginning of a line indicates the presence of a label see the GOTO and GOSUB command The slash character is used to separate the various options from the commands and the variables While only one slash is required it is also possible to include more slashes to visually separate various groups of options m multiple lines It is always possible to spread a long command over several lines or even insert blank lines within a command The semi colon character is always used to indicate the end of the command Using memory and data file variables A script can contain 2 different kinds of variables Memory and data file variables A memory variable is the name of a location in memory that stores a value The value of a memory variable can change during script execution Every variable has a name that begins with a dollar sign as its first character a data type 1 e
57. the number of times resampling is carried out This number can range from 10 to 100 000 Initial seed value SIMSTAT automatically provides a seed value for the random number generator that drives the simulation analysis Alternatively the Initial Seed Value option can be used to specify a seed value To instruct SIMSTAT to randomly 72 SIMSTAT for WINDOWS choose a new seed value click on the Shuffle button SIMSTAT outputs the seed value with the simulation results This value can be used to regenerate the same data at a later time or to compare various estimators using the same bootstrap samples Same as original When this option is enabled the size of each bootstrap sample drawn from the original data is automatically adjusted to the size of this original sample When disabled you can use the Size option to specify how large each bootstrap sample will be The sample size can range from 1 to 100 000 cases OUTPUT PAGE Descriptive statistics Enable this option to obtain a table of various descriptive statistics Thistable includes the following statistics mean standard error of the mean sum mode median standard deviation variance skewness kurtosis minimum maximum and range Confidence intervals The Confidence Intervals option allows you to display the 90 95 and 99 confidence intervals of the coefficient The program also displays bias corrected confidence intervals that apply a correction to the interval
58. the total number of observations in a specific category of the dependent variable This type of barchart is especially useful if you want to compare the proportions between different categories of the dependent variables rather than the absolute frequency Perspective This option allows you to specify whether the barchart should be displayed in 2 or 3 dimensions A sample output of inter raters analysis INTER RATERS TABLE EVENT1J2 Level of stress Judge 2 by EVENT1J1 Level of stress Judge 1 EVENT1J2 Count Low Medium High owe 12 exe Ie GO 2 00 3e0U Total EVENT1J1 i S000 24 4 28 Low 40 7 5 8 20 47 5 IO 22 2 24 Medium a STS 3 4 40 7 AU it 2 4 7 High ion 3 4 Ove Wis Column 2 28 6 59 Total 42 4 47 5 1 0 527 ROO vO INTER RATER AGREEMENT MEASURES Nominal level Pct agreement 84 7 Cohen s Kappa 74 3 Scott sUpiv 742 Free marginals 77 1 Ordinal level Krippendorf s r bar 77 1 Krippendorf s R 77 1 Free marginals 84 7 100 SIMSTAT for WINDOWS Item Analysis STATITEM v1 0 performs classical item and test analysis for multichoice item questionnaires It computes the percentage endorsement and the item total correlations for correct and alternate responses It also provides endorsement rates for various achievement levels and descriptive statistics on the total score such the mean median minimum maximum variance skewness kurtosis etc as well
59. unmebrsubghb etu 77 DGS CEI PEA VES used el e eem e tret dee ecd onde de d Rede Gate SER 80 Factor anal ysis ss s uars r exu ex eau uu eee nuu Ens aS ERE SE 81 Erequ ncleS sr REISEN REALE RE AS Ce AS EEA Re 86 Friedman Tests 2 lt ho ek ed Re yt Ba cc ehem E eR rye o dittee ni 89 GEM ANOVAJANCOWA x igttur p ERE TS E 90 Inter raters analysis 4 2139452 84 E eh a A ECRANS AES 97 Item Analysls i e heroes eror mene capter Reto Peres dos 100 Kolmogorov Smirnov 1 Sample Test 2 0 0 0 0 108 Kolmogorov Smirnov 2 Samples Test 0 cee eee eee eee ene 109 Kruskal Wallis x ssp halts ee a Se oe oN eee ed 110 LAStiti CASES T DE 111 Logistic TESTESSION xad aa be BORE TARR DOD ERE BOER ER RR RO 112 Mann Whitriey U tests ons peux boot di p eei d e VER astute RC e dats 114 MeNemat test cios s Bay Res es AS Baa UR UA sg ed Re E ERE NE ey duo 115 Median test 2er RR Eee wea ee RR EE imet pem ed 116 Moses test of extreme reactions sese eeeenee 117 Multiple fegtesslOD ro UR Eoo eed cte iced en iet doa s denda ta 118 Multiple responsesanalyses lee 125 Nonparametric matrix ooe enre er n Ta n 127 One sample chi square test 1 0 eee eee eee eens 129 Oneway ANOVA uictis sec Bes ed tad Be E ep pru rt E 130 Regression analysis nna s lw ni AIR ee eee oe ee et x 133 Reliabihty an lysls ws 2st usen toe ectetuer v E Ree uei aoe ieee 136 Runstests 5G chee SESS ee wa RSS RS Ee way Bed Pee ees 140 Sensitivity analys
60. used as a cutoff point All observations falling below this value form one group and all observations equal to or above the cutoff point form the other group If two numbers are specified cases with values equal to the first number form one group and cases with values equal to the second number form the second group Providing no values allows the analysis of dichotomous variables Proportion This option allows the specification of the test proportion This value must lie between 0 and 1 Sample output of a binomial test analysis BINOMIAL TEST AGGRESS Nb of agressive behaviors Mean 27 5424 Proportion 5000 Cases 37 Lt Mean Observed proportion 6271 22 Ge Mean Z 1 8226 2 tailed P 0684 68 SIMSTAT for WINDOWS Bootstrap analysis The BOOTSTRAP submenu gives you access to an innovative and extremely powerful statistical technique called bootstrap simulation This technique developed by Efron Efron 1981 Diaconis amp Efron 1983 can be used to assess various properties of statistical estimators such as their accuracy their sampling variability etc Typical applications include the computation of nonparametric estimates of sampling distributions the assessment of the stability of statistical estimators and the construction of nonparametric confidence intervals SIMSTAT also allows the computation of nonparametric power estimates and Type I error rates for various estimators The following section pro
61. variable names A set of consecutive variables can be designated by typing the first and last variable names separated by two dots For example the DEPRESI DEPRES29 expression refers to all the variables in the active file starting from DEPRES up to and including DEPRES29 A variable list can span over several lines filename A filename with a valid extension By default the file is assumed to reside in the starting directory To refer to a file in another location specify the full path name integer Integer value You can either use an equal sign z between the option and the integer or put the integer between parentheses real Real value May be entered in either normal or scientific notation and can be put after an equal sign or between parentheses string A string of alphabetical as well as numeric characters Some commands require the string to be enclosed between quotation marks color A keyword representing a color Valid keywords are BLACK PURPLE BLUE MAROON TEAL FUSCHIA GREEN GRAY AQUA OLIVE SILVER WHITE NAVY RED BLUE LIME Items inside square brackets are optional 174 SIMSTAT for WINDOWS A vertical bar indicates a choice between two or more items item item A horizontal three dot ellipsis means more of the preceding items can be used in a single line statement keyword A vertical three dot ellipsis is used to indicate block structured statements Textual information can be put between
62. whether this condition is associated with higher or lower values of the scale Sensitivity statistics This option allows you to obtain the level of sensitivity proportion of positive cases correctly diagnosed as true and specificity proportion of negative cases effectively diagnosed as false for each value of the quantitative measure The values are reported in terms of absolute and relative frequencies Error rate statistics This option allows you to obtain the number and percentage of false positive and false negative diagnoses for each value of the quantitative measure The display also includes likelihood ratios for both positive and negative results ROC curve This option allows you to obtain a receiver operating characteristic ROC curve that displays the relationship between the sensitivity and 1 specificity Error rate graph This option provides a line chart that displays the evolution of the percentage of false positives and false negatives for the values of the scale Sample output of sensitivity analysis SENSITIVITY ANALYSIS DISORDER by TEST Area under ROC curve T9296 Cutting Cases Control i Point Included Sensitivity Included Specificity Specificity 100 0 n 0 00 d 0S5 0 05 9 0 2 0 10 al 0795 0 05 8 0 4 Bacon T 0 95 Degas TEQ 9 0 60 3 0 86 0 14 6 0 X13 0 80 5 0 76 0 24 S O 18 Duce p 9 OS 0 43 4 0 2S 3t c9 15 O29 Oa yI S50 29 1 00 19 0 10 0 90 250 Sale HIST ZAL 0 00 LOO Cutting False False False F
63. will be removed from the model The P to Remove should always be greater than the P to Enter value Tolerance The Tolerance criterion is used to prevent the inclusion of a variable that would produce multicollinearity in the equation This measure is the proportion of variance of the variable not explained by the other independent variables already in the equation or 1 R2 In order to be entered or to remain in the equation a variable must pass this tolerance test To disable this function set the criterion value to zero ORDER PAGE When a hierarchical multiple regression method is chosen the dialog box displays an Order tab with a grid that lets you specify the order of entry of each independent variable in the in the regression model The order number can lie between 1 and 20 All variables with the same number are entered at the same time while variables that have lower numbers are entered before those that have higher numbers Variables with identical order numbers will be entered simultaneously You can enter the order by using the keyboard or the spin button located at the right edge of the grid At each step information can be generated about the degree to which a new variable or set of variables explains variance in the dependent variable Sample output of a hierarchical multiple regression MULTIPLE REGRESSION Dependent Variable AGGRESS Nb of agressive behavior Method Hierarchical entry de dede dece e de ede e d
64. window are shown and a comprehensive menu system where all menus are displayed even if unrelated to the active window While the task specific menu system may facilitate the location of a specific command by displaying fewer menu commands at a time performing a task related to another window requires you to first activate this window in order to gain access to the proper menu For example if you are browsing through the notebook and want to filter the data file you will need to activate the data window before accessing the DATA menu where the FILTER command is located The full menu system allows you to perform any action related to any of the 4 window types without changing the active window Save and restore desktop information Select this option to keep information about the sizes and locations of the 4 windows at the moment you leave SIMSTAT The next time you run the program the locations and sizes of the 4 windows will be restored ES To make the starting locations and size of those windows permanent Enable this option Adjust the size and location of the 4 windows to your preference Quit SIMSTAT Restart SIMSTAT Disable this option to prevent overwriting previously saved information 236 SIMSTAT for WINDOWS Display notebook after analysis When this option is enabled the Notebook window automatically becomes the active window after an analysis command is performed and its results added to the notebook If the D
65. you to set the compression factor to be used by SIMSTAT when creating archive copies of data files You can set the compression factor to a numeric value between 0 and 9 When set to 0 files are simply stored in the archive Setting this compression factor to 9 gives the best compression ratio but is also the slowest compression The default value is set to 6 Keep a session backup When this option is enabled SIMSTAT automatically creates a temporary compressed copy of a data file upon its opening This feature is especially useful to cancel all data transformation or editing performed during a session and restore the file as it was when you opened it It is also possible to refresh this temporary backup by using the DATA SESSION BACKUP REFRESH option This will ensure that all modifications to the data file made so far will not be lost if you later decide to revert to a previous version of the data file If you quit SIMSTAT or open another file this temporary backup file is deleted SETTING THE PROGRAM PREFERENCES 237 PRINTING PAGE The printing page allows you to set various printing options for both text and graphic outputs Header Use this header option to specify a line of text that will appear at the top of each analysis when those analyses are sent to the printer The header is composed of 3 strings All text that appears in the LEFT box will be printed on the left margin All text appearing in the RIGHT box will be printed flush
66. 000000000 67 181 Bitmap enose Selb e ete tee 10 165 166 214 BMP files 0000 10 165 166 214 Bootstrap analysis 68 73 182 183 196 217 Ref rences ss S ce Malet AN RUE EA 252 Box amp Whisker plot 774 87 185 195 Breakdown analysis 000 74 185 C Caseplot GLM ANOVA ANCOVA sss 92 197 Multiple regression 0 120 220 R gressiOn 2r ELLE E e 133 220 Carts ERE 153 166 BID NASW iss b nseni s els Delo dot rises nuh rdiet 162 Axis scaling and grid 0 0 159 Creating specific charts 0 154 Customizing charts 0 158 164 Exporting charts 0 0 0 0 000 000 165 Data point values 00 anae 160 Legend tence ted eedbadongaeaghehaanass 160 Navigating in the chart window 155 Setting global options 0 160 Titles and axis labels Lus 158 Zoominginandout llle eese 163 Chi square 0 0 79 129 186 190 CHOOSE SY 222 15 Moat Grder de ela ets 14 65 CHX HES r oe hee hone cia aie Seva 19 157 211 Clipboard 6 10 155 165 170 Cluster analysis 2 0 00 00000000008 66 187 Cohen s Kappa 83 97 187 200 217 Compression 0 20 cece eee eee 45 236 Conditional transformation 55 188 Confidence intervals Bootstrap analysis 68 70 72
67. 182 183 217 Correlations 5 e AS Sa e se aed des 75 190 Frequency analysis selle 86 195 Logistic regression 112 113 204 Oneway ANOVA sees 130 210 TOS Doe io dt cords Seca dacs fon boil e Ro MES 149 227 Confirmation 0 0 0 0 eee een 238 Contingency table 77 115 125 190 Control bars 144 145 147 223 228 Correlation Bootstrap 0000 70 183 196 217 Correlation matrix 00050 75 190 Crosstabulation 77 125 190 207 GLM ANOVA ANCOVA 92 197 Item analysis 0 0 0 0 0 0 104 201 Logistic regression 00 112 204 Multiple regression 120 220 Regression analysis 00 133 220 Reliability analysis 136 139 221 Correspondence analysis 00 5 189 Covariance matrix Correlation analysis 00 75 190 Factor analysis 2 edee cu I RR RS 81 193 GLM ANOVA 0 000002 90 196 Reliability analysis 136 138 221 Creating a new datafile 0 24 Creating specific charts 00 153 Crosstabulation 77 125 190 207 R f rences i oet oe hehe Se ay bea 249 Cubic regression llselelle eee 136 220 Cumulative distribution chart 86 195 Customizing charts 004 158 164 Data distribution
68. 211 Scott s pl suu 183 201 218 244 245 250 SCR AMES ete ttt 168 172 185 SCPC plots d uei uns seio eet odere es 81 83 194 Script leseeeee ee 167 180 Encrypted script files 0 0 168 Expression Operators and Functions 176 Introduction 0 0 0 eee eee eee 168 Memory and data file variables 174 Navigating in the script window 170 Opening an existing script 169 Recording a script 2 2 0 eee eee eee 171 Running a script 0 2 0 eee eee eee 171 Saving a script 0 0 0 eee ee eee ee 172 Syntax Convention 000 173 SEZ les 1 te cS ghana awe 168 169 172 Seasonality sks ed eh he ke ae edad 146 Security SBS kets Ma n Ao M 28 43 44 Seed value 71 73 182 183 196 218 Sensitivity analysis Lus 141 142 225 Sets of variables lees 22 46 118 Setting global options 00 0 160 iur MC ems ENS 143 226 Single case experimental design 223 251 Skewness 49 69 74 86 100 125 182 193 SNB file 1L tas fa tee ee el 19 211 Sorting 36 237 77 86 97 191 195 200 227 Sound files 2z ze eee h aces 180 213 Split half reliability 0 0 139 221 Spreadsheet 0 0000 39 41 199 DPSS Sees visti dana E hea eee 38 40 199 Standard regression 04 118 119 208 Startup options
69. 9 Navigating in the script window 0 0 0 cece eee tee teens 170 Using the RECORD SCRIPT Feature 0 0 0 cee eee eee 171 Running script sce RES SERERE E ERR Odes Bin hE Sin Re d 171 Saving Script files ome d Baked oa ok ete da ee Se eed ee ee 172 9 Script language reference soui ea u 00 0 a a eee ee een e 173 Syntax Convention mse eea e eria a E e aiia aa E D e E hn 173 Using memory and data file variables 0 0 cee eee ee eee 174 Expression Operators and Functions 176 One line descriptions of commands 0 0 0 cee eee eee eee 178 Commands description 1 0 0 0 eect cnet e eens 181 10 Customizing the tools menu lsllleleee II 233 11 Setting the program preferences 0 0 0 ee eee ee e 235 APPENDIX A xBase syntax and functions 0 0 ce eee eee ee eee 240 APPENDIX B References eener coreto cece eect he 249 APPENDIX C Limitations 2 25244 s454i44a Sot e3ma Set haa NUS ema XI S RA 253 APPENDIX D Technical support lsleeeeeee II 254 INTRODUCTION 1 INTRODUCTION Welcome to SIMSTAT for Windows This program is designed to provide an easy to use yet powerful statistical package for scientific business and engineering applications as well as for teaching purposes It provides unique features that facilitate data analysis as well as other tasks such as data preparation output management and data presentation SIMSTAT provides a wide range of stat
70. AT for WINDOWS To zoom in on a specific portion of a chart Activate the zooming feature by clicking on the button or selecting the ZOOM IN command from the CHART menu Click with the mouse on the upper left corner of the area you want to display Drag the mouse cursor to the lower left corner of the area Release the mouse button SIMSTAT automatically resets the axis minimum and maximum values as well as the increment value to coordinates near the defined area Those new axis limits are usually saved as soon as you move to another chart or switch to another window To restore the original axis scaling e Select the ZOOM RESET command from the CHART menu to reverts the effect of one or several ZOOM IN commands and restore the original axis scaling of a chart This command should be used immediately after a ZOOM IN command has been performed because any other modification to the chart or any window operation may cause the new axis limits to be saved resulting in the loss of the original scaling information WORKING WITH CHARTS 165 Exporting charts SIMSTAT lets you transfer charts to the clipboard or export them to other formats so that they can be viewed or edited by other applications SIMSTAT currently supports 3 different file formats Windows Metafiles Windows Bitmap and tab separated values files Window metafile format WMF The Windows Metafile format stores graphic images in a vector based format The result
71. ATA FILE OPERATIONS 39 ASCII data files SIMSTAT will read up to 500 numeric and alphanumeric variables from a plain ASCII file text file The file must have the following format Every line must end with a carriage return The first line must include the variable names separated by spaces tabs and or commas Variable names may have a length of not more than 10 characters Longer strings are truncated at 10 characters The remaining lines must include the numeric scores separated by spaces tabs and or commas e Each line must contain scores for one case and variables must be in the same order for all cases e All invalid scores and all blanks encountered between commas or tabs are treated as missing values A single dot can also be use to represent a missing value Comments can be inserted anywhere in the file by putting an asterisk at the beginning of the line e Blank lines can also be inserted anywhere in the file Spreadsheet data files SIMSTAT reads spreadsheet files produced by LOTUS 1 2 3 v1 1 to v5 0 SYMPHONY v1 0 and v1 1 EXCEL v2 2 to v5 0 and QUATTRO PRO v1 0 to v6 0 When a spreadsheet data file is selected the program displays a dialog box where you can specify in the case of multiple page spreadsheets the page where data are located and the range of cells to be read You must specify a valid range name or provide upper left and lower right cells separated by two periods such
72. C file SAV SPSS for Windows file CSV Comma separated values ASCII TAB Tab separated values ASCII Examples IMPORT C LOTUS DATA SURVEY WK3 IMPORT C SPSSWIN DATA IMPACT SAV INPUT Syntax INPUT Svarname AS type expression options Description The INPUT command allows you to get a value from the user and store the result in a user defined string or numeric variable When a value is already stored in the specified variable it will be presented as the default value to the user Use the CLEAR option to erase this value The LEN option can also be used to increase or decrease the maximum length of input By default the dialog box is positioned in the center of the screen However the LEFT and TOP options can be used to specify the position of the box s upper left corner The parameter for these two options is an integer value between 0 a 100 expressing a percentage of the screen height or width If the variable specified is a numeric variable you can restrict the valid range of responses by using the MIN and or the MAX options By 200 SIMSTAT for WINDOWS default the number of decimals displayed is set to 0 and user s input is restricted to integer values The DEC option can be used to alter the number of decimal places to display Options color Text color see below LEFT integer Vertical position of the box s upper left corner 0 to 100 TOP integer Horizontal p
73. CALL command If no extension is provided the program will successively look for an existing file name with an SCR and an SCZ extension If no path information is provided SIMSTAT will first look in the same directory as the calling script file and then in the default script directory Example CALL C DEMO LESSON1 SCR 186 SIMSTAT for WINDOWS CHART Syntax CHART options Description The CHART command allows you to modify various properties of the currently displayed chart or navigate through the charts either to display another chart on the screen or modify some features of a chart currently not displayed Options FIRST LAST NEX PRIOR 3D ON OFF GRIDX ON OFF GRIDY ON OFF GRIDY2 ON OFF SCALEX value value SCALEY value value INCX integer INCY integer DECX integer DEXY integer TITLE string LABELX string LABELY string LABELY2 string CHI SQUARE Syntax CHISQUARI Description ove to the first chart ove to the last chart ove to the next chart ove to the previous chart Add or remove 3D perspectiv Horizontal grid Vertical grid Second vertical grid Horizontal axis limits Vertical axis limits Increment value of the horizontal axis Increment value of the vertical axis Decimal places for values on the X axis Decimal places for values on the Y axis Title string Horizontal axis string Vertical axis string Second vertical axi
74. Extreme Differences Absolute Positive Negative KES A 2 tailed P o 12 99 00000 8220 90 2 803 000 110 SIMSTAT for WINDOWS Kruskal Wallis The KRUSKAL WALLIS one way analysis of variance by ranks is a procedure for testing whether k groups have been drawn from the same population This test is a nonparametric version of the ONEWAY analysis of variance The output displays the number of valid cases the mean rank of the variable in each group chi square and its probability with a correction for ties OPTION Range of x This option requests two values that will be treated as minimum and maximum values of the grouping or independent variable under consideration Each discrete value of the independent variable that falls within this range defines a distinct group Sample output of a Kruskal Wallis one way analysis of variance by ranks KRUSKAL WALLIS 1 WAY ANOVA AGGRESS Nb of agressive behavior by SIBLING Number of siblings Mean Rank Cases 26 45 29 SIBLING 00 cde 24 SIBLING 1 00 45 83 6 SIBLING 2 00 59 Total Corrected for Ties CASES Chi Square Significance Chi Square Significance 59 6 3480 0418 6 4123 0405 STATISTICAL ANALYSIS 111 Listing cases LIST CASES displays a listing ofthe values ofthe selected dependent and independent variables OPTION List all cases Check this box if you want to display all currently selected cases Number of cases When the List All Cases option is di
75. FUNCTION value variable or expression NORMAL x Pseudo random number generated from a normal distribution with mean of O and standard deviation of X SCRIPT LANGUAGE REFERENCE 177 UNIFORM x Pseudo random number generated from a uniform distribution between 0 and X 178 SIMSTAT for WINDOWS One line descriptions of commands You will find below a short description of all available script commands grouped in four broad categories For more detailed information on each of these commands refer to the description available in the next section where commands are presented alphabetically Database commands APPENDFROM COMPUTE DATA ENDDATA FILTER IMPORT OPEN RANK RECODE RECORD SORT VARDEF VLABELS WEIGHT Appends data in a DBF file to the current data file Stores the result of a computation in a new variable or an existing one Creates a temporary data file Filters the records Imports data from another file format Opens a data file a notebook file or a chart file Transforms the values of a variable into ranks Recodes the numeric values of a variable Moves within the records of the data file or adds a new record Sorts records on one or several variables Defines variables label missing values display width amp decimals etc Defines value labels Weights records using a variable Flow control commands CALL GOTO GOSUB IF THEN ELSE DEFINE or DIM DELAY LET QUIT RU
76. LS DEFINITION Often several variables share the same valuelabels For example some questionnaires may use a common ordinal scale for several questions Rather than retyping the same value labels over and over again SIMSTAT enables the user to establish a link between a variable without value labels and an existing one which already contains labels To establish such a link click on the down arrow button of the Linked to list box and select from the list of variables the one containing the value labels you want to use These labels will appear in the value labels list box Once a link has been established every modification made to value labels list will affect the labels associated with the original variable as well as all other variables currently linked to this variable US yo may also use this feature to copy value labels from one variable to another To do this follow the previous instructions to link the current variable to the one from which you want to copy labels The labels should appear in the value labels list Then remove this link by setting the linked to option to none DATA FILE OPERATIONS 31 Entering and Editing Data Data can be entered and edited directly within SIMSTAT with the use of a data sheet Here are some quick descriptions of the steps needed to perform specific data editing tasks To enter or edit a value Type in the value Press Tab to move one cell to the right Press Enter to move one
77. M SOS E E EE RE 9 Status Dari icu teuer stola eee ose a ectetuer desee err erbe 11 Getting help sisi s ett et ne ng os Be Re ne nee ce B6 11 2 Tutorial Performing statistical analyses 2 2 eee ee eee 13 Step 1 Opening a data file for analysis 0 0 0 eee ee eee eee eee 13 Step 2 Assigning variables 0 0 llle 13 Step 3 Performing frequency analyses 0 cee eee eee eee eee 15 Step 4 Viewing the results 0 eect ene 16 Step 5 Selecting a subset of cases 0 cece ene 17 Step 6 Performing regression analyses 18 Step 7 Keeping hard copies of results 19 Step 8 Exiting SIMSTAT 0 araa NA ee ee eee nee 19 3 Data Hleoperations eh de og hod beg VE Shen eed bane yd Foo E edly 20 Opening an existing data file ems 0 2 eee eee nee nee 21 SIMSTAT for Windows data files 0 0 eee eee 22 Navigating in the Data Window 0 eee eee eh 23 Creating a new data file 0 cee tenets 24 Dbehininig variables 25 ote o Rs S M ste mustang Ret e P ende E 27 Entering and Editing Data 2 cee ene 31 Filtering records i i5 iu ret a a ae EA 33 Sorting Tec dS zs neuer be Oe TA RES e TAGE ERIS SH Site Slee eee 36 Importing data from other applications 0 0 0 eee eee eee eee 38 Exporting data to other applications 0 0 0 cece eee eee 40 Merging data files cererea ea ea e ak wee ta es eee pee ES bh 42 Limiting access to data files lsieeeeee
78. Median 49 69 74 86 182 183 193 Median test leere 116 206 Running median 224 228 144 Split median trend 224 144 Median test 0 ccc cece eee 116 206 Memory and data file variables 174 Merging data files 0 2 0 eee eee eee 42 Metafiles 2 2 iki etr eter odere dre kiei 6 165 Missing values 22 28 53 57 75 MEDICORUM vated enh atch 127 190 201 210 219 Modifying specific chart options 160 Moses test oe BIER eis 117 207 Movie file 0 0 0 ee 180 213 Moving average 145 147 224 228 Multiple comparison procedures 130 211 250 Multiple regression 90 118 124 196 197 204 208 References 29 3 9 n RU eter een 250 Multiple responses analysis 125 207 N Navigating in the data windows 0000s 23 in the notebook 0 000 cee 61 in the chart window 0000005 155 in the script window 000 00s 170 Newman Keuls 000000005 130 211 Nonlinear regression 20 0000 133 220 Nonparametric tests Binomial test 0 0000 67 181 Friedman test 0 0 00 cee ee eee 89 195 Kolmogorov Smirnov test 108 109 202 203 Kruskal Wallis command 110 202 Mann Whitney test 114 152 205 McNemar test lesen 115 205 Median test
79. N STOP Runs another script program and returns Branches to another part of the program Branches to and returns from a subroutine Carries out a command based on a specified condition otherwise performs another command Defines a new memory variable Temporarily stops the program execution for a specific length of time Assigns a value or the result of an expression to a variable Quits SIMSTAT Runs an external DOS or Windows program Stops the script and returns to the SIMSTAT menu Interactive commands BOX ENDBOX BEEP INPUT MENU END QBOX QUESTION Displays a dialog box with textual information Generates a beep sound Gets a numeric or a string value from the user Creates and displays a bouncing bar menu Displays a dialog box with textual information Displays a multiple items question dialog box SCRIPT LANGUAGE REFERENCE 179 Statistical analysis commands BINOMIAL BOOTSTRAPI BOOTSTRAP2 BREAKDOWN CHISQUARE CLUSTER CORANAL CORRELATION CROSSTAB DESC DIVERSITY FACTOR FULL FREQUENCY FRIEDMAN GLMANOVA INTERRATERS ITEM KRUSKAL KSI KS2 LIST LOGISTIC MANN MCNEMAR MEDIAN MOSES MRESPONSE MULTREG NPAR ONEWAY PCA PCO REGRESSION RELIABILITY SCED RANDOM RUNS SENSITIVITY SIGN T TEST TIME SERIES WILCOXON Binomial test Univariate bootstrap analysis Bivariate bootstrap analysis Breakdown analysis Chi square one sample test Cluster analysis requires MVSP v2 2 Corres
80. NS Type of Analysis Sets the type of factor analysis to use The available types are Principal Components and Image Covariance Varimax Rotation Set this option if you would like to rotate your factors using varimax orthogonal rotation Number of Factors Sets the maximum number of factors to retain in subsequent analyses i e rotation and scoring The default is the number of variables in the data set Minimum Eigenvalue Sets the minimum permissible eigenvalue for a factor to be retained in subsequent analyses 1 e rotation and scoring The number of factors to retain and minimum eigenvalue criteria are evaluated in an either or fashion that is if either criterion is met the number of factors retained is cut off at that point Descriptive statistics If you would like means and standard deviations for your selected variables in the output enable this option Note that the standard deviations printed by EFA are unadjusted i e the sum of the squared deviations is divided by n rather than n 1 To obtain adjusted standard deviations or more detailed descriptive statistics see the DESCRIPTIVES or the FREQUENCY commands Correlation matrix When toggled on the correlation matrix for the variables in the data set will be included in the output Scree plot If you would like a scree plot of the eigenvalues in the output select this option 82 SIMSTAT for WINDOWS Unrotated factor solution When enabled the results
81. ONS 21 Opening an existing data file To open an existing data file select the DATA OPEN command from the FILE menu This opens an open file dialog box as shown below Open 21 x File name Folders c simstatwidata Cancel cars dbf g c sample dbf Cy simstatw Network Tir amp y data Read only List files of type Drives xBase DBF amp c disk1l voll J When this dialog box is evoked the program points to the default data directory and displays in the File Name list box all available data files in this directory To open a file double click on its name or select it and then click on the OK button If the name of the data file you want to open is not displayed type the filename in the File Name box including drive and path if necessary and select the OK button You may also use the following components to locate the data file f the file is on a different disk click on the down arrow of the Drives list box to display available drives and select the disk where the file is located Ifthe file is in a different directory double click on the directory names in the Directory or Folder to move through the directory tree Ifthe file name is displayed in the File Name list box double click on it to open the file or select it and click on the OK button If you want to open a data file used previously click on the down arrow button at the right side of the File Name edit box and sel
82. Property dialog box This allows you to adjust the viewing angles object depth and shadows of the 3D charts WORKING WITH CHARTS 163 BD View Properties X Angle Bl Iv 3D v Full 3D View Y Angle 25 IV Shadows Apply Cancel To adjust the viewing angles Check the Full 3D View box Drag the marbles to the desired angles or type the desired angles in the fields Verify the desired view shown in the sample rotation frame Click on the Apply button to modify the angles in the background chart while remaining in the rotation dialog or click on the OK button to apply these angles and exit the rotation dialog box To control the 3D depth You can also control the depth of the 3D chart with the sliding control located underneath the rotation preview When you drag the sliding control to the right you will increase the depth in comparison to the chart width Sliding it to the left will decrease the chart depth To alter the 3D shadows In this dialog box you ll find the shadows check box which will allow you to turn the shadow for the markers in 3D mode on and off Turning it off will cause the back side of the markers to paint in the same color assigned to the front of the marker Zooming in and out Sometimes you may want to zoom on a chart to view it in more detail While it is possible to do this by adjusting the axis scaling SIMSTAT provides a more convenient method using the mouse 164 SIMST
83. S The data window is a spreadsheet style data editor where values can be entered browsed and edited Each column contains data for one variable and each row contains the responses for a single case You can have only one Data window open at a time The associated DATA pull down menu offers a wide range of procedures to select a subsample of cases sort these cases transform existing variables or create new variables resulting from mathematical operations performed on existing variables DATA D XCARS DBF co IF chevrolet chevelle malibu buick skylark 320 plymouth satellite p amc rebel sst E ford torino ford galaxie 500 a chevrolet impala P plymouth fury iii al pontiac catalina amc ambassador dpl citroen ds21 pallas chevrolet chevelle concours sw il ford torino sw 0o 0o A COO CO CO CO CO cO CO co co This section introduces you to various tasks that may be performed on data files such as how to Open an existing data file Create a new data file Set the variable appearance and definition display width missing values value labels etc Enter and edit values in the data grid e Filter records e Sort the data grid on one or several variables Import from and export to other applications e Save and restore archived copies Restrict the default access to the data file Information on data transformation techniques will be introduced in the next chapter DATA FILE OPERATI
84. SIMSTAT for Windows User s Guide Designed and written by Normand P ladeau Provalis Research DISCLAIMER This software and the disk on which it is contained are licensed to you for your own use This is copyrighted software owned by Normand P ladeau By purchasing this software you are not obtaining title to the software or any copyright rights You may not sublicense rent lease convey modify translate convert to another programming language decompile or disassemble the software for any purpose You may make as many copies of this software as you need for backup purposes You may use this software on more than one computer provided there is no chance it will be used simultaneously on more than one computer If you need to use the software on more than one computer simultaneously please contact us for information about site licenses WARRANTY The SIMSTAT product is licensed as is without any warranty of merchantability or fitness for a particular purpose performance or otherwise All warranties are expressly disclaimed By using the SIMSTAT product you agree that neither Normand P ladeau nor anyone else who has been involved in the creation production or delivery of this software shall be liable to you or any third party for any use of or inability to use or performance of this product or for any indirect consequential or incidental damages whatsoever whether based on contract tort or otherwise even
85. SIS 149 T test analysis T TEST calculates either independent sample t tests or paired sample t tests to determine whether two sample means are significantly different The paired sample or correlated t test compares the means between each pair of variables assigned as dependent and independent The independent t test compares means on the dependent variable for two groups defined by values of the independent variable SIMSTAT provides two distinct tests to take into account whether the two populations from which the samples are drawn have equal or unequal variances You can also specify whether the null hypothesis should be evaluated using a one tailed directional or a two tailed non directional test OPTIONS ANALYSIS TAB Type of design This option determines whether the design includes paired correlated or independent samples Values of x For independent samples the program requires you to specify two values of the grouping or independent variable that will be used to define the two groups Direction The Direction of the statistical test can be one tailed directional or two tailed non directional Confidence interval The interval options allows you to set the width of the confidence intervals around the two measures of effect size The interval width is expressed as a percentage between and 99 This value is also used when displaying barcharts with a confidence interval or an error bar graph representing a c
86. TAT closes the existing data file and opens this new data file When an SNB extension is used the program clears the current notebook and loads the specified notebook file If the file extension is CHX then SIMSTAT clears the content of the Chart window and loads the specified chart file Examples OPEN C NDATANCARS DBE OPEN C NOU PUT R EPORT SNB open the cars dbf data file open the report snb notebook file 212 SIMSTAT for WINDOWS PCA Syntax PCA varlist options Description This procedure performs a R mode principal component analysis The component loadings are scaled to unity so that the sum of squares of an eigenvector equals 1 and the component scores are scaled so that the sum of squares equals the eigenvalue Q mode PCA will generally have the opposite scaling Requires MVSP v2 2 Options Select a transformation LOG10 Log base 10 LOGE Log base e LOG2 Log base 2 SORT Square root RATIO Logratio TRANSPOSE Transpose data CENTER Centered data STANDARDIZE Standardize data Minimum eigenvalu default all KAISER Kaiser s rule JOLLIFFE Jolliffe s rule MINEIGEN real Specify a minimum eigenvalue ACCURACY real Accuracy of solution Output of graphics PLOT Text scattergrams GPLOT Graphic scattergrams PCO Syntax PCO varlist options Description Principal coordinates analysis PCO is a generalized form of PCA Whereas PCA impl
87. V Times series of AGGRESS Times series of AGGRESS moving average Single case experimental design Error rate curves of AGGRESS The list box displays the title of all charts in the Chart window e Select the chart you want to move by clicking on its title Click on the up or down arrow buttons to move the chart up or down the list You may also drag the chart to its new location by keeping the mouse button down and dragging the chart title to its new location WORKING WITH CHARTS 157 TO WRITE THE CHARTS TO DISK Select the CHART SAVE command from the FILE menu or click on the button while the Chart window is active f the charts have never been saved before a File Save dialog box will appear Enter the name of the file under which you want to save the charts and press Enter or click on the OK button By default SIMSTAT uses the CHX extension for chart files If no extension is given the program automatically adds this extension to the end of the file name ES To save the charts under a different file name choose the CHART SAVE AS command and provide a new file name TO RETRIEVE CHARTS STORED ON DISK Select the CHART OPEN command from the FILE menu or click on the es button while the Chart window is active ES Rather than creating a new chart file you may want to add new charts to an existing chart file To do this open the existing chart file where you want the new charts to be pla
88. a data file with restricted access Choose the DATA SECURITY command from the FILE menu and enter the proper password to display the Security dialog box Enable the options you need and click on the OK button to exit the dialog box CAUTION All options modified when regaining full access become the new default access values You should take care to reset these values before closing the data file if you want to maintain this default limited access to the data file DATA FILE OPERATIONS 45 Archival backup of data files During a research project data files are frequently modified cleaning recoding mathematical transformation etc At some point you may need to return to a previous version of an existing data file to recover lost variables or cases that have been transformed or deleted or just to make routine verifications In order to do this several successive backup copies of this data file should be created and kept Another reason to make backup copies of data files is to prevent the accidental loss of an entire data file caused by a hardware failure or software malfunction SIMSTAT provides asimple archiving procedure that allows one to quickly create backup copies This procedure stores a copy of the currently active data file and all its related files structure definition value labels variable sets etc in a single compressed file SIMSTAT uses the industry standard Zip format as its own archive file format allowing you
89. able discrimination between nearly all ability levels If the average item total correlation is higher a wider range of difficulty levels is required When selecting items according to their difficulty level two additional factors should be considered While a mean difficulty level of 5096 will usually maximize the total score variance this value may need to be increased to take into account the fact that with multiple choice items a specific proportion of correct responses may be obtained simply by guessing Also the choice of an average difficulty level should take into account the fact that the scale may be constructed to differentiate between those at the lower or the higher ends of the scale In such situations 102 SIMSTAT for WINDOWS better discriminatory value will be achieved by selecting items with difficulty levels near the desired end However the percentage of correct responses is not a sufficient condition to judge the quality of an item since the number of correct responses should also be related to the level of the ability we want to measure If those who answer correctly are low on the measured ability while the subjects with higher scores choose a wrong answer or if the success on this item is the same for these two groups then there may be a problem in the item formulation or the item may be measuring a different ability unrelated to the one of interest Several methods exist to assess the relationship between level of
90. able specified on the first line of the command When a data file variable is specified the answer is automatically stored in the current record of the data file If a numeric value is already stored in variable it is used as the default answer The CLEAR option can be used to automatically erase this value before the dialog box is displayed If an answer is required and should not be skipped use the NOSKIP option to prevent the user from exiting the dialog box without providing a valid answer Options color Color of the text TOP integer Vertical position of the dialog box s upper left corner 0 to 100 LEFT integer Horizontal position of the dialog box s upper left corner 0 to 100 BEEP Sounds a beep using computer s speaker NOBUTTON Hide the OK button CLEAR Clear the content of the variable MOSKIP Prevent exiting the dialog box without providing a valid answer SCRIPT LANGUAGE REFERENCE 217 Example QUESTION SANSWER1 BLUE NOSKIP In the following expression P 0 05 What does the probability value stands for ANSWERS amp A The probability that the null hypothesis is true amp B The probability that the null hypothesis is false amp C The probability of the data given the null hypothesis amp D The probability of the null hypothesis given the data amp E None of the above END QUIT Syntax QUIT Description This command takes you out of the SIMSTAT program All opened file
91. ained and residual values for all cases This option is useful for identifying outliers 1 e cases that are not well represented by the regression model The Outliers option allows you to restrict the residual caseplot to those cases for which the absolute standardized value is greater than or equal to the specified value The Outliers value can be set between 0 and 4 standard deviations Durbin Watson statistic This option tests for the presence of autocorrelation or serial correlation in the residuals The larger the autocorrelation the less reliable the results of the analysis Residual scatterplot This procedure produces a bivariate scatterplot The predicted value is plotted along the horizontal axis and the standardized residuals on the vertical axis Residual normal plot This option displays a normal probability plot of residual values that allows you to evaluate whether those residuals are normally distributed If the residuals follow a normal distribution the data points will fall approximately along a straight line going from the lower left corner of the graph to the upper right corner Save residuals This option instructs SIMSTAT to save the predicted and residual values as new variables in the current data file Creating those variables allows you to perform further analyses such as displaying scatterplots between each independent STATISTICAL ANALYSIS 93 variable and the predicted and residual values The new variabl
92. al scores more closely Save scale totals This option allows one to save the total score computed for each subject in an ASCII data file This data file can then be opened by SIMSTAT to perform further analysis Sample output of an item analysis ITEM ANALYSIS Point Bis Alpha if US CAG Value Freq Percent Biserial Correl Delete Hi Low ARE Tos 259 65 43 34 674 62 220 10 3 SU ui Smo I 4 edm O 4 0 114 29 S cue V2 T0 63 16 ME e ev 43 4 AL es Ee 3 1 SS edu Scope Een 79 34 24 686 40 V3 T0 Tu 4 52 ede 2 274 69 be 48 2653 72 SQ 22 6 UNI is SOS 4 0 89 22 2s s V4 IPSO 14 4 SUPE cupid T zu 3 1 TOL ros ae 378 95 706 42 hdd 22 4 0 5 1 SE 421 MS 2m es 364 91 47 E 253 24 aQ SL 8 SEE Seu 4 0 5 1 ror Seo V6 DoD 8 2 LIE DO 29 2 1 ROG TON EU 15 0 SES dio Bg 389 97 LA MNT TEON 4 106 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 107 ITEM TOTAL RESPONSE BREAKDOWN 25 of respondents in each group C25 GSO CUD COO HI LOW vel 321 u3g 80 94 628 V2 57 17 86 97 40 V3 26 64 85 98 72 V4 os EDD RUD SEDIS cee V5 75 93 96 99 24 V6 96 96 96 100 4 SCALE STATISTICS Total score Percentage Biserial Item Total Mean 14 76 82 0 45 eee Median dU 85 8 45 Pee Minimum 6 00 48 0 ep zs Maximum 18 00 93 3 986 48 Variance EOZ 20era 04 mri Std Dev Bess Il 14 4 led S12 Skewness Sg AS S E Skewness Elegie Kuriosne 104 berks Rurtosts 245 Nb of c
93. alse Pct False Pos Neg Point Negatives Negatives Positives Positives Results Ratio 100 0 10 100 0 1 4 8 SES 2L 510 DIS g 90 0 al 4 8 S216 Ss Aft 18 8 0 7 70 0 1 4 8 25 8 14 7 pO 4 40 0 3 14 3 22 6 Deals 6 0 2 20 0 5 23 8 22 6 0 8 DEO al 10 0 9 42 9 Jo O52 4 0 0 0 0 15 71 4 48 4 ORO Sr 0 0 0 qo 90 5 61 3 0 0 2 0 0 0 0 2L 100 0 67 7 0 0 STATISTICAL ANALYSIS 143 Sign test The SIGN TEST tests the hypothesis that two variables have the same distribution This is assessed by comparison of the numbers of positive and negative differences between values of the two variables OPTIONS Direction This option allows you to select either a one tailed directional or two tailed non directional probability test Sample output of a sign test SIGN TEST T2DEPRES With TIDEPRES Cases 2DE eS GIDDIDRYEJESEU S ihc DEERE S IERDE oT DEBRE S je Aral IRYAS 2E TES 40 Total Z 2 4333 l tailed P DS 144 SIMSTAT for WINDOWS Single case design Single case experimental design was originally developed for the study of animal and human behaviors in the context of controlled laboratory experiments This methodology is now currently used in applied research such as the evaluation of behavior modification programs or other types of clinical interventions the effects of pharmacological agents on behaviors or the impact of educational and social intervention programs It involves repeate
94. an rank of the variable in each group chi square and its probability with a correction for ties Options VALUES integer integer Range of X PANEL Show dialog box KS1 Syntax KS1 varlist options Description The KS1 command Kolmogorov Smirnov one sample test compares the distribution of each variable against a standard normal distribution or a uniform distribution It tests whether the sample data can reasonably be thought to have come from a population having this theoretical distribution Options Type of distribution NORMAL Normal distribution UNIFORM Uniform distribution VALUES real real Mean and standard deviation of a normal distribution or minimum and maximum of a uniform distribution PANEL Display the dialog box SCRIPT LANGUAGE REFERENCE 203 KS2 Syntax KS2 varlist BY varlist options Description The KS2 command Kolmogorov Smirnov two sample test evaluates whether a dependent variable Y has the same distribution in two independent samples as defined by a grouping variable X This test is sensitive to differences in the shape location and scale of the two sample distributions Options VALUES integer integer Values of independent variable PANEL Display the dialog box LET Syntax LET Svariable Svariable value or expression Description The LET command assigns the value of the expression on the right side of the assignment operator to the var
95. and Ys can be used to split the various items into two halves The output contains summary statistics for both scales mean variance and standard deviation the correlation between those scales the Spearman Brown coefficient and the Guttman split half coefficient which does not assume that the two parts have the same variance Internal consistency This option allows the computation of Cronbach s alpha internal consistency statistics It has been shown to be equivalent to the average of all possible split half coefficients The output also includes the mean inter item correlation and the standardized item Alpha the alpha that would have been obtained if all the items had been standardized Sample output of reliability analysis RELIABILITY ANALYSIS DEP1 Depression scale item 1 DEP2 Depression scale item 42 DEP3 Depression scale item 3 DEP4 Depression scale item 4 DERS Depression scale item 5 DEP6 Depression scale item 6 DEP7 Depression scale item 7 DEP 8 Depression scale item 8 138 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 139 140 SIMSTAT for WINDOWS Runs test The RUNS TEST is a procedure that can be used to test whether the ordered sequence in which observations were obtained is random In order to be carried out this test requires that all values be dichotomized The dialog box allows you to separate observations in two distinct groups using the mean the median or a user specified value as a
96. and restores the previous value 8 SIMSTAT for WINDOWS List boxes A list box allows you to select a string from a list of valid possibilities A list box can contain a vertical scroll bar To make a selection scroll if necessary then click the item you want Drop down list boxes A drop down list box is similar to a list box but its items are hidden from view To select an item from the list click on the downward pointing arrow on the right side of the box scroll if necessary then click the item you want Check boxes A check box is a small square box with a text description to its right Check boxes operate independently of one another You can turn an option ON or OFF by clicking in the box If an option with a check box is turned on an X appears in the box Radio buttons A radio button is a small round button Radio buttons are used in groups to present mutually exclusive options Click the button to turn the option on to turn it off select a different radio button Ifa radio button option is active it contains a dark circle Spin buttons Sometimes an edit box will be presented with spin buttons to its right This indicates that a numeric value is expected You can type this numeric value using the keyboard or use the spin buttons to quickly increment or decrement the value shown in the edit box Multiple page dialog boxes Some dialog boxes may contains a fairly large number of options In order to facilitate
97. anywhere on the line where you want SIMSTAT to begin the execution of commands e Click on the button or select the RUN FROM CURSOR command from the SCRIPT menu 172 SIMSTAT for WINDOWS Saving script files To save a script file choose the SCRIPT SAVE command from the FILE menu A file save dialog box appears By default SIMSTAT uses the SCR extension for script files If no extension is given the program automatically adds this extension to the end of the file name To save the contents of the script window in an encrypted and compressed file enter a file name with a SCZ extension or select the Encrypted Script SCZ option from the Save File as Type list box and enter a valid file name without an extension SCRIPT LANGUAGE REFERENCE 173 9 SCRIPT LANGUAGE REFERENCE Syntax Convention This section outlines the syntax conventions of the various commands and options Unless specified otherwise you can type commands and options in either uppercase or lowercase letters You will find below a short description of those elements UPPERCASE Items in capital letters are keywords Keywords are a required part of the statement syntax unless they are enclosed in brackets or specified as optional lowercase italic Items in lowercase italic characters are placeholders for information you must supply in the statement Several types of information can be required such as variable a single variable name varlist One or several
98. aracter N Numeric D Date L Logical M Memo DATA FILE OPERATIONS 25 Description Character variables can contain up to 254 alphanumeric characters Numeric variables can contain floating point numbers The date type occupies 8 spaces and holds a year month and day The logical type uses a single space to store a boolean value that can be either true or false or Yes or No The memo type can contain up to 32K of text ES While SIMSTAT for Windows can perform some analyses such as frequency or crosstabulation on character variables it is often preferable to use numeric variables especially when the number of different values of this variable is limited For example rather than storing the sex of the respondent as a character variable and using male and female or F and M it is advisable to assign numeric values to this information for example 1 for male and 2 for female To facilitate theinterpretation ofthese numeric values SIMSTAT provides a way of associating an alphanumeric description of up to 60 characters with each numeric value of a variable see page 28 Length The variable length specifies the maximum number of characters that can be stored in the variable Variable lengths for date and logical field are automatically set to 8 and 1 respectively The maximum length for a character variable is 254 while the maximum length for a numeric variable is 19 Decimal The decimal column lets yo
99. ariables and then click on the button Click on the OK button to proceed to the deletion of all variables listed in the Variables To Delete list box DATA FILE OPERATIONS 33 Filtering records The FILTER RECORDS command temporarily selects cases according to some logical condition You can use this command to restrict your analysis to a subsample of cases or to temporarily exclude some subjects You may also use this feature to perform data transformations on a subsample of cases The filtering condition may consist of a simple expression or include many expressions related by logical operators i e AND OR NOT The condition expression should be a valid xBase expression evaluated as true or false and may not exceed 240 characters To obtain more instructions on expression operators evaluation rules and supported xBase functions see Appendix A When selecting the FILTER RECORDS command a filter builder dialog box is displayed see below You can directly type the filtering expression in the Filter edit box using the proper syntax or use any elements displayed on the upper part of the Filter Records dialog box to build a valid expression ilter records x HAAL 7 Iam B AGGRESS sw ATA EEE HOURSTV PHA ee GIGIE PHASE OR SEX um wj C JOG ALLTRIM string FS AT searchstring targetstring CTOD string DATE DAY datefield F Sorted DELETED zl Filter Ei
100. art Pareto chart Histogram Box amp whisker plot Cumulative distribution chart ogive Normal probability plot Barchart 2 dimensions Multiple box amp whisker plot Dual histogram Error bars Mean bars Deviation chart Scatterplot Residual caseplot Residual scatterplot Time series plot Autocorrelation plot ACF amp PACF Interrupted time series or single case experimental design ROC curve Error rate graph COMMANDS Descriptive Frequency Descriptive Frequency Descriptive Frequency Descriptive Frequency Descriptive Frequency Descriptive Frequency Descriptive Frequency Tables Crosstab Compare Breakdown Compare T test Compare T test Compare Oneway Compare T test Compare Oneway Compare Oneway Regression Regression Compare GLM Anova Ancova Regression Regression Regression Multiple Regression Compare GLM Anova Ancova Regression Regression Regression Multiple Regression Time series Time series Time series Time series Time series Single case Scale Sensitivity Scale Sensitivity WORKING WITH CHARTS 155 Navigating in the chart window All high resolution charts created during a session are displayed in the Chart window This window can be used to view the charts and perform various operations on individual charts or on the entire collection of charts For example you can modify the various chart attributes save those charts to
101. as A1 H20 If you set the Range Name list box to ALL the program attempts to read the whole file The selected range must be formatted such that the columns of the spreadsheet represent variables while the rows represent cases or records Also the first row should preferably contain the variable names while the remaining rows hold the data one case per row Cells in the first row of the selected range are treated as field names SIMSTAT will automatically determine the most appropriate format based on the data it finds in the worksheet columns If no variable name is encountered SIMSTAT will automatically provide one for each column in the defined range When reading the data for analysis all blank cells or cells that do not correspond to the variable type e g alphanumeric entries under a numeric variable or a numeric value under a string variable are treated as a missing values 40 SIMSTAT for WINDOWS Exporting data to other applications SIMSTAT allows you to export data to other applications including SIMSTAT for DOS SPSS PC DBASE PARADOX v3 0 v5 0 LOTUS 123 v1 0 v5 0 SYMPHONY v1 0 v1 1 EXCEL v2 1 v5 0 QUATTRO PRO v1 0 v6 0 COMMA SEPARATED VALUES Windows and DOS TAB SEPARATED VALUES Windows and DOS When exporting a data file SIMSTAT will use the current filtering condition to determine which records will be exported The active sorting order will also be used to control the record sequ
102. ases 400 Nb of items ie Cronbach s Alpha 694 Ferguson s Delta IAS TOTAL SCORE FREQUENCY DISTRIBUTION Cumul Cumul Value Frequency Percent Frequency Percent 6 2 35 2 5 7 1 33 3 8 I 8 6 1S 9 23 a 9 4 TO 13 EE i 10 iS 48 28 3 0 EE iil 20 EPIO 48 12 0 Eu Ip 28 3i c0 76 19 0 its 41 TONS E 29 3 EE 14 47 ig 164 Mil 0 oe 15 54 SD 218 54 5 ISIN 16 61 i555 279 69 8 Po 17 60 15 0 339 84 8 E EE 18 61 ies 400 100 0 5s TOTAL 400 100 0 400 100 0 0 100 108 SIMSTAT for WINDOWS Kolmogorov Smirnov 1 Sample Test The KOLMOGOROV SMIRNOV one sample test compares the distribution of each variable against a standard normal distribution or a uniform distribution It tests whether the sample data can reasonably be thought to have come from a population having this distribution The output displays the largest absolute positive and negative differences between the two distributions Kolmogorov Smirnov s Z and the two tailed probability test OPTIONS Distribution This option allows you to choose between distribution types normal or uniform Mean Standard Deviation These options allow you to specify the mean and the standard deviation of the hypothetical normal distribution If no value is given the observed mean and standard deviation are used Minimum Maximum These options allow you to specify the minimum and the maximum values of the hypothetical uniform distribution I
103. ate a specific string on the current page or in the entire notebook by choosing the FIND command in the EDIT menu or by pressing lt Ctrl F gt You can also use the page flip icons located at the bottom of the notebook to move within the notebook pages Left click on the s icon to move to the next page or click on the right button to move to the previous page Left click on the icon to move to the previous page Pressing with the right button moves you to the previous page To add an empty page Empty pages can be added in the notebook Those pages may be used to sketch an analysis plan write down some comments or interpretations of analysis results or simply put some reminders of things to do To add an empty page to your notebook e Make the notebook the active window Select the page after which you want the empty page to appear e Select the PAGES NEW command from the EDIT menu or click on the button If the current page is the first page of a section a dialog box will appear giving you a choice to enter the new page before or after the current page 62 SIMSTAT for WINDOWS To erase existing pages e Select the PAGES DELETE command from the EDIT menu Specify the range of pages you want to delete and select OK to proceed Note You may also use the notebook index to delete single pages or to erase an entire section Using the notebook index SIMSTAT automatically creates an index of all analysis output store
104. ays a matrix for various measures of association and concordance between two variables OPTIONS Estimator The ESTIMATOR option allows you to choose from a drop down list which measures of association are to be computed and displayed in an association matrix SIMSTAT currently supports the following measures Kendall s tau a Kendall s tau b Kendall Stuart s tau c e Somers d symmetric e Somers dyy asymmetric e Somers dy asymmetric Goodman Kruskal s gamma e Spearman s r e Pearson s r Type of matrix When set to X vs Y the association matrix displays relationships between all variables assigned as independent against all variables assigned as dependent The square matrix option produces a matrix displaying relationships between all selected variables without taking into account whether they were selected as dependent or independent Display probability value When this option is enabled SIMSTAT outputs an association matrix with counts and exact probabilities for each coefficient When disabled asterisks are used to show the probability level attained The standard matrix includes the chosen coefficient with up to 3 asterisks corresponding to the significance level attained Significance test This option allows you to specify whether the probability of the coefficients is based on a one tailed directional or two tailed non directional test Missing values This option allows you to specify wheth
105. be composed of a vertical line going from the lower left to the upper left point and a horizontal line going from the upper left to the upper right corner The higher the sensitivity and specificity of a test at each cutoff point the closest the curve will be to the upper left corner of the graph and the greater the area under the curve The overall performance of a test can be quantified by computing the area under the ROC curve AUC A perfect test would yield an AUC of 1 0 while a useless test would yield an AUC of 0 5 Comparisons can also be made among alternative tests by comparing their ROC curves or their AUC values Various methods have been proposed for computing the area under the ROC curve SIMSTAT provides two such methods The first assumes that the categorical scale of the test results from anunderlying continuous variable The second method adopts a nonparametric strategy that does not make the assumption of a bivariate normal distribution This measure of AUC is obtained by computing the area underneath the straight lines that connect the various observed points of the curves by using a trapezoidal method OPTIONS Value This option allows you to specify which value of the criterion variable will be used to identify positive diagnosis 142 SIMSTAT for WINDOWS Scale orientation This option allows you to specify whether the scale is positively or negatively related to the presence of the condition disease You must specify
106. ble becomes very low Hierarchical Multiple Regression The hierarchical multiple regression model allows you to specify the order in which the independent variables will be entered in the equation The variables can be enter one variable at a time or in sets of variables At each step information can be generated about the degree to which this new variable or set of variables adds to the prediction of the dependent variable The order of entry of variables is chosen according to theoretical or logical considerations such as the theoretical importance causal precedence or elimination of nuisance STATISTICAL ANALYSIS 119 Statistical Regression In the statistical regression model the independent variables are entered in the equation according solely to some statistical criteria obtained from the current sample SIMSTAT provides three versions of statistical regression forward selection backward deletion and stepwise selection When forward selection is used the variable that shows the largest correlation positive or negative with the dependent variable is entered in the equation provided that it meets a specified statistical criterion of significance Then a comparison is made between all the remaining variables to select the one that contributes the most to a significant increase in prediction Forward selection continues until there are no other variables that meet the entry criterion In backward deletion all variables are enter
107. bles could be printed Font size This option allows you to increase or decrease the size of the font used to print the content of the notebook A smaller font size allow you to print more information per page Chart printing The chart printing option allows you to specify how many charts should be printed on a single page When this option is set to 1 or 4 charts per page the printing page orientation is automatically set to landscape while setting the number of charts per page to 2 will force the charts to be printed in portrait mode 238 SIMSTAT for WINDOWS DATA PAGE The Data page allows you to specify information regarding the management of data files confirmation of operations etc The Confirmation section provides check boxes to mark the actions you want SIMSTAT to prompt you to confirm Record deletion Enabling this option causes SIMSTAT to prompt you to confirm the deletion of a record made with the DELETE RECORD command from the DATA menu Variable deletion Enabling this option causes SIMSTAT to prompt you to confirm the deletion of a variable or a variable list specified using the DELETE VARIABLES command Variable transformation Most commands from the Transformation submenu allow you to overwrite values of an existing variable Enabling this option causes SIMSTAT to prompt you to confirm the overwriting of those values Variable creation Some transformations may require the creation of one or several ne
108. ced before creating those charts TO MERGE TWO SEPARATE CHART FILES Open the first chart file containing the charts that should be positioned at the beginning of the new file by using the CHART OPEN command from the FILE menu or by clicking on the button while the Chart window is active e Select the CHART APPEND FROM command from the FILE menu and select the chart file that should be added to the end of the currently opened chart file TO CLEAR THE CHART WINDOW Select the CHART NEW command from the FILE menu or click on the cy button while the Chart window is active If any modification has been made to a chart in the current Chart window or if new charts have been created you will be asked if you would like to save the modifications to disk Select Yes if you want to save those modifications or No to clear the chart window without saving the changes 158 SIMSTAT for WINDOWS Customizing charts After you create one or several charts you can edit their title and axis labels choose a different font or font size adjust the scaling on either axis experiment with different colors or patterns add a legend or make other adjustments to suit your needs The current section presents the various options available for customizing charts Editing titles and axis labels Proper titles and axis labels are of utmost importance when describing the information displayed in a chart By default SIMSTAT will use variable names and
109. cell down To erase a value Position the cursor on the cell you want to erase and press the Del or the lt Backspace gt key To add a row Move to the last row then press the Dn arrow key To delete a record Position the cursor on the record or row you want to delete Choose DELETE RECORD from the DATA menu NOTE You should be aware that the record is not physically removed from the file but is simply tagged as deleted and hidden to the user To permanently remove the deleted records use the DATA PACK FILE command from the FILE menu To add new variables Choose the ADD VARIABLES command from the DATA menu A dialog box will appear that is similar to the one used to create variables for a new data file Specify the name of the new variables and set the variable types sizes and descriptions as described previously on page 24 e Click on the OK button to create these new variables and add them at the end of the current data set 32 SIMSTAT for WINDOWS To delete existing variables Choose the DELETE VARIABLES command from the DATA menu The following dialog box will appear Variable list Variables to delete Sorted Highlight the names of the variables you want to delete and click on the button to move them to the Variables To Delete list box To delete successive variables click on the first variable drag the mouse cursor down the list to highlight multiple v
110. cify the position of the window The window stays on screen until a key is pressed The DELAY option allows you to insert a minimum delay between the display of the box and the input of a valid key The colors of the text appearing in the box can be altered by specifying a color as an option The default colors can also be changed by using the SET COLOR command Options Color Text color DELAY integer Length of delay msec TOP integer Vertical position of the box s upper left corner 0 to 100 LEFT integer Horizontal position of the box s upper left corner 0 to 100 BEEP Sounds a beep using computer s speaker NOBUTTON Hide the OK button Valid Colors 216 SIMSTAT for WINDOWS Black Maroon Green Olive Navy Yellow Purple Teal Gray Silver Red Lime Blue Fuchsia Aqua White QUESTION ANSWERS ENDBOX Syntax QUESTION varname options ANSWERS ENDBOX Description This command displays a dialog box with a multiple choice question The lines between the QUESTION and ANSWERS keywords contain the text of the question while the lines between the ANSWERS and ENDBOX keywords are used to type the available answers Each answer should be entered on a separate line An optional amp character can be inserted in the line to specify an accelerator key that when pressed will select this item The maximum number of items for each question is 30 The answer provided by the user is stored as a number in a vari
111. cludes the proportions for 0 10 0 05 and 0 01 alpha levels The Alpha option lets you specify a fourth Alpha level option only available for the RANDOM SAMPLING command Power estimate The Power Estimate option gives the proportion of samples that have reached a specified alpha level The standard display includes the proportions for 0 10 0 05 and 0 01 alpha levels The Alpha option lets you specify a fourth Alpha level only available for the TWO VARIABLES command STATISTICAL ANALYSIS 73 Full analysis bootstrap analysis The FULL ANALYSIS bootstrap procedure allows one to perform various statistical analyses such as frequency analysis crosstabulation or multiple regression on bootstrap samples The program provides a complete analysis for each bootstrap sample Specific statistics can then be extracted from the listing file with the use of a text editor and then stored in a new data file for further analysis The dialog box allows you to choose which analysis to perform the sample size and the number of samples to be taken from the original sample A second dialog box allows you to control how the analysis is to be performed and what statistics are to be printed OPTIONS Analysis Choosing this option evokes a drop down list from which you can choose a specific analysis to be performed on each bootstrap subsample Same as original When this option is enabled the size of each bootstrap sample drawn from the original data is a
112. ction The first part provides detailed instructions on how to install SIMSTAT on your computer and gives a variety of information about the program Getting started The Getting Started chapter provides basic information about the program s operation Tutorial This chapter includes a short tutorial that will steer you through an easy to follow example The tutorial is designed to quickly familiarize you with the basic operation of the program If you are already familiar with the operation of SIMSTAT you may skip the tutorial section Data file operation This section introduces you to various tasks that may be performed on data files such as how to open an existing data file create a new one import a file from another application filter records or sort them on one or several variables This section also presents instructions on how to perform basic data entry and data editing as well as a description of several security features including file archiving and file access protection Data transformations This chapter presents features to perform transformation of existing variables or to create new variables from computation on existing ones 2 SIMSTAT for WINDOWS Working with the notebook The notebook window provides an efficient way to browse and manage outputs This section provides basic instructions for navigating the notebook editing its contents and using tabs and the index to manage the outputs Statistical Analysis
113. cutoff point OPTIONS Cutoff point This option allows you to specify how the values of a variable will be dichotomized The cutoff point used can be either the mean the median or a user specified value All observations falling below the cutoff point form one group and all observations equal to or above the cutoff point form the other group Value This option is used only if you have a selected value as a cutoff point The value entered in this field is used as a cutoff point All observations falling below this value point form one group and all observations equal to or above the cutoff point form the other group If the data are already dichotomous specify a numeric value that lies between those two values Sample output of runs test RUN TEST AGE Age of the child Mean 9 5424 Number of runs Sal Expected number 29 0136 Cases 25 Lt Mean 34 Ge Mean f 931197 2 tailed P 7496 STATISTICAL ANALYSIS 141 Sensitivity analysis When using a questionnaire or an instrument to identify people who suffer from a specific disease or any particular problem the measure used is often associated with some diagnostic error Two types of errors are possible 1 the instrument can falsely classify a healthy subject as suffering from the disease false positive or 2 the instrument can fail to detect a person who has the disease false negative The error rates depend on the quality of the instrument and on the cut
114. d frequencies percentages or proportions are transformed into relative frequencies The number of values provided in this field must match the number of categories in the values field or in the data file If this option is left blank equal frequencies are assumed for all categories Sample output of one sample chi square analysis CHI SQUARE TEST SIBLING Number of siblings Value Frequency Expected ZDB 2 9 19 6 Tn 24 LOGY 2 00 6 EASL Chi Square 14 881 D F 2 P 0000 130 SIMSTAT for WINDOWS Oneway ANOVA The ONEWAY ANOVA procedure performs a one way analysis of variance for a dependent variable on groups defined by a categorical independent variable It allows testing whether the means of the groups 2 or more are significantly different from each other ONEWAY produces a table including between and within groups sums of squares mean squares degrees of freedom F ratio and its associated probability You can also obtain for each group descriptive statistics including count mean standard deviation standard error and a user specified confidence interval for the mean OPTIONS Range of x This option requests two values that will be treated as the minimum and maximum values of the grouping or independent variable under consideration Each discrete value of the independent variable that falls within this range defines a distinct group If those fields are left blank all values or the independent variabl
115. d in the notebook This index is useful to get an overall view of the notebook content and to quickly locate and move to a specific page It also allows you to restructure the contents of the notebook To view the notebook index click on the Index tab at the bottom of the notebook Expanding and collapsing sections Choose a section in the outline Press Enter or double click on the section name or its folder icon To locate and display a specific page Browse through the index until you find the page you want to display Double click on the page name or icon To move a page within the notebook Position the mouse cursor over the page you want to move Press and hold down the mouse button Drag the page over the page just before its new location and release the mouse button To erase a page Select the page you want to erase e Select the DELETE command from the EDIT menu or click on the button WORKING WITH THE NOTEBOOK 63 Output management using Tabs A typical research project or analysis task can result in hundreds or even thousands of pages of statistical output Using tabs can facilitate the management of your output by allowing the creation of sections in the notebook For example you may choose to create sections for different types of analyses descriptive analysis of data reliability analysis of instruments regression analysis to test your hypothesis etc or for analysis performed on different sub
116. d mean that is to the average of all group means Each coefficient associated with a coded vector represents the deviation from the grand mean for the group associated with this vector The predicted score of a subject is obtained by adding to the grand mean the regression coefficient of the group to which the subject belongs The specific effect of the last group can be obtained by computing the summation of all the b coefficients of the variables associated with the factor and inverting the sign of the result or bj The F and the significance value associated with the coefficients cannot be used to specify whether there are significant differences among the various groups but represent instead a test of the deviation of a group mean from the grand mean When the analysis involves more than one factor the mean of each cell can be obtained by substituting for each coded vector the proper code i e 1 O or 1 in the regression equation For instance the predicted score of a subject with treatment combination A and Bj is equal to the sum of the intercept grand mean the effect of treatment i of factor A the effect of treatment j of factor B plus the effect associated with the interaction between those two treatments When a quantitative variable covariate is included in the model the predicted score can be obtained by adding to the result the value obtained by multiplying the observed value on this quantitative variable by its b coefficient
117. d objective measurement of a single subject dependent variable over a period of time interspersed with changes in the treatment condition independent variable If the application or withdrawal of the treatment condition is associated with systematic changes in the behavior of the subject then it is inferred that the treatment has caused the observed changes Whereas with traditional group design variability is treated as error which should be controlled with the use of experimental design and statistical analysis in order to identify functional relations that supersede this variability advocates of single case experimental design stress the importance of identifying and controlling the source of this variability By carefully monitoring the variability of measures taken on a single subject it becomes possible to identify the source of this variability and to develop effective methods of intervention When applied to the evaluation of complex intervention programs this method facilitates the identification of the active ingredients of the intervention The reliance of single case design on visual inspection instead of statistical analysis to identify the presence of a functional relationship also provides assurance that only potent interventions that produce clinically significant changes will be identified The generality of the findings is strengthened through systematic replications ofthe original experiment with other subjects in various settings
118. d to specify the position of the window s upper left corner The parameter for these two options is an integer value between 0 and 100 expressing a percentage of the screen height and width SCRIPT LANGUAGE REFERENCE 207 Options TOP integer Vertical position of the box s upper left corner 0 to 100 TOP integer Horizontal position of the box s upper left corner 0 to 100 Example MENU LEFT 30 TOP 20 amp Open OPEN C NDATAFILENDATAFILE amp Save GOTO SAVEDATA Save amp As GOTO SAVEAS amp Quit QUIT ENDMENU MOSES Syntax MOSES varlist BY varlist l options Description The MOSES test of extreme reactions tests whether the range of an ordinal variable is the same in a control group as in a comparison group as defined by a grouping variable The output includes counts for both groups number of outliers removed the span of the control group before and after outliers are removed and one tailed probability of the span with and without outliers By default 596 of the cases are trimmed from each end of the range of the control group to remove outliers Options VALUES int int Values of independent variable OUTLIERS integer Number of outliers to remove PANEL Display the dialog box MRESPONSE Syntax MRESPONSE MULTX MULTY FREQUENCY CROSSTAB INTERRATERS options Description The MRESPONSE multiple responses c
119. database To add new variables to an existing data file Open the data file to which you would like to append new varaibles Choose the DATA APPEND VARIABLES command from the FILE menu e Select the data file to be merged with the current file and click OK Choose among the list of variables shared by both data files those that should be used to perform the record matching and click OK If no record of the secondary file matches the key values in the current file missing values are assigned to the new variables If several records in the secondary file match the key values values are extracted from the first matching record Variables with identical names are ignored To replace the values of one of those variables with the ones in the second data file simply delete this variable from the current data file prior to the APPEND VARIABLES operation Limiting access to data files DATA FILE OPERATIONS 43 Access privileges are useful for controlling access to sensitive data in a file whether or not the file is shared on a network For example you may want to prevent modification of data files by other users or simply prevent accidental modification by yourself SIMSTAT allows restrictions to be placed on the type of activities that can be performed on a data file Full access to the data file can be regained by providing a single password To define the default limited access to a data file Open the file you want to
120. de cases for which either expression is true For example GROUP 2 OR INCOME gt 28 000 will cause the program to select cases from GROUP 2 and cases from other groups only if they have an INCOME greater than 28 000 NOT The NOT boolean operator can be used to negate a condition or exclude records meeting a specified criteria For example NOT GROUP 1 will cause SIMSTAT to exclude all records for which GROUP equal 1 Multiple level expressions Multiple parentheses levels can be used to control the order in which expressions are evaluated as in the following example GROUP 2 AND INCOME gt 28 000 OR SEX 1 AND EDUC 5 36 SIMSTAT for WINDOWS Sorting records SIMSTAT provides two different ways to arrange the records or cases of a data file in numeric or alphabetic order Simple sort The easiest and quickest way to sort the records on the values of a single variable is to double click on the variable s name displayed on the top row of the grid The first time you double click on a variable s name the records are immediately sorted in ascending order on this variable Double clicking a second time on the same column title sorts the records in descending order Complex sort It is also possible to sort the records of a file using several variables To achieve this select the SORT RECORDS command from the DATA menu When activated this command brings a dialog box similar to the one u
121. disk export them to another application using the clipboard or disk files or print them It is also possible to delete a specific chart or to modify the order of those charts in this window Chart UNTITLED l E HRSPWR 34 530 00737 ACCE v gt 5of5 TO DISPLAY THE PREVIOUS CHART Press the lt Ctrl P gt key or click on the icon TO DISPLAY THE NEXT CHART Press the Ctrl N key or click on the icon TO SELECT A SPECIFIC CHART FROM A LIST e Click on the x button to display a list of all charts Select the chart from the list 156 SIMSTAT for WINDOWS TO ERASE A SPECIFIC CHART Use the navigation keys or buttons to display the chart you want to delete e Select the ERASE command from the EDIT menu or click on the button TO REORDER THE CHARTS IN THE CHART WINDOW Select the CHART ORDER command from the EDIT menu The following dialog box appears Bar chart of AGE Pareto chart of AGE Frequency distribution of AGGRESS Box amp Whiskers plot of AGGRESS Cumulative distribution of AGGRESS Normal probability plot of AGGRESS Bar chart of SEX with AGE Box amp Whiskers plot of AGGRESS Dual histogram of AGGRESS Mean and 90 C I of AGGRESS Mean and standard error of AGGRESS Mean and standard deviation of AGGRESS Deviation chart of AGGRESS AGGRESS 8 929 1 309 HOURSTY AGGRESS 34 165 10 188 AGE 37 Residual scatterplot for AGGRESS Normal probability plot of HOURST
122. dow The data window is a spreadsheet like data editor where data values can be entered browsed and edited Each column contains data for one variable and each row contains the responses for a single case You can have only one Data window open at a time The associated Data pull down menu offers a wide range of procedures to select a subsample of cases sort cases transform existing variables or to create new variables from transformation of existing variables Notebook window The Notebook window displays the statistical output for all analyses performed during a session The notebook metaphor provides an efficient way to browse and manage output The text output of each analysis is displayed on a separate page You can turn pages either with the mouse by clicking on the page flip icons at the bottom of the notebook or by using the lt PgUp gt and lt PgDn gt keyboard keys Tabs may also be added to create sections in the notebook allowing you to store different types of output in different sections of the notebook An index of all analyses included in the notebook is also available This index can be used to quickly locate and go to a specific page move pages within the notebook or delete some pages While each page can be annotated or edited it is also possible to add empty pages for recording ideas or remarks sketching an analysis plan or interpreting results Chart window The Chart window displays all high resolution charts creat
123. e COLOR option Options DELAY integer TOP integer LEFT integer color NOBUTTON BEEP Valid Colors Length of delay msec Vertical position of the box s upper left corner 0 to 100 Horizontal position of the box s upper left corner 0 to 100 Text color see below Hide the OK button Sounds a beep using computer s speaker Black Maroon Green Olive Navy Yellow Purple Teal Gray Silver Red Lime Blue Fuchsia Aqua White Example is the regression line ENDBOX BOX TOP 10 LEFT 10 COLOR NAVY The red line drawn in the scatterplot SCRIPT LANGUAGE REFERENCE 185 BREAKDOWN Syntax BREAKDOWN varlist BY varlist options Description The BREAKDOWN command computes descriptive statistics for various sub groups within the entire sample Statistics are computed for each variable on the first list of variable within groups defined by the values of the second list grouping or independent variables This command also allows you to obtain a multiple Box amp Whisker plot that can be used to compare the distribution of the dependent variable among several sub groups Options DETAIL Detailed statistics RANGE integer integer Range of X BOXPLOT Box amp Whisker plot PANEL Display the dialog box CALL Syntax CALL filename Description This procedure executes another script file After executing the external script file the program continues at the statement following the
124. e and setting its radio button to more than or equal Variable label This option lets you enter an alphanumeric description of the variable of up to 60 characters SIMSTAT will use this label when displaying statistical results or charts DATA FILE OPERATIONS 29 Value labels The value labels feature allows you to assign to each value of a numeric variable a string of up to 60 characters To define new value labels for the current variable click on the Edit button next to the value label option The following dialog box will appear alue labels SEX x Linked to None Value Label 1 Male 2 Female TO ADD A NEW LABEL Enter a numeric value in the Value edit box Enter the description associated with this numeric value in the Label edit box Press Enter or click on the Add button to accept this definition TO DELETE AN EXISTING VALUE LABEL nthe value label list box select the label you want to erase The caption of the ADD button will change to DELETE e Click on the DELETE button 30 SIMSTAT for WINDOWS TO MODIFY AN EXISTING LABEL Enter the numeric value you want to modify in the Value edit box Enter the new label in the Description edit box Press Enter or click on the Add button SIMSTAT will ask you to confirm the replacement of the existing value label Click on the Yes button to replace the existing value label USING AN EXISTING VALUE LABE
125. e ede e de e de e de dede e de eee de eee ende STEP LR KKK KKK KK kk kck ck ck ck kk kckck ck ck kck kckck ck ck kk KK Variable s entered on Step 1 SEX Sex of the child 122 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 123 124 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 125 Multiple responses analyses The MULTIPLE RESPONSE procedure allows you to obtain descriptive analyses and crosstabulation analyses on variables which can legitimately have more than one response These multiple responses are stored in as many variables as necessary Suppose you conduct a survey on the brand of cereal that peoples ate for breakfast in the previous week allowing respondents to give up to 3 brands of cereal One way to code this information in a data file would be to create as many variables as there are brands of cereal and record for each of those variables whether the respondent ate that particular brand or not Another way would be to create 3 variables corresponding to the maximum number of responses and to enter a code corresponding to each brand of cereal One problem with this type of coding is that in order to obtain a simple frequency analysis you would have to perform 3 different frequency analyses and compute the total frequency by manually aggregating those 3 tables The difficulties increase if you choose to perform a crosstabulation of this variable with another one The MULTIPLE RESPONSES procedure allows you to easily obta
126. e independent variable among several sub groups OPTIONS Range of X This section requests two values that will be treated as minimum and maximum values of the grouping or independent variable under consideration Each discrete value of the independent variable that falls within this range defines a distinct group If those fields are left blank all values of the independent variable will be included Statistics When set to Brief the mean standard deviation minimum and maximum values and the number of cases are displayed on a single line To obtain additional statistics such as the skewness kurtosis mode median etc set this option to Detailed Box amp Whisker plot The Box amp Whisker Plot option allows you to obtain a multiple box amp whisker plot that can be used to compare the distribution of the dependent variable among several sub groups Sample output of a breakdown analysis BREAKDOWN HOURSTV Hours per week spent watching TV With SIBLING Number of siblings Variable Mean Std Dev Mimimum Maximum Cases Entire Population C502 So O 5 580 2900 39 SIBLING SUD 3 us digo 5 50 SU 29 SIBLING TE 15 92 dm Toc vox wur 24 LSTBLING 2 00 16 42 3 47 ISO 2272 510 6 STATISTICAL ANALYSIS 75 Correlations The CORRELATION command produces a matrix of Pearson product moment correlation coefficients for all pairs of dependent and independent variables You can request either exact probabilities for the coeff
127. e into a character string formatted according to the Windows settings For example if the date format was American and the date field contained March 21 1995 DTOC datefield would return 03 21 1995 DTOS DateField Converts an xBase date into a string formatted according to standard xBase storage conventions CCYYMMDD For example December 21 1993 would be returned as 19931221 Indexes that contain date elements should use the DTOS function which naturally collates into oldest date first EMPTY Field Reports the empty status of any xBase field Character and date fields are empty if they consist entirely of spaces Numeric fields are empty if they evaluate to zero Logical fields are empty if they evaluate to False Memo fields that contain no reference to a memo block in the associated memo file are empty IF Logical True Result False Result This is the immediate if function If the Logical expression is true return the True result otherwise return the False result The types of the True Result and the False Result must be the same i e both numeric or both strings etc The logical expression must of course evaluate as True or False IF DATE CTOD 12 31 93 O This Year Last Year APPENDIX A XBASE SYNTAX AND FUNCTIONS 245 IIF Logical True Result False Result Supported exactly like IF as noted above INDEXKEY Returns the current index key as a string Same as ORDKEY
128. e section COMPUTING VALUES or QUICK TRANSFORM below 50 SIMSTAT for WINDOWS Computing values The COMPUTE command allows the transformation of existing variables or the creation of numeric variables by performing numeric transformations on existing variables This command offers more than 50 operations and functions including numerical operators trigonometric transformations cos sin log etc statistical functions mean minimum maximum across variables or cases etc date and random number operations Conditional transformations can also be performed using an IF THEN ELSE logical structure To perform a numeric transformation select the TRANSFORM VARIABLE COMPUTE command from the DATA menu When this command is evoked the following dialog box appears Variable transformation x ABS numeric a Sume x ca BG 7 Acos numeric NO1 ASIN numeric SEX ELI s s 7 seen AB NE COUNT varlist CSC numeric AGGRESS FL s DAY date PHASE EXP numeric BEHAVIOR FACT numeric oR LAG variable LN numeric LOG numeric MAX varlist zi Transformation Store in F Conditional transformation NUMERIC 12 4 Formula v OK Precision x Cancel Help Store in This edit box allows you to specify the target variable where the computation results will be stored If the target variable already exists the program will ask you if you wa
129. e the Edit button SIMSTAT displays the Tool Properties dialog box with information for the selected program Modify the program property and click on the OK button to save those new properties SETTING THE PROGRAM PREFERENCES 235 11 SETTING THE PROGRAM PREFERENCES The PREFERENCES command gives access to a multi page dialog box where you can set global options affecting the program s working environment file handling and printing The current section provides a description of each option available in the Preferences dialog box ENVIRONMENT PAGE Show toolbar This option turns the main window s toolbar at the top of the screen on or off Turn off the toolbar to gain free work space Show status bar This option turns the status bar at the bottom of the screen on or off You can gain extra free work space by hiding the status bar Display hints This option allows you to toggle the display of Help Hints Beep on error This option allows you to determine whether or not the computer sounds a beep when an error message appears on the screen Show notebook scroll bars This option allows you to specify whether to display scroll bars on the notebook Displaying scroll bars allows you to scroll the output horizontally or vertically with the mouse Hide menu items unrelated to the active window This option allows you to choose between two different menu systems a task specific menu system where only menus related to the active
130. e the script language is that you may want to write sets of procedures that would be executed by someone else less familiar with statistics or with the operation of SIMSTAT But SIMSTAT s script language goes beyond the simple automation of statistical analysis and provides some commands that allow you to write interactive tutorials demonstration programs or even small applications to be used by someone else These special commands allow you to display textual information wait for a key to be pressed ask questions construct bouncing bar menus play music or sound files display graphics or animations etc Using normal and encrypted script files SIMSTAT normal script files are plain text files with an SCR extension They are usually created and edited from within the program but may also be created or edited using almost any word processor or text editing program However if you use a word processor make sure that you save the script file as a plain text file SIMSTAT can also execute encrypted and compressed script files with an SCZ extension Once you have developed a program you may want to prevent others from altering your source file or simply hide its contents One reason would be to make sure that no one else will commercialize your entire script or parts of it under their own name This encrypting feature may also be useful to prevent unauthorized changes to the original program or in the context of computer assisted instruction to
131. e values or a value range an equal sign and the new value Each value on the left of the equal sign is recoded into the value on the right The recoding proceeds from left toright and stops after a transformation occurs The SYSMIS and MISSING keywords can be used to represent missing values while the ELSE represents all non specified values Examples RECODE SESTATUS 1 2 1 3 4 5 2 6 10 3 ELSE MISSING To store the new codes into another variable use this command in combination with the COMPUTE command as in the following example COMPUTE SESTATCAT SESTATUT RECODE SESTATCAT 1 2 1 3 4 5 2 6 10 3 CE LSE 220 SIMSTAT for WINDOWS RECORD Syntax Description RECORD option The RECORD command allows you to move the cursor within the currently opened data file to make a specific record active You can also use this command to add a new record at the bottom of the data file Options FIRST LAST NEX PRIOR NEW REGRESSION Syntax Move Move Move Move to TO TO to the the the the first record last record next record previous record Create a new record REGRESSION varlist BY varlist Description options The REGRESSION command produces simple regression analysis for each pair of dependent independent variables SIMSTAT lets you choose among linear and
132. e will be included An ANOVA test is then performed on these groups Descriptives The Descriptives option displays the count mean standard deviation standard error and a user specified confidence interval around the mean of the dependent variable for each group Post hoc tests This option allows you to perform a post hoc multiple comparison between all group means The program gives a choice between four different methods each one using different criterion for computing the significance level or constructing confidence intervals of the difference between two means The Least significant difference LSD test computes a confidence interval and a standard Student s t test between all group means It is the most powerful a posteriori contrast test but as the number of pairwise comparisons increases so does the probability that at least one of the confidence intervals or significance tests is in error For this reason this test is usually recommended only when it is applied to comparisons that are planned before observing the data The Newman Keuls test applies different criteria depending on the number of steps between the two group means The further apart those two group means are from each other the larger the difference between those two groups must be in order to be significant This procedure should be used only when the group sizes are equal The Tukey s honestly significant difference HSD test uses a single criterion for all compariso
133. ect the filename 22 SIMSTAT for WINDOWS SIMSTAT for Windows data files SIMSTAT for Windows uses the industry standard xBase file format as its own format Thus the program can read DBF files created by almost any program that can create these files SIMSTAT supports dBase files containing up to 1022 fields provided that the maximum record length is less than 64k SIMSTAT will also create several associated files in order to store important information about the data file The following list describes the various file extensions that are used Extension Description STR Files with an STR extension contain all variable information defined by the user such as variable labels missing values and display format display width number of decimal places alignment This file also contains information about access restrictions set by the user VLBFiles with this extension contain all value labels defined by the user SET Set files contain information about defined sets of variables IDX Files with an IDX extension are automatically created by SIMSTAT to keep track of the filtering and sorting conditions set by the user FPT Files with an FPT extension are used to store texts entered in memo fields US When movin g data files to a new location you must also move these related files and place them in the same directory as the DBF data file To facilitate this task you may use the ARCHIVES BACKUP and ARCHIVES RESTORE command
134. ed during a session The associated CHART menu can be used to modify almost any chart feature such as the axes the labels or the legends You may also reorder the charts save those charts on disk print them or export them either to disk or to the clipboard in Windows Metafile Windows bitmap or tab separated values formats Script Window The script window is used to enter and edit SIMSTAT commands These commands can either be read from a script file on disk typed in by the user or automatically generated by the program When used with the RECORD feature the script can also be used as a log window to keep track of the analyses performed during asession These commands may then be executed again providing an efficient way to automate statistical analyses Additional commands also allow one to create demonstration programs computer assisted teaching lessons and even computer assisted data entry programs 6 SIMSTAT for WINDOWS Changing the active window To change the active window you can use one of three techniques Click on a visible part of another window Choose another window from the WINDOWS menu Click on the proper icon on the toolbar see below Script Chart Data Notebook Working with pull down menus To select a menu item Point to the menu name and click the left mouse button or press the Alt key and the underlined letter in the menu name Point to the command you want to execute and cl
135. ed initially Then variables that do not meet the statistical criterion are sequentially removed Stepwise selection is a combination of the forward and backward procedures With this procedure a variable is selected to be entered in the equation in the same manner as with forward selection but after each entry of a new variable each variable already in the equations is examined so that if it no longer contributes significantly to the regression itis removed The criteria used to enter or remove a variable from the equation can be specified by the user Much controversy surrounds the use of this type of multiple regression One of the criticisms of this technique is that it capitalizes on chance and thus offers a solution that often overfits the sample data and cannot be generalized to the population or even be replicated on another sample In a sense the solution obtained from the sample may be very unstable The reason that those procedures have been included here is that there remain some rare situations in which these techniques are needed such as when one wants to select a limited number of variables among a set of good predictors mainly for practical reasons In the author s opinion hierarchical regression is probably the most highly recommended procedure However this procedure requires clarification of the logic and theory behind the data collection OPTIONS ANALYSIS PAGE Method The Method list box option allows you choose amon
136. eee nee nee 43 Archival backup of data files 2 2 ec eh 45 Creating and using variable sets 0 2 eee eee teenies 46 4 Data transformation 0 0 hrs 48 Examining data distributions 1 0 llle 49 Computing values i d ege Re eR ee a a ee ie 50 Performing conditional transformations 0 0 cece eee eee 55 Quick transformations T a hr s 56 Recoding values of a variable 2 e en a ccc eee eee 57 Transforming a variable into ranks 0 0 eee eee eee 58 Dummy recoding of nominal variables 0 0 cece eee ee eee 58 N rmerlefecoding 11e Ree petto Sacer gentetaec sale pires e BERGER 59 5 Working with the notebook llleeeeeeeeee I Ie 60 Navigating in the notebook lsllseeeeeee eee eee eens 61 Using the notebookndex i e c DAL EU ese SR RE RR 62 Output management using Tabs 0 0 cece eee ee 63 Rounding numerical values 0 0 0 0 eee te e 64 6 Statistical analysis 52i e ica e ba due salva b ped a S 65 Assigning variables for statistical analysis 0 0 2 0 cece eee eee eens 65 Binomial tests eeeh ilh A eA E ALIVE bu 67 Bootstrap analysis vir sot dece gba ex EEPIEAGUDVS Iu EA SRE ESOS REESE 68 Full analysis bootstrap analysis 0 0 2 eee ee eee eens 73 Breakdown nalysis 333 8955 Races e Ae e ce d Ren C RT ee 74 Correlations 1 22 ia as Oh ere ea pee Ria aed pi ESTEE 75 CrosstaDs 25 1 2 ce nemerTreduere isduemerirode cup bie
137. eek from 1 to 7 TODAY Returns the current date Missing Value Constant When the target variable already exists and a transformation expression is left blank the value of the target variable remains unchanged To erase the previous value and leave the variable blank use the SYSMIS keyword SYSMIS System missing value DATA TRANSFORMATION 55 Performing conditional transformations Conditional transformations allow you to apply a numeric transformation to a subset of records meeting a specific criteria When the CONDITIONAL TRANSFORMATION check box is enabled the single FORMULA edit box is replaced with the following 3 edit box IF The IF edit box contains a logical expression up to 250 characters long specifying criteria for records on which to apply a specific transformation This expression must be evaluated as true or false It can be a simple logical expression including the following 3 components a variable name a relational operator and a variable name or a numeric constant It can also be a complex expression composed of many simple expressions related with logical operators AND OR NOT Parentheses can be used to specify the order in which expressions are evaluated For example GROUP 1 AND AGE gt 30 OR GROUP 2 This expression can also include any xBase function provided that the final result of the expressions can be evaluated as true or false For more information on valid xBase expressions
138. eger This integer value specifies how many times the case will be duplicated If the value is less than 1 the case is excluded from the analysis To assign a variable as a weighting variable highlight in the variable list the variable you want to use as a weighting variable and click on the ES button next to the Weight box To remove the weighting variable select this variable and click on the Ej button next to the Weight box STATISTICAL ANALYSIS 67 Binomial test The BINOMIAL TEST allows you to assess whether the observed number of cases in a dichotomous variable is the same as that expected from a specified binomial distribution The dialog box allows you to specify whether the comparison will be made on the observations above or below the mean the median or a user specified cutoff value or to select observations equal to two values It also allows specification of the test proportion OPTIONS Cutoff point This list box allows you to specify how the values of a variable will be dichotomized The cutoff point used can be either the mean the median or a user specified value All observations falling below the cutoff point form one group and all observations equal to or above the cutoff point form the other group The value mode can also be used to restrict the analysis to two groups defined by distinct values Values This option is used only if you have selected Value as a cutoff point If only one number is provided it is
139. elihood ratio statistics Contingency coefficient Kendall s tau b Kendall s tau c Goodman Kruskal s gamma Somers d pearson s r 2 dimension barchart Overlapping bars Stacked bars 100 stacked bars 2 D perspectiv 3 D perspectiv Display the dialog box The DATA command allows you to define temporary data to be analyzed The first line following the DATA keyword must contain the variable names while the remaining lines hold the data SIMSTAT writes the embedded information to a temporary file named TEMP DBF and automatically opens it for analysis For more information on the proper format to use see information on ASCII files format 192 SIMSTAT for WINDOWS Example DATA CYLINDERS MPG ACCEL 4 15 0 6 3 4 17 2 5 9 6 13 2 5 3 6 12 4 6 2 6 12 4 5 8 ENDDATA DEFINE or DIM Syntax DEFINE varname AS NUMERIC STRING Description The DEFINE command allows you to explicitly declare a new memory variable and specify its type Using a separate DEFINE statement for each variable along with an explanation of the variable s purpose at the beginning of each subroutine and function makes the script easier to debug and maintain Examples DEFINE SANSWER AS NUMERIC DEFINE SUSERNAME AS STRING DELAY Syntax DELAY integer Description The DELAY command causes a delay in a program for a specified number of milliseconds Example Delay 5000
140. elimination Stepwise selection Enter all variables P value to enter P value to remove Minimum tolerance value Show each step ANOVA table Test of changes in R square History of changes in R Summary ANOVA table Variables in the equation Variables not in the equation Width of confidence interval Casewise plot of residuals Outliers criterion s d Durbin Watson statistic Residuals scatterplot Normal plot of residuals Save predicted values and residuals Display the dialog box NOTE Syntax SCRIPT LANGUAGE REFERENCE 209 NOTE expression options Description The NOTE command displays a single line string on the background screen By default the note line is displayed horizontally centered and at the bottom of the screen with a font size of 10 points The TOP and LEFT options can also be used to specify the position of the text s upper left corner The parameter for these two options is an integer value between 0 and 100 expressing a percentage of the screen height and width Other options allow you to control the size style and color of the displayed text Options TOP integer LEFT integer SIZE integer COLOR color ITALIC BOLD HIDE Examples NOTE Monthly analysis NOTE HIDE Valid Colors Black Maroon Green Gray Silver Red Vertical position of the text s upper left corner 0 to 100 Horizontal position of the text s upper left corner 0 to 100
141. emporarily exclude some subjects These conditions are specified in a logical expression that may consist of a simple expression or include many expressions related by logical operators AND OR XOR Most xBase functions see page 240 can be used with this command provided that the final expression can be evaluated as either true or false The selection stays effective until the logical expression is changed or the selection deactivated Type of analysis Principal Component Image Covariance Analysis aximum inimum Perform Display Display Display Display Display Display Display Save Display number of factors to retain igenvalue to retain varimax rotation a scree plot in output descriptive statistics correlation matrix unrotated factor solution rotated factor solution factor scoring weights factor scores factor scores dialog box When used alone this command deactivates the previous selection Examples FILTER FILTER AGE 10 OR deactivate the previous filter SI EX 1 FREQUENCY Syntax FREQUENCY varlist Description SCRIPT LANGUAGE REFERENCE 195 options The FREQUENCY command performs various frequency and descriptive analyses on specified variables This command also provides a choice of graphs for either categorical or numeric variables Options TABLE NOTABLE PTILES integer BARCHART
142. ems should be divided The dialog box provides a choice of various item statistics e g mean minimum maximum standard deviation inter item variance covariance and correlation matrices total scale and item total statistics The maximum number of items is limited to 90 items per scale STATISTICAL ANALYSIS 137 OPTIONS Item statistics This option allows the display of summary statistics on each item in the scale including the item mean minimum maximum standard deviation and mean correlation with other items It also provides various statistics for the whole scale such as the item means variances covariances and correlations Inter item correlations This option displays an inter item correlation matrix that allows the identification of items that have small correlations with the other items Variance covariance This option computes an inter item variance covariance matrix Item total statistics This option displays various statistics that allow the evaluation of the effect of each item on the reliability of the scale The output consists of the average score of the scale and its variance when the item is excluded the correlation of the scores on this item with the sum of the scores of the remaining items and the Cronbach s alpha that would result from the exclusion of this item Split half reliability This option allows one to assess the correlation between two parts of the scale When selecting the variables the Xs
143. ence in the new file format To export data to any of these applications Set the filtering and sorting conditions of the active data file to display the records as they should be exported Choose the DATA EXPORT command from the FILE menu e Select the file format you want to create using the Save As File Type drop down list Type a valid filename with the proper file extension If necessary use the Directories and Drives boxes to specify the storage location of your choice e Click on the OK button You will find below a list of available file formats along with information about the export function limitations SIMSTAT for DOS v1 0 v3 5 amp SPSS PC maximum of 500 variables can be exported Only the first user missing value is kept If an alphanumeric field is longer than 8 characters then all values will be truncated to 8 characters DATA FILE OPERATIONS 41 ASCII data files maximum of 500 variables will be exported Variable descriptions missing values definition and value labels are not supported All spreadsheet formats maximum of 255 variables will be exported When exporting to a file format supporting multiple pages the data are saved to the first page Variable descriptions missing values definition and value labels are not supported Creating a file with subsets of cases The export feature can also be used to create a new data file with an identical structu
144. endorsement and the measured ability A simple method consists of making sure that the percentage of correct responses to an item is higher for those who are high on the measured construct than for those who are low on this construct In the absence of an external criterion a second method uses the total score on all items as the discriminatory criterion In this situation we simply compute the total scores on the test for each subject and retain two distinct groups each representing a specific proportion of the examinees say 10 25 or 5046 positioned at the upper and lower end of the ability scale Once the upper and lower groups have been composed we can obtain an index of discrimination for an item by computing the difference in the percentage of correct responses between the upper and the lower groups This index often designated as the D index varies from 1 0 to 1 0 a positive value indicating that the item correctly discriminates according to the measured construct The greater the difference the better the item is able to discriminate between subjects On the other hand a negative value suggests a negative discrimination favoring those in the lower group and is strong evidence for a problematic item Ebel 1965 suggested the elimination or complete revision of items with a discrimination index less than 20 and the revision of those with an index value between 20 and 30 While the D index is easily computed and interpreted it suffers
145. er you want to exclude cases with missing values by either PAIRWISE or LISTWISE deletion If you select pairwise deletion a case is excluded if it has a missing value on either of the two variables used to compute a given coefficient However this case can be included in the computation of the other coefficients Listwise deletion excludes cases containing missing data from the computation of all the measures included in the matrix 128 SIMSTAT for WINDOWS Sample output of a nonparametric association matrix tau c STATISTICAL ANALYSIS 129 One sample chi square test The ONE SAMPLE CHI SQUARE TEST allows you to assess whether there is a difference between the observed number of cases in various categories and the expected frequencies in those same categories The dialog box allows you to restrict the test to specific values and to specify the expected frequencies OPTIONS Values This option allows you to specify the values that define the various categories For example if you enter the following string T 15 2273 the analysis will be restricted to the cases with values equal to one of these four categories This option can also be used to include categories for which there are no cases If you leave this field blank all distinct values encountered in the variable will form a separate category Frequency This option allows the specification of the distribution against which the sample will be tested All values expecte
146. erformed and what statistics are to be printed If no sample size is specified option SIZE the bootstrap sample size is automatically adjusted to the size of the original sample Options SIZE integer Size of each sample SAMPLING integer Number of samples SEED integer Initial seed value PANEL Display the dialog box GLMANOVA Syntax GLMANOVA varlist BY varlist options Description The GLMANOVA command provides a General Linear Model implementation of analysis of variance and covariance It can handles balanced and unbalanced ANOVA designs and support models with categorical and or quantitative variables The procedure can also be used to perform standard multiple regression problems that involve interaction terms Various adjustment methods for unequal cell size are provided including a hierarchical strategy where the user can set the order of entry of each variable in the model The various options allow you to display standard tables as well as various outputs usually found in ANOVA ANCOVA or multiple regression analyses Options COVAR varlist SCRIPT LANGUAGE REFERENCE 197 Quantitative variables INTERACTION var var var var STEP Show statistics at each step ULTREG ultiple regression statistics EQUATION Regression equation CI integer Confidence interval EAN Adjusted means CHANGE Test of changes in R square CPLOT Residuals casep
147. ertical scroll bars and use those scroll bars to browse through the active page see Setting Program Preferences section page 235 You may also use the lt PgUp gt or lt PgDn gt keys to move one screen at a time If you press the lt PgUp gt key while the cursor is already on the top of the page the program will bring you to the previous page while pressing the lt PgDn gt key when the cursor is located at the bottom of a page will move you to the next page if any To move from one page to another you may also click with the mouse on the page corner icons located at the bottom of the notebook VS Rather than moving from one button to the other in order to move forward and backward in the notebook pages you can click on the same button using the right button of the mouse to perform the reverse action To view the histograms produced during the same analysis activate the chart window The chart window displays the last chart produced To display the previous chart in the window press the lt Ctrl P gt combination key or click on the icon To display the next chart in the window press the lt Ctrl N gt combination key or click on the icon To select a specific chart from a list of all charts click on the button to display a list box and select the chart from the list When an image is displayed in the Chart windows it is possible to customize its axis scaling or labels change the colors and fonts and modify several ot
148. es COMPUTE TOTSCORE DEP1 DEP2 DEP3 DEP4 LOG DEP5 5 Conditional transformation can be performed by using successive FILTER and COMPUTE commands such as in the following examples FILTER AGE 8 COMPUTE TOTSCORE AVG DEP1 DEP10 FILTER AGE gt 8 COMPUTE TOTSCORE AVG DEP1 DEP15 or FILTER RELIGION 1 COMPUTE PREDFACT 1 235 FILTER NOT RELIGION 1 COMPUTE PREDFACT 0 656 CORANAL Syntax SCRIPT LANGUAGE REFERENCE 189 CORANAL varlist options Description The correspondence analysis or reciprocal averaging procedure performs several varieties of correspondence analysis including detrended correspondence analysis DCA Correspondence analysis in general is well suited for working with count or presence absence data whereas PCA is geared more towards measurement data on a continuous scale although PCA can also be performed on count and binary data Requires MVSP v2 2 Options LOG10 OGE LOG2 SQRT RATIO 3 1 RANSPOSI GI ECIPROCAL JACOBI PLOT GPLOT the following options apply DETRENDING SEGMENTS integer CYCLES integer DOWNWEIGHT T the following options apply KAISER JOLLIFFE MINEIGEN real ACCURACY real ERC RA CO AS H
149. es are named PREDxxx and RESIDxxx where xxx stands for a serial number between 001 and 999 that corresponds to the number of analysis performed during a single command This number is automatically reset to 001 after each command To prevent overwriting those variables you must rename them using the DEFINE VARIABLE command ORDER amp INTERACTION PAGE Type This column allows you to specify whether the variable is nominal or quantitative To toggle between these two types of variables click on the appropriate radio button at the bottom of the grid Order When a hierarchical approach is chosen see Adjustment Method this option allows you to specify the sequence of entry of the variables in the model The order for each variable can lie between 1 and 6 All variables with the same number are entered at the same time while variables that have lower numbers are entered before those that have higher numbers Variables for which the order is set to 6 are entered at the same time as the interaction variables You can enter the order at the keyboard or use the spin button located at the right edge of the grid to increase or decrease the order Interactions This option allows you to specify which interactions are tested For example to test for interactions between factors A and B and factors A B and covariate C enter Interactions AB ABC Sample output of aGLM ANOVA ANCOVA analysis GLM ANALYSIS OF VARIANCE Dependent Variable
150. es based on a small number of subjects Another common discrimination index is the point biserial correlation between an item and the scale total when the item is omitted The item is removed from the total score since keeping it would produce artificially inflated correlation coefficients A positive value suggests that those STATISTICAL ANALYSIS 103 who answered the item correctly scored relatively high on the scale total A negative value indicates that those who answered the item correctly have obtained relatively low scores on the scale total The biserial correlation coefficient is an index derived from the point biserial correlation This coefficient assumes that the variable measured by a dichotomous item response is in fact a continuous variable that is normally distributed Sincethe reduction of a continuous variable into a dichotomy has the effect of reducing its correlation with other variables the computation of the biserial correlation consists of applying a correction to the point biserial that tries to estimate what would have been the value of this correlation if the item had not been dichotomized One drawback of this correction is that the biserial correlation is no longer bounded by the 1 0 and 1 0 limits and can take values lower than 1 or higher than 1 INAPPROPRIATE USE OF ITEM ANALYSIS When a scale is used to assess the knowledge of individuals on a particular subject for a purpose other than selection or discriminati
151. f decimal places for newly created variables Changes the title line on printed output Number of charts per page 1 2 or 4 Enables disables demonstration mode 0 to disable or 1 to 30000 milliseconds options The SIGN test procedure tests the hypothesis that two variables have the same distribution This is assessed by comparison of the numbers of positive and negative differences between values of the two variables The probability test performed can be either one or two tailed Options 1TAIL 2TAIL PANEL Direction of the test Display the dialog box SCRIPT LANGUAGE REFERENCE 227 SORT Syntax SORT varname or expression Description This command allows you to arrange cases in numeric or alphabetic order If only one variable is specified the sorting will be done on this variable in ascending order To sort the records in descending order or to specify a more complex sort involving several variables you need to specify a sort expression This expression can include almost any supported xBase function Examples SORT AGE Sort on AGE in ascending order SORT DESCEND AGE Sort on AGE in descending order SORT SEX 100 AGE Composite xBase expression that will first sort the records on sex and then on age STOP Syntax STOP Description Immediately stops the batch command and returns to the SIMSTAT user interface T TEST Syntax T TEST varlist BY varl
152. f no values are given the observed minimum and maximum are used Sample output of a Kolmogorov Smirnov goodness of fit test KOLMOGOROV SMIRNOV GOODNESS OF FIT TEST AGGRESS Nb of agressive behavior Type of Distribution Normal Mean 27 542 Std Dev 16 578 Most Extreme Differences Absolute Positive Negative Iai A 2 tailed P GARES 16418 ilg 1L eL PISIS STATISTICAL ANALYSIS 109 Kolmogorov Smirnov 2 Samples Test The KOLMOGOROV SMIRNOV two samples test evaluates whether a variable has the same distribution in two independent samples as defined by a grouping variable This test is sensitive to differences in the shape location and scale of the two sample distributions The output displays the count in each group the largest absolute positive and negative differences between the two groups Kolmogorov Smirnov s Z and the two tailed probability for each variable OPTIONS Values of x This option requires two independent variable values The cases that match the first value on the independent variable form one group and the cases that match the second value form the second group The order in which values are specified determines which difference is the largest positive and which is the largest negative Sample output of a Kolmogorov Smirnov two samples test KOLMOGOROV SMIRNOV TWO SAMPLES TEST AGGRESS Nb of agressive behavior With SEX Sex of the child Cases 29 Sup c 1 00 30 SEX 2 00 59 Total Most
153. fidence interval Those bars can be used as a judgemental aid to identify a change in the series The Minimum and Maximum options allow you to specify on which observations in the series the mean and confidence limits will be calculated If those fields are left blank the first and the last observations will be used For example specifying the limits 1 and 10 tells SIMSTAT to compute the mean and the confidence interval for the first 10 observations Width This option allows you to set the confidence interval width for the control bars This interval width is expressed as a percentage and must lie between 1 and 99 Sample output of time series analysis TIME SERIES FORCASTEING SALES Nb of sales per month Smoothing Moving average 2 4 4 2 Mean 27 5424 Std dev 16 5783 Plot of time series Case Value 0000 66 0000 p 4 4R p 4 4R p 4 I 16 000 2 46 000 i 3 66 000 4 66 000 X 5 26 000 we 6 25 000 i 7 53 000 tg 8 46 000 fx 9 56 000 10 265 000 i TL 66 000 L2 36 000 ial T 46 000 j 14 35 900 z 15 46 000 PES 16 46 000 fe SE 30 000 i 18 46 000 D 18 6 000 gt 20 46 000 hs 21 S2 OOO oer D 26 000 Bie 23 26 000 ah ck 4 4 4 4 Smoothed value Observed value Plot of autocorrelations Lag COLT ig gt L 5 15 0 0 FI E BSS 005 Y EU Eon 148 SIMSTAT for WINDOWS STATISTICAL ANALY
154. from a major drawback When comparing only the two most extreme groups much information is discarded such as the percentage of success in intermediate groups or among the subjects within a group STATITEM offers several ways to examine the distribution of correct responses more closely The program can provide detailed information about the distribution of correct responses by breaking down the total number of examinees into several groups from 2 to 10 of similar size and displaying the percentage of correct responses for these groups It is also possible to display even more detailed information on the success rate of an item for every value of the total score allowing one to assess whether the item can provide adequate discrimination of subjects all along the scale To simplify the examination of the relationship between the proportion of success to an item and the total score STATITEM also provides a graphical representation of this relationship as an empirical item characteristic curve also called an item test regression curve which displays for each total score level the percentage of persons who responded correctly This curve allows one to examine the item s difficulty and discriminatory properties It also provides a complete picture of the relation between item performance and the total score STATITEM can display an empirical curve computed from the original scores or a smoothed version that may be used to eliminate noise caused by percentag
155. g 5 different strategies of multiple regression Hierarchical entry forward selection backward deletion stepwise selection and standard regression Show steps This option requests the printing of various statistics at each step of the regression The information output at each step can include an ANOVA table a test for change and various statistics for the variables in the equation and for those not yet in the equation ANOVA table The ANOVA Table option allows printing of an ANOVA table that includes regression analysis and residual sum of squares mean square F and probability value associated with F 120 SIMSTAT for WINDOWS Test of change This option allows printing an ANOVA table that tests whether the new variable s in the model significantly increase R square above the R squared predicted with the variables already in the equation The table includes the resultant R squared the R squared change the sum of squares the F ratio and its probability value Variables in the equation This option displays various statistics computed on the variables in the equation including each unstandardized regression coefficient or Bj its standard error and confidence limits the standardized coefficient or beta whole partial and semi partial correlation coefficients tolerance level F ratio and its probability Variables not in the equation This option displays the tolerance level the F ratio and its probability for each var
156. g various estimator distributions obtained for different sample sizes can prove useful in planning research by allowing the researcher to determine the sample size needed to achieve a desired precision level It can also be used for power estimation allowing comparison of the power attained using various estimators and or sample sizes Researchers thus have an empirical basis for choosing between two different statistical strategies In addition unlike standard approaches to power estimation which rely on numerous assumptions including normal data distributions bootstrap power estimates make no distribution assumptions STATISTICAL ANALYSIS 71 Teaching Tool As a teaching tool bootstrap simulation would be effective in illustrating to new statistics students concepts such as sampling theory or the central limit theorem It would provide a simulation of the sampling process of an experiment allowing the students to visualize the sampling variability of given estimators By increasing or decreasing sample size the student can observe how these changes affect the variability of estimators or the statistical power of an experiment Additionally bootstrap would be effective in demonstrating how outliers can affect estimation and how data transformation can improve population estimates Monte Carlo investigations Bootstrap might also be handy for the researcher interested in studying the effect of violation of the normality assumption on some e
157. generic text document that can be pasted into or imported by a wide range of charting and spreadsheet applications TO TRANSFER A SINGLE CHART TO THE CLIPBOARD Display in the Chart window the chart you want to transfer Select the COPY AS command from the EDIT menu e Select the format in which you want to transfer the chart 166 SIMSTAT for WINDOWS TO EXPORT A SINGLE CHART TO FILE Display in the Chart window the chart you want to export Select the CHART EXPORT command from the FILE menu Activate the Current Chart radio button to specify that only the current chart should be exported to disk Setthe List of File Type list box to the format in which you want to export the current chart Enter a valid file name and click OK to exit the dialog box and export the chart TO EXPORT ALL CHARTS IN THE CHART WINDOW TO FILES Make the Chart window active e Select the CHART EXPORT command from the FILE menu Activate the All Charts radio button e Set the List of File Type list box to the format in which you want to export the charts Enter a valid prefix that will be used to create file names and click OK to exit the dialog box and export the charts This prefix can have a maximum length of 5 characters A serial number between 001 and 999 will be automatically added to this prefix to create unique file names NOTE Great care should be taken when exporting several charts in BMP format since this single co
158. ght and width Other options allow you to control the size style and color of the displayed text Options TOP integer LEFT integer SIZE integer COLOR color ITALIC BOLD HIDE Examples H p E Monthly analysis ITLE HIDE Valid Colors Black Maroon Green Gray Silver Red Vertical position of the text s upper left corner 0 to 100 Horizontal position of the text s upper left corner 0 to 100 Size of the font between 6 and 100 Color of the font see below Displays the string in italic characters Displays the string in bold characters Hides the string SIZE 10 COLOR YELLOW BOLD Olive Navy Yellow Purple Teal Lime Blue Fuchsia Aqua White 230 SIMSTAT for WINDOWS VARDEF Syntax VARDEF varlist options Description The VARDEF command can be used to assign a value description or define the display width number of decimal places or missing value for one or several variables Options LABEL string Variable label WIDTH integer Display width DECIMAL integer Number of decimal places to display MISSING real User defined missing value Examples VARDEF AGE LABEL Age of the child WIDTH 5 DECIMAL 2 VARDEF SEX RELIGION INCOME MISSING 9 VLABELS Syntax VLABELS varlist real string real string Description The VLABELS command can be used to assign alphanumeric strings to value
159. gram NBAR integer Nb of bars intervals NORMAL Normal curve PANEL Displays the dialog box SCRIPT LANGUAGE REFERENCE 183 BOOTSTRAP2 Syntax BOOTSTRAP2 varlist BY varlist options Description The BOOTSTRAP2 command performs bootstrap resampling to estimate the distribution of various estimators in a given population e g correlation Tau b Cohen s Kappa The program draws a specified number of pairs of observations from the sample and computes the estimator for the subsample This procedure is performed many times 10 to 30 000 times The options allow you to display information about the estimator distribution including descriptive statistics frequency table percentile table and histogram of the estimator distribution The program can also compute nonparametric and bias corrected bootstrap confidence intervals and the power rate of some estimators for 3 to 4 alpha levels If no sample size is specified option SIZE the bootstrap sample size is automatically adjusted to the size of the original sample Options SIZE integer Size of each sample SAMPLING integer Number of samples SEED integer Initial seed value Choice of statistics TAU A Kendall s Tau A TAU B Kendall s Tau B TAU C Kendall Stuart s Tau C D SY Somers D symmetric D XDEP Somers D X dependent D YDEP Somers D Y dependent GAMMA Gamma RHO Spearman s Rho R Pearson s r SLOPE Regression
160. grid 159 Editing titles and axis labels 158 eoa if EEE E eed 59 90 Eigenvalue 81 84 189 193 212 Encrypted script files 0 04 168 Endorsement rate 2 2 0 0 eee ee eee 104 Entering and editing data 31 Error bar chart 130 149 210 227 Error r tes 2 vo ro e 68 141 225 Excel files 0 eee eee eee 38 40 199 Exponential regression 52 133 176 220 Exportation Charts toD Bae AES 155 165 166 Data to other applications 40 41 Expression operators and functions 176 240 F Factoranalysis issssciic ARR ER 81 193 Filtering recotds 22cm 33 194 Font 2 oar 158 161 209 229 238 Forward selection llle 119 208 Frequency analysis 15 86 97 125 186 195 Friedman test eei ck Ee E RR 89 195 Full analysis bootstrap 00 73 196 Further reading inas Aa 249 252 G Gamma 69 78 127 183 191 210 218 Getting helps cvs If xy E a AEK 11 GLM ANOVA ANCOVA command 90 196 250 Global options 20 0000 160 225 235 GO TO command cee eee eee 60 H Headers wrrklsrkehifem4a er waa kode ee ABS 225 237 Help eera rire ra Rex EU PER 11 Hinis Sr a es 9 225 235 Hierarchical regression 90 118 119 208 Histogram 49 72 87 150 183 184 195 217 227 I Image covariance
161. gularities in the curves caused by small numbers of subjects used in the computation Item total response breakdown This option allows one to separate the whole sample into several groups of equal size and to compare the percentage of cases who scored positively on each item The Number of groups specifies the number of groups to create This number must lie between 2 and 10 For example setting this option to 4 will create four different groups based on the obtained total score each group representing 2596 of the respondents The program will then display a table that indicates for each item of the scale the percentage of correct responses for these four groups A discrimination index is also computed for each item by computing the STATISTICAL ANALYSIS 105 difference in the percentage of endorsement between the groups with the highest and the lowest scores Scale statistics The Scale Statistics option displays various descriptive statistics on the distribution of the scale total mean median minimum and maximum standard deviation standard error skewness kurtosis etc The Cronbach s alpha internal consistency statistic and the Ferguson s delta discrimination index are also computed This option also provides summary statistics on the percentages and on the biserial and point biserial correlations between each correct response and the scale total Total score distribution Activate this option to scrutinize the distribution of tot
162. h of these correction techniques free marginal adjustment assumes either that all categories on a given scale have equal probability of being observed or that the judges have not based their decisions on information about the distribution for their ratings Scott s pi adjustment does not assume that all categories have equal probability of being observed but does assume that the distributions of the categories observed by the judges are equal Cohen s kappa adjustment does not assume that all categories have equal probability of being observed nor that the distribution of the categories is equal for the two judges It does however in computing the chance factor take into account the differential tendencies of the judges SIMSTAT also offers three adjustments for ordinal level variables These are similar to the previous measures except that they also take into account the ordinal nature of the scales by adjusting the weights assigned to various levels of agreement They apply the same tree model of chance agreement used in the previous measures of nominal data Free marginal adjustment for ordinal level variables also assumes that all categories on a given scale have equal probability of being observed Krippendorf s R bar adjustment is the ordinal extension of Scott s pi and assumes that the distributions of the categories are equal for the two sets of ratings Krippendorf s r adjustment is the ordinal extension of Cohen s Kappa in that it adjusts fo
163. h string is contained within a target starting from the right side of the target string If found the function returns the position of the search string within the target string relative to 1 If not found the function returns 0 zero RAT ab abzaba returns 4 RECCOUNT Returns the number of records in the table as a long integer RECNO Returns the current physical record number as a long integer RIGHT String Length Returns the rightmost characters of the expression for the defined length RIGHT xyzabc 3 returns abc SELECT Returns the workarea number for the current workarea as a long integer SPACE Length APPENDIX A XBASE SYNTAX AND FUNCTIONS 247 Returns a string consisting entirely of spaces for the defined length STOD String The inverse of DTOS STOD converts a string formatted according to standard xBase storage conventions CCYYMMDD to an xBase Date formatted according to the Windows settings STR Number Length Decimals Converts a number into a right justified string with decimal digits following the decimal point The total length of the string is defined by the length parameter STR RECNO 5 0 isa common indexing element that ensures creation of unique keys if appended to another field element An index key using this expression could be built with NAME STR RECNO 5 0 If the decimals parameter is omitted the function defaults to zero decimal places If t
164. havior change New York Pergamon Press GRESHAM F M 1991 Moving beyond statistical significance in reporting consultation outcome research Journal of educational and psychological consultation 2 1 1 13 MICHAEL J 1974 Statistical inference for individual organism research Mixed blessing or curse Journal of Applied Behavior Analysis 7 647 653 SIDMAN M 1960 Tactics of scientific research Evaluating experimental data in psychology New York Author s Cooperative Press REFERENCES 253 RELIABILITY ANALYSIS CARMINES E G amp ZELLER R A 1979 Reliability and validity assessment Beverly Hill Sage Publication NUNNALLY J C 1964 Educational measurement and evaluation New York McGraw Hill BOOTSTRAP SIMULATION DIACONIS P amp EFRON S 1983 May Computer intensive methods in statistics Scientific American 116 130 EFRON B 1981 Nonparametric estimates of standard error The jackknife the bootstrap and other resampling methods Biometrika 68 589 599 EFRON B amp GONG G 1983 A leisurely look at the bootstrap the jackknife and cross validation American Statistician 37 36 48 4 EFRON B amp TIBSHIRANI R J 1993 An introduction to the bootstrap New York Chapman amp Hall MOONEY C Z amp DUVAL R D 1993 Bootstrapping A nonparametric approach to statistical inference Beverly Hill Sage Publication PELADEAU N amp LACOUTURE Y 1993
165. he length parameter is omitted as well the length of the result is the length of the field STRZERO Number Length Decimals Converts a number into a zero padded right justified string with decimals digits following the decimal point The total length of the string is defined by the length parameter STRZERO 1234 10 2 returns 0001234 00 If the decimals parameter is omitted the function defaults to zero decimals If the length parameter is omitted as well the length of the result is the length of the field SUBSTR String Start Length Returns a portion ofthe string expression starting at the defined start location for the defined length SUBSTR xyzabcd 3 4 returns zabc TIME Returns the system time as a string in the form HH MM SS TRANSFORM Expression Picture 248 SIMSTAT for WINDOWS Transform converts strings and numeric values into formatted character strings The function transforms the result of the first expression in accordance with the second picture string The picture string is made up of two parts The first part is the Function string and it is optional for both strings and numeric values as long as the second Template string is present A character string transformation picture may consist of only a Function string or only a Template or both A numeric picture must contain a Template string the Function string is optional A logical value must contain only a Template string
166. he recoding proceeds from left to right and stops after a transformation occurs The SYSMIS and MISSING keywords can be used to represent missing values while the ELSE keyword represents all non specified values Once a valid target variable and a recoding numeric expression have been entered you can leave this dialog box and perform the recoding by clicking on the OK button If the recoding expression is invalid a message is displayed and the exit is not performed To exit from the dialog box without performing the recoding click on the CANCEL button 58 SIMSTAT for WINDOWS Transforming a variable into ranks The RANK command replaces the values of a variable by their rank If ties occur the mean rank of tied values is used Missing values are excluded To transform a variable into ranks Position the data sheet cursor on the variable you wish to transform into ranks e Select the TRANSFORM VARIABLE RANK command from the DATA menu Enter the new variable name under which the rank should be stored and click OK to proceed to the transformation Dummy recoding of nominal variables The DUMMY RECODING command automatically creates as many dummy variables as needed to represent the values of a nominal variable To create such dummy variables position the cursor on the nominal variable you wish to recode and select the TRANSFORM VARIABLE DUMMY RECODING command from the DATA menu The following dialog box will appear Dummy
167. he RECORD SCRIPT command to automatically generate commands Write script files to disk The final part of this chapter provides programming information including a reference section with a description of all available commands their syntax and related options 168 SIMSTAT for WINDOWS Introduction to the scripting feature While SIMSTAT pull down menus and open panels allow you to do almost everything you need it is also possible to perform analyses using the SIMSTAT batch command language But why should you bother using this script language if you can do everything you need with the user interface First using the script language allows you to keep track of what you did during a session This may prove very useful if you want to come back later and inspect what you did or if you have to provide to someone else a detailed description of your analyses The second advantage is that it allows you to automate statistical processing of your data files Often you will have to perform an identical series of analyses on the same data file on several occasions such as every month or every year It may become easier and faster to resubmit the same script file than to remember all you did and do it again with the menus and dialog boxes You may also want to perform similar analyses on different data files In that situation modifying an existing script file may be more efficient than starting from scratch with the menus Another reason to us
168. he current scale A list box also allows you to choose among 3 different line styles to draw those grid lines 160 SIMSTAT for WINDOWS Controlling the legend display This command brings a dialog box that allows one to specify whether a legend should be displayed on the chart and to set the legend position egend options xi r Appearance Show legend Location C Left C Right C Bottom To resize the width or height of a legend simply drag its border to resize it To set a default legend location upon creation of new charts see the GLOBAL OPTIONS command Displaying data point values Some chart types give you the option of displaying the numeric value associated with each data point When such an option is available the DATA LABELS command in the CHART menuis enabled Selecting this command immediately displays those data labels To remove the data labels select the DATA LABELS command again Modifying specific chart options The SPECIFIC OPTIONS dialog boxes control the display of properties that are specific to the currently displayed chart type For most chart types the options available in this dialog box are the same as those displayed in the dialog box used to create the charts For example if the current chart is a histogram this dialog box gives you the option of removing or adding the normal distribution curve or changing the number of intervals WORKING WITH CHARTS 161 Sett
169. he filename extension is used to determine the graphic type of the file Up to 5 different pictures can be displayed on the same screen This command can be used with the CURRENTCHART option to display a bitmap copy of the currently active chart When used with this option the size of the displayed picture will be equal to the size of the chart window s content see the WINDOW command for instruction on how to change the size of a chart window By default the pictures are displayed in the middle of the screen The TOP and LEFT options can also be used to specify the position of the picture s upper left corner 214 SIMSTAT for WINDOWS Options NO1 NO5 Picture number LEFT integer Vertical position of the picture s upper left corner 0 to 100 TOP integer Vertical position of the picture s upper left corner 0 to 100 CURRENTCHART Display the currently active chart HIDE Hide the picture Examples PICTURE NO1 C NSIMSTATNSCRIPTNEARTH BMP TOP 10 LEFT 20 PICTURE NO2 CURRENTCHART TOP 30 LEFT 60 PICTURE NO1 HIDE PLAY Syntax PLAY filename options Description The PLAY command allows you to play multimedia files such as WAV sound files MDI music files or AVI movie files If no directory is provided the program successively looks for the file in the active directory and in the directory where the current script file is located When playing an AVI movie file the viewer is positioned i
170. he keyboard or select a variable name or function Numeric boolean and relational operators buttons Clicking on any relational or boolean operation or on any numeric button inserts the corresponding symbol in the edit box at the current caret position Usually when part of the filtering expression is highlighted pressing any keyboard key or clicking on any variable name or numeric button will replace the highlighted text with the character or expression associated with this key or button However when choosing a function requiring a parameter enclosed between parentheses the highlighted text will not be deleted but will be used instead as the new parameter of this function Examples of a simple filtering expression Here is an example of a simple filtering condition GROUP 1 This expression tells SIMSTAT to select only those records which have a value for GROUP that is equal to 1 DATA FILE OPERATIONS 35 Using boolean operators in the filtering condition The AND and OR switches are logical operators that allow the evaluation of 2 or more logical expressions AND The AND operator instructs SIMSTAT to include only the cases for which both expressions are true For example GROUP 2 AND INCOME gt 28 000 will cause the program to select only those cases from GROUP number 2 that have an INCOME greater than 28 000 OR The OR operator instructs SIMSTAT to evaluate both logical expressions and to inclu
171. he probability significance for the whole model omnibus test Equation When enabled this option displays various statistics computed for the variables included in the model including the regression coefficient or B its standard error and confidence limits the standardized coefficient or beta whole partial and semi partial correlation coefficients F ratio and its probability SIMSTAT generates coded vectors to represent independent categorical variables factors and uses multiple regression techniques using those vectors in order to accomplish ANOVA or ANCOVA problems and obtain a regression equation that can be used for prediction Effect coding is used to create the vectors such that for each vector created cases of one group are assigned 1 s while all other cases are assigned O s with the exception of the cases of the STATISTICAL ANALYSIS 91 last group which are coded as 1 s The following table illustrates the result of an effect coding where 3 vectors X1 X2 and X3 are created in order to represent the various values contained in the GROUP variable 1 1 0 0 1 1 0 0 2 0 1 0 2 0 1 0 3 0 0 1 3 0 0 1 4 1 1 1 4 1 1 1 The regression equation obtained with this method provides much useful information When the sample sizes are equal and there is no covariate the intercept is equal to the grand mean of the dependent variable When the analysis involves unequal cell sizes the intercept is equal to the unweighte
172. her features of the chart All these customization options can be accessed from the CHART menu It is also possible to save these charts on disk export them either to files or to the clipboard or print them For more information on the various chart options available see the section entitled Working with Charts on page 153 TUTORIAL 17 Step 5 Selecting a subset of cases For our second analysis we will perform two regression analyses where AGGRESS number of aggressive behaviors will be used as the dependent variable while AGE and HOURSTV will be treated as two separated independent variables or predictors For the purpose of this demonstration we will restrict the analysis to male subjects To select this subgroup choose the FILTER command from the DATA menu This command displays a dialog box like the one below ilter records x All E HAE ane puli PHASE OR s L SIBLING NOT o E ALLTRIM string n AT searchstring targetstring CTOD string DATE DAY datefield IV Sorted DELETED zl Eilter El Apply 8 Ignore x Cancel Help The Filter edit box allows you to specify a condition that must be met in order to include the case in the analysis The sex of the child has been stored in a numeric variable named SEX using 1 to designate boys and 2 for girls To select the boys for the next analysis type the following condition
173. highlighted Toreplacethe highlighted text with a value an expression or a variable name simply type in the proper text or select a variable name or function Numeric arithmetic boolean and relational operators buttons Clicking on any relational or boolean operation or on any numeric button inserts the corresponding symbol in the edit box Usually when part of an expression is highlighted pressing any keyboard key or clicking on any variable name or numeric button will replace the highlighted text with the character or expression associated with this key or button However when choosing a function requiring a parameter enclosed between parentheses the highlighted text will not be deleted but will instead be used as the new parameter of this function 52 SIMSTAT for WINDOWS You will find below a list of valid symbols and functions along with a short description of each as well as some additional syntax information Arithmetic Operators You can use any of the following symbols in a transformation formula Addition Subtraction Multiplication Division Exponentiation When a transformation expression is evaluated exponentation is performed first followed by division and multiplication and finally addition and substraction Multiple parentheses levels can be used to control the order in which expressions are evaluated as in the following examples 2 3 4 2 returns 48 2 3 4 2 returns 28 2 3 4 2
174. hoose an appropriate cutoff point The program provides for each value of the quantitative measure the level of sensitivity proportion of positive cases correctly diagnosed as true and specificity proportion of negative cases correctly diagnosed as false and the percentage of false positives and false negatives This command also allows you to obtains a ROC receiver operating characteristic curve and an Error rate graph Options VALUE real Criterion value HIGH Scale orientation ascending SSTA Sensitivity statistics ESTA Error rates statistics ROC Roc curve graph ERROR Error rates graph PANEL Display the dialog box SET Syntax SET options Description The SET command can be used to change various global program options such as the display of the toolbar the status bar or the help hints It can also be used to modify options used when printing notebook pages or charts such as the page header or the number of charts per page The HEADER option is used to specify a line of text that will appear at the top of each printed analysis result By default this command alters the contents of the header printed at the center of the page The sections of the header printed to the left or the right edge of the page remain unchanged However you can also modify those sections by using the less than and the greater than gt character All text that appears to the left of the first lt character will be pr
175. iable not yet in the equation History When the History option is activated a summary report of the changes in R squared at each step of the regression is printed Summary ANOVA This option allows the printing of a detailed ANOVA table that includes the mean square the sum of squares the F ratio and its probability for each variable or set of variables in the equation Confidence interval This option allows you to set the width of the confidence interval for the unstandardized regression coefficients This interval is expressed as a percentage and must lie between 1 and 99 DIAGNOSIS PAGE Residual caseplot This option allows the output of a casewise plot of standardized residuals including the predicted obtained and residual values for all cases This option is useful for identifying outliers 1 e cases that are not well represented by the regression model The Outliers option allows you to restrict the residual caseplot to those cases for which the absolute standardized value is greater than or equal to the specified value The Outliers value can be set to between 0 and 4 standard deviations Durbin Watson statistic This option tests for the presence of autocorrelation or serial correlation in the residuals The larger the autocorrelation the less reliable the results of the analysis Residual scatterplot This procedure produces a bivariate scatterplot The predicted value is plotted along the horizontal axis and the
176. iable on the left side of the operator The assignment statement must do the following Identify the variable that receives the value Use the assignment operator to separate the variable from the expression End with the expression that determines the variable s value Examples LET SAGE 12 LET SCLASSNAME STATISTICS 101 You can also use a variable on both sides of the first assignment statement that uses it For example the following statement increases the value of the variable COUNT by one LET COUNT COUNT 1 In addition to basic arithmetic operations you can use various functions including mathematical procedures trigonometric functions and random number functions on the right side of the expression see page 176 for a list of available functions 204 SIMSTAT for WINDOWS You may also use parentheses to control the evaluation sequence such as in LET COUNT SLAST 2 1 1415926 LET SNEWAGE SQRT DB AGE 2 NORMAL 2 LIST Syntax LIST varlist N integer Description The LIST command displays a listing of the values of the specified variables The N option can be used to restrict the listing to the first N cases in the file LOGISTIC Syntax LOGISTIC varlist options Description The LOGISTIC command performs a logistic multiple regression Requires LOGISTIC v3 11ef Options VALUES in
177. ian distance Cosine theta distance anhattan metric distance Canberra metric distance Chord distance Chi square distance Average distance Mean character difference distance Pearson product moment correlation Spearman rank order correlation Percent similarity coefficient Gower general similarity coefficient Sorensen s coefficient Jaccard s coefficient Simple matching coefficient Yule coefficient Nei amp Lei s coefficient Clustering method default UPAIR Unweighted pair group Unweighted centroid Weighted pair group Weighted centroid inimum variance 188 SIMSTAT for WINDOWS NEAR Nearest linkage FAR Farthest linkage REEDESC Create tree description file REEORDER Create tr order fil RANDOMIZE Randomize input order CONSTRAIN Constrained clustering Output of graphics PLOT Text dendrograms GPLOT Graphic dendrograms COMPUTE Syntax COMPUTE varname expression Description The COMPUTE command allows the transformation of existing values of a variable or the computation of a new variable SIMSTAT offers more than 50 operations and functions including numerical operators trigonometric transformations cos sin log etc statistical functions mean minimum maximum across variables or cases etc and date and random number operations For more information on the available transformation functions available see the Computing values section page 50 Exampl
178. icients or a display of asterisks that indicate the probability levels attained You can also tell the program to calculate probabilities using one directional or two tailed tests non directional and to display cross product deviation and covariance tables OPTIONS Type of matrix When set to X vs Y the correlation matrix displays correlations between all variables assigned as independent against all variables assigned as dependent The square matrix option produces a matrix displaying correlations between all selected variables without taking into account whether they were selected as dependent or independent Confidence intervals The confidence intervals option allows you to display confidence intervals of the correlations The interval Width is expressed as a percentage between 1 and 99 Display probabilities values This option determines the content of the correlation matrix When disabled the display includes up to 3 asterisks to indicate the significance level attained for each correlation coefficient If you enable this option the program prints a matrix including the number of cases used to compute each coefficient and the estimated probability of the correlation Significance test This option specifies whether the probability of the correlations is based on a one tailed directional or two tailed non directional test Missing values The Missing Values option allows you to specify whether you want to exclude
179. icitly uses either a covariance or correlation matrix PCO allows you to input any matrix of metric values PCO may be used with any of the distances calculated by MVSP except for the squared Euclidean distance Of the similarity measures only Gower s is metric PCO is calculated as a Q mode eigenanalysis therefore it only gives the eigenvectors not scores Note that a PCO of Euclidean distances will give the same results as a Q mode PCA Requires MVSP v2 2 SCRIPT LANGUAGE REFERENCE 213 Options Transformation LOG10 Log base 10 LOGE Log base e LOG2 Log base 2 SORT Square root RATIO Logratio TRANSPOSE Transpose data Select a coefficient default EUCLID EUCLID Euclidian distance STEUCLID Standardized Euclidian distance COSINE Cosine theta distance MANHAT anhattan metric distance CANBER Canberra metric distance CHORD Chord distance CHISOR Chi square distance AVERAGE Average distance MEAN Mean character difference distance GOWER Gower general similarity coefficient Minimum eigenvalue default none KAISER Kaiser s rule JOLLIFFE Jolliffe s rule MINEIGEN real Specify a minimum eigenvalue ACCURACY real Accuracy of solution Output of graphics PLOT Text scattergrams GPLOT Graphic scattergrams PICTURE Syntax PICTURE NOx filename options Description The PICTURE command allows you to display Windows Metafiles WMF Windows bitmap BMP or icon ICO files T
180. ick the left mouse button again or press the underlined letter in the command you want Once in a main menu select a sub menu command by typing the underlined letter of the command name You can also use the Up and Dn arrow keys to highlight the command you want and press Enter to select your choice Use the Right or Left arrow keys to move from one pull down menu to another Press Esc or click anywhere outside the pull down menu to close it When a menu item is dimmed the command is currently unavailable For example when no data file has been opened several commands of the EDIT DATA and STATISTICS menu are dimmed Some menu items also display a keyboard key or key combination to the right of the command name When such a keyboard shortcut exists you can bypass the menu system and choose those commands by pressing their associated keyboard shortcuts Working with dialog boxes Some SIMSTAT commands in submenus are followed by an ellipsis three dots which indicates that a dialog box appears when you choose this command Dialog boxes allow you to specify various options used to adjust the program s operation to customize the analysis output or to specify conditions to be fulfilled by an analysis GETTING STARTED 7 Click on any item to position the cursor on it You may also use the following keyboard keys to navigate through dialog boxes Key Tab lt Shift Tab gt Effect Moves to the next
181. in frequency and crosstabulation analyses by treating responses stored in several variables as if they were stored in a single variable The dependent X and independent variable Y designations see Choose X amp Y command are used to gather all the variables together OPTIONS Independent dependent variables as multiple responses These two options allow you to specify which type of variable is to be treated as multiple responses Pressing the OK button accesses a second dialog box that allows you to alter any options normally available in the standard FREQUENCY CROSSTAB or INTER RATERS analysis Sample output of Multiple responses frequencies MULTIPLE RESPONSE FREQUENCIES NBREST1 Nb of meals in restaurants this week NBREST2 Nb of meals in restaurants last week Valid Valid Pet Of PED Or PED OT POETOE Value Frequency Responses Cases Responses Cases None 0 29 24 6 49 2 24 6 49 2 3E 53 44 9 ag n 44 9 DAC 2 36 S0 I EPIS EI TOTAL 118 199 9 20070 100 9 200 0 Mean i059 Variance ee Kurtosis i 1169 Median 1S0 Std Dev 743 Gd IDE 451 Mode 1 008 Std Rrr 068 Skewness eu DUE Minimum 2000 Sum 125 000 S E Skew M Maximum 2 000 Range 2 000 Valid IIS UD 95 Confidence Interval for the mean 9253 to OSA Sample output of a multiple responses crosstab analysis MULTIPLE RESPONSE CROSSTAB DAY Day or The week 126 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 127 Nonparametric matrix The NPAR MATRIX displ
182. ing file is much smaller than a bitmap format and the resolution of the graphic image becomes device independent In other words its resolution will be determined by the output device This file format may be inserted in most Windows word processing program modified using many drawing packages such as Corel Draw or Microsoft PowerPoint or paint programs e g Microsoft Paint Windows bitmap format BMP Charts saved in a bitmap format are stored as series of pixels of different colors The charts are stored as they are displayed on the screen Therefore their resolution is determined by the resolution of the screen as well as the size of the graph This means that greater resolution can be achieved by maximizing the Chart Window before exporting the chart to disk or to the clipboard However you should be aware that bitmap files take a lot of disk space and increasing the size of a bitmap image can increase its required disk space by several hundred kilobytes Like the Metafile format bitmap files may be imported into most Windows word processing drawing or painting programs Tab separated values format TAB Rather than exporting charts in a graphic format you may prefer to extract the values of a chart and use those values in another charting program such as Harvard Graphics Chart XL Microsoft PowerPoint or a spreadsheet program with advanced charting features Exporting a chart to a tab separated values format produces a
183. ing global options Selecting the GLOBAL OPTIONS command from the CHART pull down menu displays a multi page dialog box containing options that are shared by all charts Any change made to these options will affect all charts in the current Chart window These options include frame and series colors series markers titles and label font sizes and styles hart Preferences x Frame colors Series color Lines Trend or regression lines Smoothed curve Control bars Normal curve lines Colors fFonts Misc Adjusting the color of various chart components SIMSTAT allows you to change the color of various chart components such as the chart itself the frame and legend backgrounds the series or the reference lines drawn over some chart types To change the color of an item e Click on this item in the Chart Preferences dialog box This will display a standard color dialog box Select the new color and click on the OK button to leave the Color dialog box The Color Scheme option on the next page of the Chart Preferences dialog box allows you to use either colors patterns or both to differentiate the various series or bars in the chart 162 SIMSTAT for WINDOWS Adjusting line style and width You can change the line style of reference lines by clicking on the drop down arrow key of the proper Type list box and selecting the line style you want To hide a reference line se
184. ing unsatisfactory items or by removing redundant ones 1 e items that are equivalent to those that remain and thus provide no further information ASSESSING ITEM CONTRIBUTION TO INTERNAL CONSISTENCY Because a scale is developed to measure an attribute every question should tap an aspect of that attribute e g attitude behavior knowledge and ideally only that attribute Since all items must reflect the attribute they should share a common variance and should be related to the other items and consequently to the scale as a whole The correlation between an item and the total score omitting this item can be considered as an indication ofthis relationship Another desired attribute of any scale is its reliability expressed as the stability of the score when the same instrument or an alternate form is administered to the same individuals The Cronbach s alpha represents the most common index of the overall reliability or more precisely the internal consistency of a scale The alpha coefficient is also an indicator of the scale precision and can be viewed as the theoretical maximum value a correlation with this scale can take STATISTICAL ANALYSIS 101 Usually the longer the scale the higher the value of the alpha coefficient Consequently we may be tempted to always prefer a long version of a questionnaire to a shorter one There are at least two problems with this solution First it is often desirable for practical reasons to have
185. inimum display width for the entire data file This option has been implemented in order to circumvent a performance problem associated with the grid in SETTING THE PROGRAM PREFERENCES 239 the 16 bit version of SIMSTAT This speed problem occurs when working with large data files involving several hundred variables Increasing this value reduces the number of columns displayed and thus increases the redrawing speed of the grid 240 SIMSTAT for WINDOWS APPENDIX A XBASE SYNTAX AND FUNCTIONS 241 APPENDIX A XBASE SYNTAX AND FUNCTIONS The following section provides a description of xBase syntax rules used in the FILTER SORTING and COMPUTE commands and a detailed description of each xBase function NOTE This appendix was reproduced from the Apollo v2 0 user s manual with the authorization of Successware 90 Inc Expression Operators and Rules Operators used in xBase expressions are standard in every xBase dialect String Operators Joins two strings Trailing spaces in the strings are placed at the end of each string Joins two strings and removes trailing spaces from the string preceding the operator and places them at the end of the string following the minus sign operator Numeric Operators t Addition Subtraction Multiplication Division Exponentiation or Relational Operators Equal to Exactly equal to lt gt Not equal to Not equal to Not equal to lt Less than
186. inted on the left margin Also all text appearing to the right of the last gt character will be printed flush right Special codes can be inserted in the header to display the source data file f the time t and date d when the analysis was done or the notebook page number p 226 SIMSTAT for WINDOWS The VARSIZE and DECIMALS parameters are used to specify the default physical size and number of decimal places of newly created variables It is recommended to set those parameters before creating a new variable with the COMPUTE command The SET command also allows you to run a script in demonstration mode by using the SET DEMO integer command where integer stands for the number of milliseconds between dialog boxes In this mode BOX and QBOX no longer stop until the user presses ENTER or clicks on the OK button but will be displayed only for a specified length of time To disable the DEMO mode set the number of milliseconds to zero Options TOOLBAR ON OFF STATUS ON OFF HINTS ON OFF BEEP ON OFF SIZE integer DECIMAL integer 2m EADER string HARTSPERPAGE integer EMO integer Q SIGN Syntax SIGN varlist BY varlist Description Turns the main window s tool bar on off Turns the main window s status bar on off Turns the display of help hints on off Beep on error Default physical size of newly created variables Default number o
187. into account distribution properties that are generally considered as contaminating factors such as skewness ceiling effects outliers etc This feature makes bootstrap estimations adequate even when data are not normally distributed In fact bootstrapping can even be used to describe the sampling distribution of estimators for which sampling properties are unknown or unavailable SIMSTAT implementation of bootstapping SIMSTAT provides automatic bootstrap analysis for seven descriptive estimators of a single variable and almost thirty estimators involving two variables Those estimators are STATISTICAL ANALYSIS 69 One variable estimators Mean e Median Variance Standard deviation Standard error e Skewness e Kurtosis Two variables estimators e Kendall s tau a and b e Kendall Stuart s tau c e Symmetric and asymmetric Somers d Goodman Kruskal s gamma e Student s t and F Pearson sr Spearman s rs Regression slope and intercept Mann Whitney s U e Wilcoxon s W Difference between means Difference between variances e Sign test e Kruskal Wallis ANOVA Median test Percentage of agreement Cohen s kappa Scott s pi e Krippendorf s R and r bar e Free marginal nominal and ordinal levels The number of bootstrap samples for a single analysis can range from 10 to 100 000 The output of a simulation analysis can consist of various results including descriptive statistics frequency tables histograms
188. is i aide Te prese eter eg genet Meg estt oo e eet E 141 SIENTES jest e eb iade See RR ERO E E E ORO E ons dean CE SR 143 Single case design lt r ora EnA xu SAN eR SINE ee MEE eee A 144 Time series analysis i laete Red REA E MR een 146 T3t st analyslS nebst RES EYT Id UeEY RES SE EYES dde ems 149 Wilcoxon test sisi e mte e de et detti fed ete ibat pate Lb REL 152 7 Wotking with charts ossai ecc tee ee dears den de beue qox 153 Creatins specific Charts ascenso ee eie xevbedekeribfeierek teda yi 153 Navigating in the chart window eeeeeeee e 155 Customizing Charts E ESAE VER MIR EN Eee RE 158 Editing titles and axis labels 0 0 oe eee cee eee nee 158 Editing axis scaling and grid 1 2 eee eee 159 Controlling the legend display 0 0 0 cece eee 160 Displaying data point values 0 eee eee eee ene 160 Modifying specific chart options 0 0 eee eee eee eens 160 Setting global options 0 0 cece eee tee eee nes 160 3D VIEW ae perene niae a de e e Rede ke CR de seg tales GS Ge E fa te ed 162 Zooming mand out 2 2 ru Ar IE ER A RS oe Sew nb E 163 Expottitie charts i25 uus uu eL A ERE be ee d 165 8 Usine SCEIDIS o oer oec ng m eod este e ufi aah atten ere ee se ak hale fees 167 Introduction to the scripting feature 0 eee eee eA 168 Using normal and encrypted script files 02 0 cece eee eee 168 Opening an existing script 2 0 ee eee cette I e 16
189. is extension to the end of the file name Enter the name of the file under which you wish to save the notebook and press Enter or click on the OK button to save the file Choose the CHART SAVE command from the FILE menu to save all charts in the Chart window in a file name The default extension for chart files is CHX If no extension is given the program automatically adds this extension to the end of the file name To print the numerical results or the charts Activate the Notebook window to print numerical results or the Chart window to print charts Select the PRINT command from the FILE menu or click on the printer button on the tool bar Set the desired options and click on the OK button to start printing Note By default SIMSTAT is configured to start printing each notebook page at the top of a new page and to print 2 charts per page To change these options select the PREFERENCES command from the FILE menu see section Setting Program Preferences at page 235 Step 8 Exiting SIMSTAT To exit from SIMSTAT choose the EXIT command from the FILE menu If during a session the contents of any window have been modified and have not been saved then dialog boxes will appear asking you if you would like to save those changes to disk Choose Yes to save the changes to the documents and exit SIMSTAT or choose No to continue to exit without saving the documents 20 SIMSTAT for WINDOWS 3 DATA FILE OPERATION
190. is formula can contain existing variable names numeric constants arithmetic operators or any supported numeric trigonometric or statistical functions You can type the numeric transformation in the Formula edit box directly using the proper syntax or use any element displayed on the upper part of the Variable Transformation dialog box to build a valid expression To restore previously used numeric transformations click on the down arrow button located at the right of the Formula edit box Oncea valid target variable and a numeric expression have been entered you can leave this dialog box and perform the transformation by clicking on the OK button If the transformation expression is invalid a message is displayed and the dialog box remains displayed Toexit from the dialog box without performing a transformation click on the CANCEL button Building a valid transformation expression The upper part of the dialog box contains various elements to help you build a valid numeric expression Variable name list box Double clicking a variable name from the list box located to the left of the dialog box inserts that name in the edit box at the current caret position Function list box A list of valid numeric functions is displayed to the right of the dialog box Double clicking on a function name in the list box inserts that function at the current caret position When a function requires one or several arguments the argument section remains
191. isplay Graph Window option is enabled see below and new charts are created during the analysis the chart will become the active window Display graph window on graph creation When this option is enabled the Chart window automatically become the active window after a new chart is created Activate script recording on start up Selecting this option enables the RECORD script feature This feature automatically generates script commands for actions undertaken using pull down menus and dialog boxes and appends those commands to the end of the script window DIRECTORIES BACKUP TAB Data files directory By default when the DATA OPEN command is activated the program looks in the drive and directory from which the program was started This option allows you to specify another default drive and or directory to be used by the program Output files directory The Output Files directory specifies the name of the directory where both notebook files and chart files are located If no directory information is provided the dialogs of the OPEN and SAVE AS commands start on the disk and directory from which SIMSTAT was started Script files directory The Script Files directory specifies the name of the directory where the Script files are located If no directory information is provided the dialogs of the OPEN and SAVE AS commands start on the disk and directory from which SIMSTAT was started Backup compression factor This option allows
192. ist options Description The T TEST command calculates either independent sample t tests GROUP or paired sample t tests PAIRED to decide whether two sample means are significantly different The paired sample or correlated t test compares the means between each pair of dependent and independent variables The independent t test compares means on the dependent variable for two groups defined by values of the independent variable SIMSTAT provides two distinct tests to take into account whether the two populations from which the samples are drawn have equal or unequal variances You can also specify whether the null hypothesis should be evaluated using a one or a two tailed test 228 SIMSTAT for WINDOWS Options GROUP PAIRED Grouped or paired t test VALUE integer integer Values of X CI integer Confidence interval width 1TAIL 2TAIL Direction of the test ERRORCHART Error bar graph BARCHART Display mean bars UPPER Display upper error bar only barchart CIBAR Confidence interval SE Standard error SD Standard deviation HISTOGRAM Dual histogram NBAR integer Number of bars intervals NORMAL Normal curve VERTICAL Vertical bars PANEL Display dialog box TIME SERIES Syntax TIME SERIES varlist options Description The TIME SERIES command allows the examination oftime series Available options offer various transformations to remove trends or seasonal dependence in a series and
193. istics including summary statistics crosstabulation inter rater agreement statistics frequency and breakdown analysis n way analysis of variance and covariance paired and independent t tests linear nonlinear and multiple regression analysis time series analysis and many nonparametric analyses Also SIMSTAT provides powerful bootstrap simulation analyses These simulations are based on resampling procedures and can be used to provide nonparametric estimates of sampling distributions to assess the stability of multivariate solutions or to perform nonparametric power analysis The program also includes a powerful script language that allows automation of statistical analyses and creation of interactive tutorials demonstration programs with multimedia features and even computer assisted testing or interviewing programs Using this manual This user s manual provides all the information you ll need to install and use SIMSTAT The manual assumes that you are familiar with the basics of operating an IBM PC or compatible under Windows 3 1 or later If you are unfamiliar with mouse actions such as double clicking and drag and drop operations or if you re not familiar with windows management consult your Windows user s manual This manual is not intended to teach statistics and assumes that you have sufficient knowledge of statistics to choose the analysis appropriate for your needs The manual is divided into the following sections Introdu
194. ited it is also possible to add empty pages for recording ideas or remarks sketching an analysis plan or writing down interpretation of results 3 Notebook UNTITLED Of x ml m FREQUENCIES CYLIND Valid Frequency Percent Percent 4 1 0 51 0 3 t 7 6 20 26 26 TOTAL 0 100 0 Variance Kurtosis Std Dev 8 H Kurt Skewness S E Skew Valid The following section provides basic instruction to navigate in the notebook and edit its contents use tabs and the index to manage output add or remove notebook pages round numeric values WORKING WITH THE NOTEBOOK 61 Navigating in the notebook To move the caret to a specific location you can click with the mouse on that location You can also use the following keys to navigate in the notebook Up Move one line up Down Move one line down lt PgUp gt Move up one screen If the cursor is on the last line of the page pressing this key moves to the next page lt PgDn gt Move down one screen If the cursor is on the first line of the page pressing this key moves to the previous page lt Home gt Move to the beginning of the line lt End gt Move to the end of the line lt Ctrl Home gt Move to the first line lt Ctrl End gt Move to the last line lt Ctrl Right gt Move to the beginning of the next word lt Ctrl Left gt Move to the beginning of the previous word lt Ctrl Enter gt Insert a page break It is possible to loc
195. itted iteratively to assess the effect of each variable This option is not available when the model contains interactions Confidence interval This option specifies the width of the confidence intervals for the coefficients Tolerance This option specifies the convergence criterion Iterations cease when the largest relative change in any coefficient between successive iterations is smaller than the specified tolerance The maximum number of iterations is set at 50 and cannot be altered 114 SIMSTAT for WINDOWS Mann Whitney U test The MANN WHITNEY U test evaluates the hypothesis that two independent samples have the same distribution The Mann Whitney U is the nonparametric version of the t test for independent samples This test is performed on the dependent variable divided into two groups as defined by values of the independent grouping variable OPTIONS Values of x This option requires two independent variable values The cases that match the first value on the independent variable form one group and the cases that match the second value form the second group Direction This option allows you to select either a one tailed directional or two tailed non directional probability test Sample output of a Mann Whitney U Wilcoxon rank sum W test analysis MANN WHITNEY U WILCOXON RANK SUM W TEST AGGRESS Nb of agressive behavior With SEX Sex of the child Mean Rank Cases 40 48 20 epe 3L O10 19 87 30 SEX
196. l appear in it e Set the Sort By option to Value and the Type option to Ascending Activate the Descriptive Statistics check box Deactivate the Confidence Interval and Percentiles Table check box The dialog box looks similar to the one shown above All graphing options are located in the second page of the dialog box To activate this page click on the Charts tab at the bottom of the dialog box To produce a histogram make sure that the histogram check box is the only one containing a check mark If other options are enabled click on those check boxes to deactivate them After setting the proper options you must activate them and tell SIMSTAT to perform the analysis by selecting the OK button SIMSTAT computes the requested statistics and appends the numerical results to the Notebook window while the histograms are automatically added to the Chart window 16 SIMSTAT for WINDOWS Step 4 Viewing the results To view the numerical output of the analysis just performed make the notebook window active by either clicking on a visible part of this window selecting it from the WINDOWS menu or clicking on the notebook icon at the right end of the toolbar The notebook displays the last page which contains the frequency table and descriptive statistics of SIBLING Use the arrow keys i e Up Dn Left and lt Right gt to navigate through this page If you prefer you may also set the program to display horizontal and v
197. labels as well as some predefined strings to provide such descriptions To add or edit those titles select the TITLES command from the CHART menu The following dialog box will appear itle editing x Titles Top Error rate curves of AGGRESS Percentage of error Bottom Aggressivity of the child OK x Cancel Help This dialog box provides 4 edit fields where you can create or edit the existing chart title and the labels on the left bottom and right axis You can enter several lines of text for each title by pressing the Enter key at the end of a line before entering the next line Font buttons on the right side of each edit box allow you to quickly change the font size or style of the related title Please note that the font setting is a global option and will be applied to all charts in the Chart windows WORKING WITH CHARTS 159 Editing axis scaling and grid The AXIS command on the CHART menu allows you to specify various axis options such as the axis limits increments and the number of decimal places used to display axis values You can also control the display of horizontal and vertical grids When this command is invoked the following dialog box appears Axis setting x Axis selection X OK C X axis ei voc Scaling Limits X cancel Linear Minimum 24 Help C Logarithmic am Increment 2 Decimal DE Grid Horizontal grid Solid
198. lays a bitmap file on screen Prints the contents of a window Saves the contents of a window on disk Creates a blank background screen Displays a text string at the top of the screen Controls the size and location of any SIMSTAT window Command descriptions APPENDFROM Syntax AP Description PENDFROM filename SCRIPT LANGUAGE REFERENCE 181 The APPENDFROM command appends data from an existing DBF data file to the current data file All variables with matching names and types are appended to the current file If variable length differs data in the destination data file is either truncated or padded with spaces Deleted records in the source file are not appended to the target file Example OPE APP BEEP N C NSIMSTATNDAT l ANANNUAL DBF ENDFROM C SIMS Syntax BE EP Description ATNDATANDE CEMB ER DBF Generates a beep signal through the computer s speaker BINOMIAL Syntax BINOMIAL varlist BY varlist Description options The BINOMIAL test allows you to assess whether the observed number of cases in a dichotomous variable is the same as that expected from a specified binomial distribution The observations can be divided either below or above the mean the median or a user specified cutoff value Alternatively the analysis can also be restricted to cases equal to two specified values The user can also specify the test pr
199. le including between and within groups sums of squares mean squares and degrees of freedom F ratio and its associated probability You can also obtain for each group descriptive statistics including count mean standard deviation standard error and a user specified confidence interval for the mean Various measures of effect size and post hoc comparisons can also be obtained SCRIPT LANGUAGE REFERENCE 211 as well as various graphs such as a barchart representing the mean of each group an error bar diagram or a deviation chart where each bar represents either the standard deviation the standard error or a user specified confidence interval Options RANGE integer integer DESC CI integer LSD N K TUKE SCHEFFE ERRORCHART BARCHART UPPE CIB Si S N D R AU E DEVCHART lgo D z Eri E OPEN Syntax OPEN filename Description Minimum and maximum values Descriptive statistics Confidence interval width Post hoc test Least significant difference Newman Keuls Tukey s H S D Scheff test Error bar chart Displays mean bars Displays only upper portion of error bars Confidence interval Standard error Standard deviation Link means Deviation chart Display the dialog box The OPEN command reads a file The extension of the file is used to determine what kind of file should be opened When the file name includes a DBF extension SIMS
200. lect None as the new line style When a solid line style is chosen you can use the Width spin button to set the thickness of the line to a value between 0 and 9 Setting the font size and style of charts components To change the title font the axis labels or the value labels Select the second page of the Chart Preferences dialog box by clicking on the Fonts Misc tab n the String list box select the text component you want to modify by clicking on its name Change the font name size color or style Miscellaneous chart options Color scheme This option allows you to use either colors patterns or both colors and patterns to differentiate the various series or bars in a chart 3 D frame When this option is enabled SIMSTAT draws a 3D frame Point marker size This option lets you specify the size of point markers for all series of a chart Marker size must be set between 1 and 20 Default legend location This group of options lets you specify whether to display a legend when a new chart is created and its default location These options apply to piecharts bar charts and Pareto charts 3D View Some charts can be diplayed using a 3D perspective When such an option is available the button on the toolbar is enabled You can click on this button to turn on off the 3D perspective for the current chart Selecting the 3D VIEW command from the CHART menu or clicking on the e button gives access to the 3D View
201. line M Axis selection This group of radio buttons allows you to specify the appropriate axis that you want to customize In some charts only the Y axis will be available Linear or Logarithmic scale In some types of chart you can choose between a linear or logarithmic axis In a linear scale the value of each major division is exactly the same In a logarithmically scaled axis each major division of the axis represents 10 times the value of the previous major division Minimum and Maximum SIMSTAT automatically adjusts the axis scales to fit the range of values plotted against it To manually set these values type the desired minimum and maximum for the axis selected Increment SIMSTAT automatically selects the initial increment value used for the axis By default this increment value is set to display 10 tick marks per axis Increasing or decreasing this value affects the distance between these tick marks as well as their number Grid lines are also affected by modification of this value Decimal This option allows you to increase or decrease the number of decimal places used to display the values on the axis Grid lines This option lets you turn horizontal Y axis and vertical X axis grid lines on and off Grid lines extend from each tick mark on an axis to the opposite side of the graph To increase or decrease the number of grid lines per axis or the distance between those lines change the Increment value of t
202. llowing conventions are used throughout the manual MONOSPACE is used to indicate what is to be typed on the keyboard bold text is used to highlight keywords and identify text displayed by the program e Words between the less than and the greater than gt characters indicate a key to be pressed For example Enter mean you have to press the Enter key A dash between key names means you have to press two keys simultaneously For example Ctrl x means that you have to press the lt Ctrl gt key and hold it down while you press the x key GETTING STARTED 3 1 GETTING STARTED In this chapter we ll get you started using SIMSTAT by showing how to install the program and how to perform some basic operations This chapter also includes a short tutorial that will guide you through an easy to follow example of an analysis This tutorial will allow you to quickly become familiar with the steps required to perform statistical analyses If you are already familiar with the operation of SIMSTAT you may skip this section The SIMSTAT package Your SIMSTAT package includes the following the SIMSTAT user s manual two 3 inch program disks the Provalis Research license agreement and registration card send this in to become a registered owner and receive technical support and information about upgrades and related products System requirements SIMSTAT requires a computer running Windows 3 1 or later 4MB or
203. lot OUTLIERS real Outliers criterion s d DURBIN Durbin Watson statistic PLOT Residuals scatterplot PPLOT Residuals normality plot SAVE Save predicted and residuals Adjustment method REGRESSION Regression approach NONEXP Nonexperimental HIERARCHICAL Hierarchical ORDER varlist varlist Order of entry hierarchical varlist All variables enumerated after the character are entered with the interactions PANEL Display the dialog box GOTO Syntax GOTO label Description This command branches to another part of the program and continues processing the commands at that point The line that the program is to switch to is marked with a label preceded by a colon Example GOTO JanuaryStat JanuaryStat 198 SIMSTAT for WINDOWS GOSUB Syntax GOSUB label Description This command branches to another part of the program and continues processing the commands at that point until the RETURN keyword is encountered The program then continues execution at the statement following the GOSUB command IF THEN ELSE Syntax IF expression THEN command ELSE command Description The IF command allows you to specify a condition that must be met for a command to be carried out If the condition if true the first command is executed If the condition is false the command appearing after the ELSE keyword is executed If the keyword ELSE is not used nothing is performed and the execution contin
204. ls along the vertical axis The graph can be printed in text mode graphics mode or both Residual normal plot This option displays a normal probability plot of the residuals that allows you to evaluate whether those data are normally distributed If the residuals follow a normal distribution the data points will fall approximately along a straight line going from the lower left corner of the graph to the upper right corner Save residuals This option instructs SIMSTAT to save the predicted and residual values as new variables in the current data file Creating those variables allows you to perform further analyses such as displaying scatterplots between each independent variable and the predicted and residual values The new variables are named PREDxxx and RESIDxxx where xxx stands for a serial number between 001 and 999 that corresponds to the number of regression analyses performed during a single command This number is automatically reset to 001 after each command To prevent overwriting those variables you must rename them using the DEFINE VARIABLE command STATISTICAL ANALYSIS 135 Sample output of a regression analysis 136 SIMSTAT for WINDOWS Reliability analysis Multiple item additive scales are often used to measure various characteristics of a subject One desirable feature of such scales is their reliability The reliability of a scale refers to the consistency of the scores obtained when the scale is administered to
205. lyses in SIMSTAT require you to distinguish between dependent and independent variables some analyses do not require such a distinction This is the case for several commands involving a single variable per analysis such as LIST DESCRIPTIVE FREQUENCY TIME SERIES BINOMIAL CHI SQUARE TEST and RUNS TEST With those commands variable assignments to the dependent or independent list boxes are simply ignored and the procedure is applied successively to all the independent variables followed by all the variables assigned to the dependent list box The RELIABILITY the ITEM ANALYSIS and all multivariate analyses available from the OTHER submenu i e Factor analysis cluster analysis etc also do not make a distinction between dependent and independent variables and will include all variables in both lists in a single analysis However when performing a split half reliability test using the RELIABILITY command the two list boxes are used to determine which variables will be included in each versions of the instruments To remove variables from the dependent or independent list Select the variables you want to remove and click on the El button To quickly remove a single variable from the list double click on the variable name To apply an integer weight to your cases SIMSTAT allows you to select a variable that will be used to weight the cases When SIMSTAT reads a case the value of the weighting variable for this case is truncated to an int
206. ma or tab delimited text The program can read data stored in the following formats SIMSTAT for DOS v1 0 v3 5 SPSS PC SPSS for Windows PARADOX v3 0 v5 0 LOTUS 123 v1 0 v5 0 SYMPHONY v1 0 v1 1 EXCEL v2 1 v5 0 QUATTRO PRO v1 0 v6 0 COMMA SEPARATED VALUES Windows and DOS TAB SEPARATED VALUES Windows and DOS To import data from any of these applications Choose the DATA IMPORT command from the FILE menu Select the file format using the List File of Type drop down list Select the file you want to import and click OK You will find below detailed information about specific file formats such as their formatting requirements if any the supported features and limitations SIMSTAT for DOS files SIMSTAT imports files from its DOS version with SIM extensions The program will import variable and value labels as well as user missing values associated with each variable SPSS PC and SPSS for WINDOWS files SIMSTAT reads compressed and uncompressed SPSS PC system files with SYS extensions The program also supports variable and value labels as well as user missing values associated with each variable SPSS PC files are saved in uncompressed format SIMSTAT can also read compressed and uncompressed SPSS for Windows data files SAV extension containing less than 1022 variables The program will import variable and value labels as well as user missing values associated with each variable D
207. manage those archive files outside of SIMSTAT using any application than can manipulate these files Also the high level of compression achieved on typical SIMSTAT data files usually more than 90 95 allows you to create backup copies of large data files on a single diskette or keep several copies of your data file on your hard drive without sacrificing precious disk space Another benefit of this archiving feature is to facilitate the transfer of your data file and all its related files to another computer by creating a single file that includes all related files and can easily fit on a single diskette To make an archived copy of the current data file Select the DATA ARCHIVE BACKUP command sequence from the FILE menu A Save File dialog box will be displayed Set the drive and directory setting to the location where you want to store the compressed file e Enter a valid file name and click OK To restore a file from an archived copy Select the DATA ARCHIVE RESTORE command sequence from the FILE menu An Open File dialog box will appear displaying all ZIP files Select the proper ZIP file and click on the OK button A second dialog box will appear to let you specify the location where data should be restored The default location is the same directory as the archive file f needed change the drive and directory setting and click on the OK button to proceed with the extraction 46 SIMSTAT for WINDOWS Crea
208. meter of this function DATA FILE OPERATIONS 37 Once a valid sorting expression has been entered you may use the Sort Order option to specify whether the records should be sorted in ascending or descending order To execute the sorting and leave the dialog box click on the APPLY button If the sorting expression is invalid a message is displayed and the dialog box remains displayed To temporarily deactivate the current sorting expression click on the IGNORE button The sorting string is kept in memory and may be reactivated by choosing the SORTING RECORDS command again Press the CANCEL button to abort the current sorting operation leave the dialog box and restore the previous sorting order Examples When the following sorting expression is used BIRTHDAY and the Ascending radio button is activated the records are sorted in ascending order by the variable birthday In the following sorting expression S EX 100 D ESC END AG E where the variable SEX is a numeric variable 0 male and 1 female SIMSTAT sorts the records with male subjects at the top of the data grid ordered from the oldest to the youngest and all females at the bottom of the grid in the same order 38 SIMSTAT for WINDOWS Importing data from other applications SIMSTAT allows you to import data files from other statistical programs spreadsheet and database applications as well as from plain ASCII data files com
209. mmand may produce several megabites of bitmap files USING SCRIPTS 167 8 USING SCRIPTS The script window is used to enter edit and execute commands Those commands can be read from a script file on disk typed in by the user or automatically generated by the program When used with the RECORD feature the script window can also be used as a log window to keep track of the analyses performed during a session Those commands may then be executed again providing an efficient way to automate statistical analysis Additional commands also allow one to create demonstration programs computer assisted teaching lessons and even computer assisted data entry programs Script file UNTITLED Mie x OPEN D SIMSTAT DATA SAMPLE DBF FREQUENCY SIBLING HOURSTU AGGRESS TRBLE FREQ DESC BOXPLOT HISTOGRAM NBAR 26 T TEST AGGRESS by SEX GROUP 8 1 2TAIL CI 95 ERRORCHART SE HISTOGRAM NBAR 15 INPUT NAME AS STRING What is your name PLAY AIPLAINE AUI PLAY MUSIC WAU QUESTION SITEH1 What does the P value stands for ANSWERS f The probability that the null hypothesis is true B The probability that the null hypothesis is false C The probability of the data given the null hypothesis is ALI This section introduces you to various tasks that may be performed with script files such as how to Open an existing script file e Navigate in the script window and edit its contents Execute a script or only part of it Use t
210. n allows you to specify whether the error bars displayed with the mean bars should be displayed above and below the mean or only above it Link means When chosen this option connects the means with a line Deviation chart This option displays a bar chart where each bar represents the deviation of the group mean from the grand mean 132 SIMSTAT for WINDOWS Sample output of a one way analysis of variance Regression analysis STATISTICAL ANALYSIS 133 REGRESSION produces a simple regression analysis for each pair of dependent independent variables The output includes the Pearson product moment correlations the intercept and slope of the regression line and an ANOVA table for the equation The dialog box allows you to obtain bivariate scatterplots to select one tailed directional or two tailed non directional tests of probabilities and to request standardized residuals plots OPTIONS Type of analysis This option allows you to choose between 8 types of nonlinear regression The following types of regression will allow the assessment of different degrees of curvilinearity among the dependent independent relations or to obtain an equation that expresses various forms of bivariate relations The following table presents the types of regression and their corresponding equations TYPE REGRESSION EQUATION Linear Y a bx Quadratic Y a bx bx Cubic Y a bix b x b x 4th degree polynomial Y a b 5th degree
211. n the middle of the screen The TOP and LEFT options can also be used to specify the position of the window s upper left corner The parameter for these two options is an integer value between 0 a 100 expressing a percentage of the screen height and width Options TOP integer Vertitical position of the AVI movie Screen s upper left corner 0 and 100 LEFT integer Horizontal position of the AVI movie Screen s upper left corner 0 and 100 Examples PLAY C NWINDOWSNTADA WAV PLAY SONATA MDI PLAY INTRO AVI TOP 10 LEFT 20 SCRIPT LANGUAGE REFERENCE 215 PRINT Syntax PRINT windowtype options Description The PRINT command can be used to send the contents of a window to the printer By default all contents are printed To restrict the printing to some pages of a notebook or to a specific number of charts use the FROM and TO options to specify the lower and upper limits Options Window types DATA Data spreadsheet window NOTEBOOK Notebook Statistical results window SCRIPT Script log window CHART Charts window FROM integer Start printing at page chart TO integer End printing at page chart Examples PRINT NOTEBOOK PRINT CHART FROM 3 TO 5 QBOX Syntax QBOX string options Description The QBOX command allows you to display a single line message or question to the screen By default the window is positioned in the middle of the screen The LEFT and TOP options can also be used to spe
212. nce and correlation matrices total scale and item total statistics It also allows you to verify the reliability of the scale through the use of a split half method or by computing internal consistency measures Each selected variable is considered as a single item of the scale The first and second lists or variables can be used to specify the division of items in two different subscales to be used in a split half method Options ITE Item statistics CORR Inter item correlation matrix COVAR Variance covariance matrix TOTAL Item total statistics SPLIT Split half reliability ALPHA Cronbach s Alpha PANEL Display the dialog box RETURN see GOSUB 222 SIMSTAT for WINDOWS RUN Syntax RUN program parameters Description The RUN command runs another Windows or DOS program If the program is not in the program directory or in the current script directory you will need to specify the full path of the program Examples RUN SIMCALC EXE RUN C WINDOWS WRITE EXE MYDOC WRI RUNS Syntax RUNS varlist BY varlist options Description The RUNS test is a procedure to test whether the ordered sequence in which observations were obtained is random In order to be performed such a test requires that all values be dichotomized into two categories The options allow you to separate observations into two distinct groups using the mean the median or a user specified value as a cutoff point
213. nd one tailed probability of the span with and without outliers OPTIONS Values of x This option requires two independent variable values The first value of the independent variable defines the control group and the second value defines the comparison group Outliers to remove This option allows you to determine the percentage of extremes cases to be excluded from the analysis If this field is left blank 596 of the cases are trimmed from each end of the range of the control group to remove outliers Sample output of a Moses test of extreme reactions analysis MOSES TEST OF EXTREME REACTIONS HOURSTV Hours per week spent watching TV With SEX Sex of the child Cases CONTROL 29 SEX 1 00 EXPERIMENTAL 20 SEX 2 00 59 oueu 1 tailed P Span of Control Group 20299 52 Observed 2633 5 After removing 1 Outlier s from each end 118 SIMSTAT for WINDOWS Multiple regression Multiple regression analysis is a statistical technique that allows you to assess the relationship between one dependent Y variable and several independent variables also called predictors For example you may want to predict the aggressiveness of children from several independent variables such as gender age number of siblings and the time spent watching TV The technique can be used to analyze how various combinations of independent variables correlate with one anothers as well as with the dependent variable Multiple regression is an exten
214. not accepted FAX 514 899 1750 Compuserve E mail X simstat Internet E Mail support simstat com Internet Web site http www simstat com Standard mail Provalis Research 2414 Bennett Street Montr al QC HIV 3S4 CANADA INDEX A ACH plot s oie bib ec eee ede A 146 228 Agreement measures 000 000 0085 97 99 Alpha Bootstrap analysis 69 70 72 183 217 Cronbach s Alpha 100 107 137 139 221 247 ANCOVA co biker ENERO 90 93 196 References 2 0 0 0 0 eee 250 251 ANOVA Friedman test 0 0000 89 195 GLM ANOVA ANCOVA 90 93 196 Kruskall Wallis 005 110 202 Multiple regression 119 120 208 ONEWAY 0 eee 130 210 Regression analysis 00 133 220 Ref rences ce xRLKC eiui 250 251 Append data i2 neue ama axes 42 181 Archival backup of data files lesse 45 Archives 1105 mau REA RU ona 22 45 ASCH files pats tt Ne EC 39 42 191 Autocorrelation 0 000 eee eee 146 148 AVIfiles xcLlim oR AG a Sh aia a 180 213 BRIS cat Ped a had ded 158 159 162 164 186 193 B Background color 00e eee eee 162 224 B ack p 3 sas bae a a ee 3 45 236 Backward selection 00005 119 208 Barchart Crosstabulation 000 78 79 191 Frequency 0 0 00 cece 86 87 195 Inter raters analysis 0 98 99 200 Binomial test 00
215. ns regardless of the distance between the group means This test keeps the experiment wise error rate equal to a The Scheff s test is used to obtain simultaneous confidence intervals for differences between all pairs of means while keeping the overall error rate to It is a more conservative test and will lead to fewer significant STATISTICAL ANALYSIS 131 differences This test can be used with equal or unequal group sizes and with planned or unplanned comparisons Confidence interval This option allows you to set the confidence interval width for the means and for the pairwise comparisons of those means post hoc tests This interval width is expressed as a percentage and must lie between 1 and 99 This value is also used when displaying bar charts with a confidence interval or an error bar graph representing a confidence interval around the mean Mean Error bar graph This option displays a mean bar and or an error bar representing the variability of the mean or the values in each group Type of error This option allows you to select whether the error bar will represent the standard deviation the standard error or a user defined confidence interval The width of this interval is set by the interval option With bar chart This option allows you to draw solid bars where each bar represents the mean of a separate group Upper error bar only When the bar chart option is selected and an error bar is requested this optio
216. nt to Programs point to the SIMSTAT for Windows item and then the SIMSTAT for Windows program icon Startup options You can edit the program properties dialog box to add a script filename to the SIMSTAT s parameter list When you do this SIMSTAT will automatically execute the script when it starts up By creating shortcuts or icons of the program with different script file names as parameter it becomes possible to execute those scripts from the desktop with a single mouse click The working environment When you call up SIMSTAT the first screen you see is the introductory splash screen with product version information This screen disappears and brings you to the working environment see below which is divided into four parts ool bar Menu bar E 8 Simstat for Windows File Edit Data Statistics Chan Script Tools Windows Help alel s beee e aea Chart UNTITLED lt Script UNTITLED Notebook UNTITLED buick skylark 320 C plymouth satellite Status panel Vorking space GETTING STARTED 5 1 Themain menu bar at the top of the screen gives access to nine pull down menus from which commands can be evoked 2 The tool bar provides quick mouse access to frequently used commands 3 Thestatus panel located at the bottom of the screen displays various information about the program s operation 4 The central part of the screen is the working space SIMSTAT s working space contains four windows Data win
217. nt to replace its values with those produced by the transformation Ifthe variable does not exist the program will ask you to confirm the creation of this new variable If you answer yes to this question the new variable will be appended to the left end of the data sheet see Setting Program Preferences on page 235 for information on how to disable these confirmation dialogs When a new numeric variable is created the program needs to know in advance its physical size and precision The current default size and number of decimal places used to store the result of a transformation is DATA TRANSFORMATION 51 indicated in the upper right corner of the Transformation group box To modify these values click on the precision dialog box and enter the new default size and number of decimal places Conditional transformation This check box allows you to choose between a simple computing formula or a conditional transformation When this option is disabled a single edit box i e Formula is shown and the transformation is performed on all currently selected records Enabling this option allows you to perform a conditional transformation using an IF THEN ELSE logical structure The single edit box is replaced by 3 edit boxes allowing you to specify a logical condition and two transformation formulas see conditional transformation below Formula This field contains the computing formula used to compute the value of the target variable Th
218. ommand allows you to obtain frequency analyses and crosstabulation analyses on variables which can legitimately have more than one response These multiple responses are stored in as many variables as necessary Choosing all these variables as dependent X or independent Y variables allows you to gather all these responses and treat them as if they were stored in a single variable 208 SIMSTAT for WINDOWS Options MULTX MULTY PANEL FREQ varlist CROSSTAB varlist BY varlist Example options MRESPONSES MULTY Treat X as multiple responses Treat Y as multiple responses Display the dialog box options CROSSTAB INCOME1 INCOME2 BY SEX TABLE DFREQ MULTREG Syntax MULTREG varlist BY varlist Description options The MULREG command allows you to perform multiple regression analysis to predict a dependent variable from many independent variables SIMSTAT provides various regression methods including standard regression hierarchical entry forward selection backward selection stepwise selection The options also allow you to display a wide range of statistics and perform various tests on residual values Options HIERARCHICAL FORWARD BACKWARD STEPWISE STANDARD PIN real POUT real STEP TOLERANCE real Type of regression Hierarchical entry Forward selection Backward
219. on New York Holt Rinehart and Winston 252 SIMSTAT for WINDOWS GENERAL LINEAR MODEL amp ANALYSIS OF VARIANCE COVARIANCE COHEN J 1968 Multiple regression as a general data analytic system Psychological Bulletin 70 426 443 COHEN J amp COHEN P 1983 Applied Multiple Regression Correlation analysis for the Behavioral Sciences 2nd Ed Hillsdale N J Lawrence Earlbaum IVERSEN G R amp NORPOTH H 1976 Analysis of Variance Beverly Hill Sage Publication PEDHAZUR E J 1982 Multiple Regression in Behavioral Research 2nd Edition New York Holt Rinehart and Winston OVERALL J E amp SPIEGEL D K 1969 Concerning least squares analysis of experimental data Psychological Bulletin 72 311 322 ITEM ANALYSIS CROCKER L amp ALGINA J 1986 Introduction to classical amp modern test theory Fort Worth Harcourt Brace Jovanovich College Publishers EBEL R L 1965 Measuring educational achievement Englewood Cliffs Prentice Hall HENRYSSEN S 1971 Gathering analyzing and using data on test items In R L Thorndike Ed Educational measurement 2nd Edition Washington D C American Council on Education STREINER D L amp NORMAN G R 1989 Health measurement scales A practical guide to their development and use Oxford University Press SINGLE CASE EXPERIMENTAL DESIGN HERSEN M amp BARLOW D 1976 Single case experimental designs Strategies for studying be
220. on it may be inappropriate to use some of the techniques of item selection Examples of this situation include mastery learning or any other instructional intervention that attempts to teach a new skill and for which achievement is measured using a criterion referenced test In these situations a test administered prior to the instructional intervention would yield a very low percentage correct for each item and item total correlations near zero reflecting the fact that examinees knew nothing about the subject matter and responded almost randomly to the test These items would indicate the examinees do need such training so that discarding the items based on their low discriminatory value or their low correlation with the total score would clearly be inappropriate Furthermore performing the same analysis after a successful training would produce very high percentages of success An item successfully answered by a great majority of individuals would be considered to have little or discriminatory value and would produce either a low or an indeterminate item total correlation Again discarding this item would mean the removal of a question that measures the mastery of the skill which is exactly what we want to assess In this situation using the internal consistency index would also be inappropriate since in the ideal situation where an instructional intervention brings about for all participants perfect mastery of a skill that was nonexistent before wo
221. on barchart Overlapping bars Stacked bars 100 stacked bars 2 D perspectiv 3 D perspectiv Display the dialog box The ITEM command perfroms a classical item analysis for multiple choice item questionnaires Requires STATITEM v1 0 Options KEYFILE T LISTWISI ISTAT ALTERNATIVE HILOW integer D IRATI ICC SMOOTHED CI S BREAKDOWN integer TSTAT TFRI GI Q SAV PAN E pd E Specify that key responses are stored in a key file Eliminate cases with missing values Display item statistics Display statistics for alternate items Set the discrimination index to a percentage between 10 and 50 Display item total response rates Display item characteristic curves Apply smoothing to item characteristic curves Display an endorsement breakdown table for up to 10 different groups Display detailed statistics of total scores Display a frequency table of total scores Save the total scores in a data file Display the dialog box 202 SIMSTAT for WINDOWS KRUSKAL Syntax KRUSKAL varlist BY varlist options Description The KRUSKALL command performs a Kruskall Wallis one way analysis of variance by ranks is a procedure for testing whether k groups have been drawn from the same population This test is a nonparametric version of the one way analysis of variance The output displays the number of valid cases the me
222. onding character to upper case comma or a space in Europe in a numeric template separate the elements of a number period or comma in Europe in a numeric template specify the decimal position fills leading spaces with asterisks in a numeric template as the leading character in a numeric template results in a floating dollar sign being placed in front of the formatted number Example Where phone is a character field holding a phone number with no formatting characters transform phone QR 444 returns 909 699 6776 If the formatting characters were actually present in the field the 9 R function would be omitted For numeric fields transform 123456 78 9 999 999 99 returns 123 456 78 TRIM String Removes trailing spaces from the string expression UPPER String Converts the string expression into upper case Character fields used in index expressions should always be converted to upper case to insure correct collating sequence VAL String Converts a string of numeric characters into its equivalent numeric value The conversion stops at the first non numeric character encountered or the end of the string VAL 123ABC returns a value of 123 YEAR DateField Returns the year portion of an xBase date as an integer 250 SIMSTAT for WINDOWS APPENDIX B REFERENCES STATISTICAL TEXTBOOK AGRESTI A amp FINLAY B 1986 Statistical methods for the social
223. onfidence interval around the mean CHART TAB Mean Error bar graph This option displays a mean bar and or an error bar representing the variability of the mean or of the values in each group Type of error This option allows you to select whether the error bar will represent the standard deviation the standard error or a user defined confidence interval The width of this interval is set by the interval option With barchart This option allows you to draw bars where each bar represents the mean of a separate group You can use the Type of Error option to add error bars 150 SIMSTAT for WINDOWS Upper error bar only When the bar chart option is selected and an error bar is requested this option allows you to specify whether the error bars displayed with the mean bars should be displayed above and below the mean or only above Link means When chosen this option connects the two means with a line Dual histogram This option displays a dual histogram representing the distribution of two numeric variables You must supply the Number of bars bars to plot A Normal curve can also be superimposed on the histogram The horizontal and vertical radio buttons allow you to select between horizontal histograms where the two charts are displayed side by side or vertical histograms displayed one above the other Sample output of an independent samples t test INDEPENDENT SAMPLES T TEST AGGRESS Nb of agressive behavior by SEX Se
224. ons allow you to specify legal values of the binary response By default those values are set to 0 and 1 Constant This option specifies whether or not to include a constant in the model Interaction This option allows to you specify which interactions should be included in the model Variables are designated by a single uppercase letter and are grouped together by an character Multiple interactions may be specified on the same line For instance the following expression A C A B C designate a two way interaction between the first and the third variables on the list and a three way interaction between the first three variables Classification table When enabled this option constructs a classification table observed response vs P observed response 1 with probabilities grouped in tenths determined from the fitted regression model to aid in assessing the adequacy of the fitted model If 3 or more rows have positive totals a Hosmer Lemeshow goodness of fit statistic Hosmer and Lemeshow 1989 sec 5 2 2 is computed along with its P value STATISTICAL ANALYSIS 113 Likelihood ratio This option allows you to display the likelihood ratio statistics for the significance of each variable They can be used as a check on the corresponding Wald statistics which Hauck and Donner 1977 have shown to be misleading sometimes This option was implemented as a separate command because of the work it generates a different model must be f
225. oportion 182 SIMSTAT for WINDOWS Options Cutoff value VALUE real real Cuttoff value values of X MEAN Mean MEDIAN Median PROPORTION real Test proportion from 0 to 1 PANEL Displays the dialog box BOOTSTRAP1 Syntax BOOTSTRAP1 varlist options Description BOOTSTRAPI performs bootstrap simulation to estimate the distribution of descriptive statistics in a population e g mean median variance The program draws a specified number of observations from the sample and computes the estimator for the subsample This procedure is performed many times 10 to 30 000 times The options allow you to display information about the estimator distribution including descriptive statistics frequency table percentile table and histogram of the estimator distribution The program also computes nonparametric and bias corrected bootstrap confidence intervals If no sample size is specified option SIZE the bootstrap sample size is automatically adjusted to the size of the original sample Options SIZE integer Size of each sample SAMPLING integer Number of samples SEED integer Initial seed value Choice of estimator MEAN Mean VARIANCE Variance STDDEV Standard deviation STDERR Standard error MEDIAN Median KURTOSIS Kurtosis SKEWNESS Skewness DESC Descriptive statistics INTERVAL real Confidence intervals PTILES integer Percentile tabl HISTOGRAM Histo
226. options allow one to control the size and location of each window and to execute various window related actions tile cascade minimize etc The parameter for TOP LEFT WIDTH and HEIGHT options is an integer value between O and 100 expressing a percentage of the screen height and width 232 SIMSTAT for WINDOWS Options DATA NOTEBOOK SCRIPT CHART ALL TOP integer EFT integer WIDTH integer HEIGHT integer NORMAL INIMIZI AXIMIZI Lu lu Og CASCADI EJ TILE Examples WINDOW ALL MINIMIZED Window types Data spreadsheet window Notebook Statistical results window Script log window Charts window All four windows Vertical position of the window s upper left corner 0 to 100 Horizontal position of the window s upper left corner 0 to 100 Relative width of the window 0 to 100 Relative height of the window 0 to 100 Returns the active window to its size and position before you chose the Maximize or inimize command Reduces the active window to an icon Enlarges the active window to fill the available space Arranges open windows in an overlapped fashion Arranges multiple opened windows so they fit next to each other on the desktop and do not overlap WINDOW CHART TOP 10 LEFT 10 WIDTH 50 HEIGHT 50 CUSTOMIZING THE TOOLS MENU 233 10 CUSTOMIZING THE TOOLS MENU The TOOLS pull down menu can be used to run some SIMSTAT add in modules
227. or all selected records missing values are excluded To compute a statistic on several variables for each record individually see the previous section Statistical Functions across variables Syntax FUNCTION Variable VMEAN Mean VCOUNT Count VSUM Sum VSTDEV Standard deviation VVAR Variance 54 SIMSTAT for WINDOWS VMIN Minimum VMAX Maximum ZSCORE Normalized score LAG Lag Random Number Functions Syntax FUNCTION value variable or expression NORMAL X Normal pseudo random number with mean of 0 and standard deviation of X UNIFORM X Uniform pseudo random number between 0 and X Date Functions Most date functions can be computed on either date or numeric variables Numeric variables are automatically transformed into a date corresponding to the number of days since 1 1 0001 The TODAY function does not have any argument while the YRMODA function takes 3 arguments separated by commas Each arguments can be either a numeric variable a constant or an expression All other functions require a single argument that can consist of a date a numeric variable a constant or an expression YRMODA Argument yy mm dd Converts 3 values variables or expressions into a numeric value expressing the number of days since 1 1 1900 YEAR Returns the year of a date MONTH Returns the month of a date DAY Returns the day of a date WEEKDAY Returns a numeric value representing the day of the w
228. or off use the Show Tool Hints option in the Preferences dialog box The following list describes the various icons used by SIMSTAT Click To i 5 lt B em EX e B The following eight buttons may be present no matter which window is currently active However their action differs depending on which window is active Create a new file You may use this button to clear the content of the currently active window Open an existing file of the same type as the active window SIMSTAT displays the Open dialog box in which you can locate and open the desired file Save the active window document with its current name If you have not named the document SIMSTAT displays the Save As dialog box Print one or more pages or charts of the active window Cut the selected text Notebook or Script window or selected cell Data window Copy the selected text Notebook or Script window or selected cell Data window Paste text or a value from the clipboard Erase the selected text Notebook or Script window selected cell Data window or chart When viewing the notebook index pressing this button erases the currently selected page 10 SIMSTAT for WINDOWS Notebook window buttons Add a tab to the Notebook Add a blank page to the Notebook raj Select variables to be included in subsequent statistical analysis Chart window buttons cu Copy a bitmap image of the current chart to the clipboard
229. or to transfer execution to an external Windows or DOS application such as a file manager a calculator or even a word processor To add programs to delete programs from or edit programs on the TOOLS menu choose the TOOLS SETUP command from the TOOLS menu The following dialog box will appear Tools Setup x Tools amp Simcalc Calculator To add a program to the TOOLS menu Choose Add SIMSTAT displays the Tool Properties dialog box where you specify information about the program to be added Tool Properties Eg Program information Tite Program C WINDOWS CALC EXE Parameters 0 0 0 0 0 0 0 0 0 Iv Wait for execution to complete 234 SIMSTAT for WINDOWS Title Enter a name for the program you are adding This name will appear on the TOOLS menu You can add an accelerator to the menu command by preceding that letter with an ampersand amp Program Enter the location of the program you are adding You must include the full path to the program Click the button to search your drives and directories to locate the path and file name for the program Parameters Enter parameters to pass to the program at startup For example you might want to pass a file name when the program launches To delete a program from the TOOLS menu Select the program to delete e Click on the Delete button To change a program on the TOOLS menu e Select the program to change Choos
230. ords in the data file you need to use the RECORD command SIMSTAT provides several predefined memory variables holding time and date related information Variable Returns SCURRENT YEAR Current year SCURRENT MONTH Current month number from 1 to 12 SCURRENT DAY Current day of the month from 1 to 31 1j SCURRENT WEEKDAY Current day of the week from 1 to 7 SCURRENT TIMI ES Current time of the day in seconds with one decimal place tenth of a second SCURRENT RECORD Current record number NB RECORDS Number of records currently displayed 176 SIMSTAT for WINDOWS Expression Operators and Functions Some script commands such as LET or IF may include expressions where basic arithmetic trigonometric or random number operations are performed Those expressions may contain the following operators or functions ARITHMETIC OPERATOR Addition Subtraction Multiplication Division Exponentiation NUMERIC AND TRIGONOMETRIC FUNCTIONS Syntax FUNCTION value variable or expression ABS ACOS ASIN ATAN csc COS EXP FACT LN LOG MOD10 RND SEC SORT SOR SIN TAN TRUNC Absolute ArcCosine ArcSine Arctangent Cosecant Cosine Exponential Factorial Natural logarithm Base 10 logarithm Modulus Round Secant Square root Square Sine Tangent Truncate RANDOM NUMBER FUNCTIONS Syntax
231. osition of the box s upper left corner 0 to 100 CLEAR Clears previous value LEN integer Maximum number of characters DEC integer Number of decimal places displayed MIN real Minimum value MAX real Maximum value Examples INPUT NAME What is your name LEN 50 CLEAR INPUT DB AGE How old are you MIN 5 MAX 25 Valid Colors Black Maroon Green Olive Navy Yellow Purple Teal Gray Silver Red Lime Blue Fuchsia Aqua White INTERRATERS Syntax INTERRATERS varlist BY varlist options Description The INTERRATERS command produces an inter rater agreement table that consists of a square table where rows and columns contain the same categories used in both variables Options Sort table by FREQ Ascending frequency DFREQ Descending frequency VALUE Ascending value DVALUE Descending value TABLE Display table ROWPCT Row percentages COLPCT Column percentages TOTPCT Total percentages EXPECTED Expected values RESID Chi square residuals SRESID Standardized residuals PCTAGREE KAPPA SCOTT NFREE KRBAR KR OFREE BARCHART VERLAP TACKED 00 D D WNFWNO 1j PANEL ITEM Syntax ITEM varlist options Description SCRIPT LANGUAGE REFERENCE 201 Percentage of agreement Cohen s kappa Scott s pi Free marginal nominal agreement Krippendorff s r bar Krippendorff s R Free marginal ordinal agreement 2 dimensi
232. pondence analysis requires MVSP v2 2 Correlation matrix Contingency crosstabulation Descriptive analysis Diversity Index computation requires MVSP v2 2 Factor analysis requires EFA v3 0 Full analysis on bootstrap samples Frequency analysis Friedman test GLM analysis of variance and covariance Interraters crosstabulation Classical item analysis requires StatItem v1 0 Kruskal Wallis Anova Kolmogorov Smirnov one sample test Kolmogorov Smirnov two sample test Listing of data Logistic regression requires Logistic v3 11 Mann Whitney U Wilcoxon W test McNemar test Median test Moses test of extreme reactions Multiple responses analysis Multiple regression analysis Nonparametric association matrix Oneway analysis of variance Principal component analysis requires MVSP v2 2 Principal correspondence analysis requires MVSP v2 2 Linear amp nonlinear regression analysis Reliability analysis Single case experimental design Random bivariate bootstrap analysis Runs test Sensitivity analysis ROC curves Sign test T Test independent and paired Time series analysis Wilcoxon signed rank test 180 SIMSTAT for WINDOWS Environment and Multimedia control commands CHART NOTE PLAY PICTURE PRINT SAVE SCREEN TITLE WINDOW Controls the appearance or display of the current chart Displays a text string at the bottom of the screen Plays a WAV sound file a MDI music file or an AVI movie file Disp
233. presented in alphabetical order The sole exceptions to this rule are the BOOTSTRAP resampling procedures that are described under the BOOTSTRAP heading Assigning variables for statistical analysis Use CHOOSE X Y from the STATISTICS menu to select the variables to be used in subsequent analyses You can also press the F3 function key or click on the button to access this function When this command is selected the program displays a dialog box Choices Dialog x Variable list Independent VM OK Numeric M m rr Dependent T m l Sorted Miles per gallon 66 SIMSTAT for WINDOWS The list box located at the left shows the available variables The drop down list at the top of this box allows you to display only specific types of variables such as numeric string date etc or the variables belonging to a user defined set of variables The Independent and Dependent list boxes located on the right of the dialog contain the variables that will be treated as independent and dependent By default the variable names are displayed in the same order as they appear in the data file You may also use the Sorted check box to display the variable names in alphabetical order To assign variables as dependent or independent Select the variables in the Variable list box and move them into the Independent or Dependent box by clicking on the gt button next to the proper list box While most statistical ana
234. provide a diagnostic for those transformations by displaying autocorrelation and partial autocorrelation function plots of the transformed series This command also allows the application of two smoothing methods i e moving average and running median to identify trends in noisy time series data Control bars representing the mean and the confidence limits can also be displayed over th Options EAN DIFF integer e series LOGLog transformation Remove the mean SEASON integer Len PLOT Plo ACF Autocorrelation function PACF Par LAG integer Typ AVG int int RMED int int RBAR Con Number of difference operations gth of seasonality t the series tial autocorrelation function Number of lags for ACF and PACF e of smoothing oving average Running median trol bars PCT integer LOW integer HIGH integer PANEL TITLE Syntax SCRIPT LANGUAGE REFERENCE 229 Width of interval Lower value Higher value Display the dialog box TITLE expression options Description The TITLE command displays a single line string on the background screen By default the title line is displayed horizontally centered and at the top of the screen with a font size of 24 points The TOP and LEFT options can also be used to specify the position of the text s upper left corner The parameter for these two options is an integer value between 0 to 100 expressing a percentage of the screen hei
235. r Values of independent variable 1TAIL 2TAIL Direction of the test PANEL Display the dialog box 206 SIMSTAT for WINDOWS MEDIAN Syntax MEDIAN varlist BY varlist options Description The MEDIAN test is a procedure for testing whether two or more independent groups differ in central tendencies It tests the likelihood that those groups were drawn from populations with the same median The output displays the number of cases greater than less than and equal to the median for each category of the grouping variable Also displayed are the median chi square degree of freedom and probability value Options EXTENDED Extended median test VALUES int int Values or range of X PANEL Display the dialog box MENU ENDMENU Syntax ENU options ENDMENU Description The MENU command allows you to define a bouncing bar menu Every line between the MENU and ENDMENU commands will become a menu item The maximum number of items is 30 An optional amp character can be inserted in the command file to specify an accelerator key that when pressed will select this item A single line command can be associated with a menu item by putting a command between this menu item and the associated command A menu separator can also be added by putting a character on a separate line By default the menu is displayed in the middle of the screen The TOP and LEFT options can also be use
236. r WINDOWS Read only When checked this option prevents a variable from being modified This option is useful to prevent accidental or unauthorized changes to the values of a variable To prevent the modification of an entire data file use the DATA SECURITY command from the FILE menu or open the file as Read Only Rename button You can use this button to change the name of the current variable When you click this button you will be asked for a new variable name This name must not exist in the current data file and should follow the basic rules for valid variable names Alignment The alignment option lets you specify whether the values of this variable should be displayed on the grid aligned to the left at the center or flushed to the right of the column By default numeric values and dates are flushed to the right of the column while strings are aligned to the left Display width The display width option lets you adjust the display width of the current variable This option is used exclusively to control how the variable is displayed in the data grid and does not affect the physical size of the variable or its internal precision While it may be possible to set this option to zero or one character the actual minimum display width is equal to the number of characters in the variable name For example the minimum display width of a variable named AGE will be 3 Decimals When the variable is numeric the decimals option is used
237. r name already exists you will be asked to confirm the overwriting of this file DATA FILE OPERATIONS 27 Defining variables In addition to the specification of the physical structure of the variable it is also possible to specify a variable s display format width and number of decimal places missing values and value labels To define these attributes use the following steps Position the cursor on the variable you want to define Choose the DEFINE VARIABLE command from the FILE menu or click on the a button on the upper left corner of the data grid This displays the Variable Definition dialog box as shown below ariable definition AGE Ed Display Type NUMERIC 8 0 Read only n Rename Alignment Left C Center Right Display width o Decimals e E Labels Variable Age of the child Value labes Edit Missing values C Equal to Less than or equal to sd C Equal to More than or equal to sd x Cancel Help Navigating through variables The and icons located at the bottom of the dialog box can be used to move to the previous or the next variable in the data file To locate a specific variable click on the Find button A dialog box appears that allows you to select a variable name from a list of all variables in the data file To quickly locate a variable name you can also type its first letters until the variable name appears in the list box 28 SIMSTAT fo
238. r the differential rating tendencies of the judges in the computation of the chance factor OPTIONS TABLE PAGE Agreement table This option requests the output of an agreement table Sort by Use this option to tell the program whether the row and columns of the table should be sorted by the values of the variable or by frequency Type The type option allows you to specify whether the rows and columns of the agreement table should be sorted in ascending or descending order 98 SIMSTAT for WINDOWS Table content group The Table Content group box allows you to request other statistics in addition to frequencies to be included in the cells of the table You can obtain as many of the statistics as desired by enabling the corresponding check box Row percentages Column percentages Total percentages Expected frequencies Chi square residuals e Standardized chi square residuals NOTE The expected frequencies displayed in the inter rater agreement tables do not necessarily correspond to the expected frequencies used in the above correction techniques Rather they correspond to the values used in the computation of chi square statistics used in contingency tables However those values coincide with those used in the computation of Cohen s Kappa and Krippendorf r STATISTICS PAGE The STATISTICS page allows you to request various nominal and ordinal level measures of agreement Percentage of agreement Cohen s
239. re as the current data file but with only a subset of cases To create such a file Set the filtering and sorting conditions of the active data file to display the records as they should be saved in the new data file Choose the DATA EXPORT command from the FILE menu Set the Save As File Type drop down list to DBASE Type a valid filename including the DBF file extension If necessary use the Directories and Drives boxes to specify the storage location of your choice e Click on the OK button 42 SIMSTAT for WINDOWS Merging data files There are circumstances when two or more data files need to be merged together For example you may need to append data from additional subjects or from another time period to the end of an existing data file or you may need to add new information on existing individuals by adding new variables Simstat provides two merging methods to either add new records or new variables to an existing data file To add new records to an existing data file Open the data file to which you would like to append records Choose the DATA APPEND RECORDS command from the FILE menu Select the data file to be merged with the current file and click OK All variables with matching names and types are appended to the current data file If variable length differs data in the merged database is either truncated or padded with spaces Deleted records in the merged file are not appended to the current
240. restrict access to Choose the DATA SECURITY command from the FILE menu The following Security dialog box will appear becutity options x Default access to database v Modify data IV Create new variables Iv Delete existing variables v Create new records Iv Delete existing records v Print or export data base v Modify variable information Full access password fF o x Cancel When first accessed all options are checked indicating that no restriction to the data file has been defined Disable the privileges you want to associate with the password For example if you disable the Delete Existing Variables and Delete Existing Records options it will be possible for any user to edit data create new variables or modify the variable definitions missing values labels display widths etc However they will not be able to delete any existing record or variable in the data file 44 SIMSTAT for WINDOWS Type the password that will give full access to the data file and click on the OK button to exit the dialog box NOTE You can limit default access for users without the need to provide a password by leaving the Full Access Password edit box empty By doing this the access will be limited every time the data file is opened but it will still be possible for anyone to modify this access privilege by altering the options in this Security dialog box To gain full access to
241. s E varlist BY varlist options The CHISQUARE command performs a one sample chi square test that allows to assess whether there is a difference between the observed number of cases in various categories and the expected frequencies in those same categories The options allow you to restrict the test to specific values and to specify the expected frequencies Options VALUES real real FREQ real real PANEL CLUSTER Syntax SCRIPT LANGUAGE REFERENCE 187 Expected values Expected frequencies Display the dialog box CLUSTER varlist options Description This procedure performs hierarchical agglomerative cluster analysis of a distance or similarity matrix Seven forms of clustering are presently available the four average linkage procedures unweighted pair group unweighted centroid weighted pair group and weighted centroid or median nearest and farthest linkage and minimum variance Requires MVSP v2 2 Options LOG10 OGE LOG2 SQRT RATIO TRANSPOSE EUCLI SEU D C STEU COSI MANH CANB CHOR CHIS R N R A C A UCENTER WPAIR WCENTER MINVAR Select a transformation Log base 10 Log base e Log base 2 Square root Log ratio Transpose data Select a coefficient default EUCLID Euclidian distance Squared Euclidian distance Standardized Euclid
242. s are closed before exiting the program RANDOM Syntax RANDOM varlist BY varlist options Description The RANDOM command is similar to the BOOTSTRAP2 command but simulates the null assumption that there is no difference or relation in the population The program draws from the sample and for each variable a specified number of random observations with replacement and computes the estimator for the subsample The procedure is performed many times 10 to 30 000 times The options allow you to display information about the estimator distribution including descriptive statistics frequency table percentile table and histogram ofthe estimator distribution The program also computes nonparametric and bias corrected bootstrap confidence intervals and Type I error rate for up to 4 alphalevels If no sample size is specified option SIZE the bootstrap sample size is automatically adjusted to the size of the original sample 218 SIMSTAT for WINDOWS Options SIZE integer SAMPLING integer SEED integer TAU A TAU B TAU C D SY D XD D YD U Ug Q D Ss Pee n tt o w INTERCEPT D 3j DESC CI real PTILES HISTOGRAM VERT HORIZ IN real INC real NBAR integer NORMAL ERROR real PANEL Size of each sample Number of samples Initial seed value Choice of statistics Kendall s Tau A Kendall s Tau B Kendall Stuart s Ta
243. s of a variable The RECODE command provides an easy way to make multiple changes to the values of numeric variables or to collapse values of a continuous variable into categories To activate this command Position the data sheet cursor on the variable you want to recode e Select the TRANSFORM VARIABLE RECODE command from the FILE menu The following dialog box will appear Recoding values of BEHAVIOR x Store in BEHAVIOR Expression 0 1 2 3 4 M K x Cancel Help Store in This edit box allows you to specify the target variable where the computation results will be stored By default the value is set to the name of the variable containing the values that will be used for recoding To store the result in another variable change this text box to the new target variable name If you keep the original variable name or specify an existing variable name the program will ask you if you want to replace its values with those produced by the transformation If the variable does not exist the program will ask you to confirm the creation of this new variable Expression The recode expression can consist of several transformations enclosed in parentheses each including one or more values or a value range an equal sign and the new value For example 122 221 345 23 6 10 4 ELSE SYSMIS Each value on the left of the equal sign is recoded into the value on the right T
244. s of one or several numeric variables When more than one variable is specified the first variable in the list will contain the value labels and all the other variables will be linked to this first variable Example LABEL DEPRESSION1 TO DEPRESSIONS Never Sometimes All the time NROS SCRIPT LANGUAGE REFERENCE 231 WEIGHT Syntax WEIGHT variable Description The WEIGHT command allows you to designate a weighting variable When the command is used alone the weighting is turned off Example WEIGHT GROUPNO use values in GROUPNO as weights WEIGHT turns weighting off WILCOXON Syntax WILCOXON varlist BY varlist options Description The WILCOXON matched pairs signed ranks test is a procedure used to test whether two related samples have been drawn from the same population Like the sign test it computes the difference between the values of the two variables but takes into account the magnitude as well as the direction of the differences The Wilcoxon signed ranks test is the nonparametric version of the t test for paired samples The probability test performed can be either one or two tailed Options 1TAIL 2TAIL Direction of the test PANEL Display the dialog box WINDOW Syntax WINDOW WindowType options Description The WINDOW command provides complete control of the display of any SIMSTAT window The various
245. s to locate the data file If the file is on a different disk click on the down arrow of the Drives list box to display available drives and select the disk where the file is located Ifthe file is in a different directory double click on the directory names in the Directory list box also called Folder list box to move through the directory tree Step 2 Assigning variables The next step is to choose the variable s on which you will perform the analysis Select the CHOOSE X Y command from the STATISTICS menu to display the Variables Selection dialog box 14 SIMSTAT for WINDOWS hoices Dialog x Variable list Independent 4 OK El pee a Dependent T Integer weight Sorted This dialog box contains among other The one located to the left of the dialog box contains a list of all variables in the data file The selection of variables for upcoming analysis is carried out by moving the proper variable s name s from this list to the independent or dependent list box When doing descriptive analyses on separate variables it does not matter whether a variable is assigned as dependent or independent For the current example we will thus move the variables to the independent variable list box To achieve this Highlight the AGE variable age of the child in the variable box by clicking once on it Click on the button just beside the independent list box to move the highlighted variable
246. s which will allow you to store all the necessary files in a single archive file and will automatically restore those files in the location of your choice DATA FILE OPERATIONS 23 Navigating in the Data Window To move the caret to a specific cell you can click on that cell with the mouse You can also use the following keys to navigate in the spreadsheet Up Down or Enter Right or Tab Move one cell up Move one cell down When you reach the last line and hit the down arrow key a new case is created Move one cell to the right When you reach the end of a row these keys bring you to the first column of the next row If you reach the last cell and hit one of these keys a new row is created Left or Shift Tab Move one cell to the left If the cursor is on the first column PgUp lt PgDn gt lt Home gt lt End gt lt Ctrl Home gt lt Ctrl End gt lt Ctrl G gt pressing either of these keys brings you to the last column of the previous row Move one screen up Move one screen down Move to the first variable column Move to the last variable column Move to the first record Move to the last record Search for a variable name You can also move to a specific value or string within a selected column or anywhere in the spreadsheet by evoking the FIND command in the EDIT menu 24 SIMSTAT for WINDOWS Creating a new data file To create a new data file select the DATA
247. sabled this option allows you to specify how many cases to include in the listing Sample output of list cases LISTING OF DATA N 10 SEX AGE SIBLING HOURSTV AGGRESS 1 1 00 6 00 00 3 2 51910 3L OW 2 1 00 TIXDO 00 dU 0 46 00 3 1 00 9 00 00 23189100 66 00 4 1500 10 00 1 00 156550 66 00 5 ROO 10 00 TB Le ene 26 00 6 1 00 10 00 00 iL 50 26 00 7 EO TOLSDU 2o 00 2250 53 00 8 100 TETBDS LSU Jen 46 00 2 1 00 9 00 1 00 14 00 56 00 10 1200 3500 00 16 00 26 00 112 SIMSTAT for WINDOWS Logistic regression The LOGISTIC REGRESSION command fits a multiple logistic regression model on a binary response variable with one or several explanatory variables Output includes the likelihood ratio statistic for overall significance parameter estimates exponentiated parameter estimates which are the odds ratios corresponding to a unit change in the independent variables Wald statistics for assessing the effects of independent variables and confidence intervals for the regression parameters For more detailed information about logistic regression and the statistics computed see the LOGISTIC user s manual NOTE LOGISTIC v3 11Ef is a shareware program written by Dr Gerard E Dallal You can obtain a copy of the program on any Internet SIMTEL FTP site such as oak oakland edu or by writing to Gerard E Dallal 54 High Plain Road Andover MA 01810 USA OPTIONS Value for success and Value for failure These two opti
248. samples or at different time periods etc To create a new tab e Make sure the Notebook window is the active window Select the TABS ADD command from the EDIT menu or click on the icon Enter the new tab name and its page location Press Enter or click on the OK button to create the new tab Once your tabs are created you can quickly move to a specific section by clicking on its tab at the bottom of the notebook When activated a section become the default output location This means that the output of all analyses performed while this section is active will be appended to the end of this section To delete an existing tab but not its contents Click on the tab you want to delete or select any page in its section e Select the TABS DELETE command from the EDIT menu To delete an entire section Click on the Index section to display the notebook index e Select the tab of the section you want to delete e Select the ERASE command from the EDIT menu or click on the icon To modify a tab name or change its location Click on the tab you want to modify or select any page in its section Select the TABS EDIT command from the EDIT menu Modify the tab name or its page number and click OK to confirm the changes 64 SIMSTAT for WINDOWS Suggestion Itis advisable to keep a copy ofthe data used with each analysis output To achieve this you can create a special DATA section and use the LIST command to
249. sed for filtering records This dialog box allows you to create a custom sorting expression by selecting information and elements displayed on the form This expression can include almost any supported xBase function The sort builder dialog box contains the following elements Variable name list box Double clicking on a variable name from the list box located on the left of the dialog box inserts that name in the edit box at the current caret position Function list box A list of valid xBase expressions is displayed to the right of the dialog box Double clicking on an xBase function from the box inserts that function at the current caret position When a function requires one or more arguments the argument section remains highlighted To replace the highlighted text with a value an expression or a variable name simply type the proper text on the keyboard or select a variable name or function Relational operations and numeric buttons Selecting any relational operation or numeric button inserts the corresponding symbol in the edit box Usually when part of the sorting expression is highlighted pressing any keyboard key or clicking on any variable name or numeric button will replace the highlighted text with the character or expression associated with this key or button However when choosing a function requiring a parameter enclosed between parentheses the highlighted text will not be deleted but will be used instead as the new para
250. see appendix A Then Else The THEN and ELSE edit boxes contain the transformation expressions you want the program to apply See the previous section Building a valid transformation expression for information on syntax rules and available functions If the logical expression entered in the IF edit box is evaluated as true the transformation expression in the THEN edit box is computed if not the expression in the ELSE field is used If an edit box contains no transformation expression the existing value remains unchanged For example the following setting STORE IN HEIGH IF SEX Female THEN HEIGHT 1 3 ELSE Will replace for each record where the character variable SEX is equal to Female the current value stored in variable HEIGHT with this same value multiplied by 1 3 all other records remain unchanged 56 SIMSTAT for WINDOWS Quick transformations The QUICK TRANSFORM command allows immediate transformation of the current variable using various functions Z Score ZSCORE X or X VMEAN X VSTDEV X Remove mean X VMEAN X Reverse VMAX X X 1 Add constant X constant Square root SORT X Natural logarithm LN X Inverse 1 X For the square root natural logarithm and inverse transformations if the variable contains values less than 1 SIMSTAT asks whether to add a constant in order to bring this minimum to 1 DATA TRANSFORMATION 57 Recoding value
251. senting each unique alphanumeric value found in the original string variable Value labels corresponding to each numeric value will be automatically created and stored with the variable If the name of an existing variable is provided you will need to confirm the overwriting of values in this variable When the variable name supplied is the same as the original string variable the program assumes that you want to transform the current string variable into a numeric variable and will erase the original variable and add a new numeric variable of the same name This new variable will be located at the left end of the data worksheet 60 SIMSTAT for WINDOWS 5 WORKING WITH THE NOTEBOOK The Notebook window displays the statistical output for all analyses performed during a session The notebook metaphor provides an efficient way to browse and manage outputs The text output of each analysis is displayed on a separate page You can turn pages with the mouse by clicking on the page corner icons at the bottom of the notebook or by using the lt PgUp gt and lt PgDn gt keyboard keys Tabs may also be added to create sections in the notebook allowing you to store different types of analyses in different sections of the notebook An index ofall analyses included in the notebook is also provided This index can be used to quickly locate and go to a specific page move pages within the notebook or delete some pages While each page can be annotated or ed
252. sion of bivariate regression The result of regression is an equation that represents the prediction of the dependent variable from several independent variables The regression equation takes the following form Y A B X B X BEX where Y is the predicted value of the dependent variable A is the intercept the value of Y when all values of the independent variables are zero X represents the observed value of the independent variables or predictors and B is the coefficient assigned to each of the independent variables The goal of the regression technique is to find the values of all B that produce prediction scores that most closely fit the actual values of Y SIMSTAT provides three broad classes of multiple regression standard regression hierarchical regression and statistical stepwise regression Each class differs in the way the independent variables are selected to be included in the equation Standard Multiple Regression In standard regression all independent variables are entered into the regression equation at the same time In this type of regression some independent variables may be entered in the model even if their simple correlation with the dependent variable is negligible It is also possible that some variables with initial high correlation with the dependent variable are entered but since they are also highly correlated with other predictors their unique contribution to the prediction of the dependent varia
253. sizes that range from 1 to 100 000 observations Random sampling Another aspect of bootstrapping is that it rests on the assumption that the original sample is representative of the population SIMSTAT offers a modified bootstrap sampling process that rests on the null assumption that there is no difference or relation between variables in the population While in bootstrap sampling the drawing is achieved by drawing vectors of scores for a particular subject the RANDOM SAMPLING procedure draws individual subject scores for each variable independently Consequently while a standard bootstrap simulation on a correlation between two variables would yield coefficients that fluctuate around the correlation that exists in the original sample the RANDOM SAMPLING procedure would produce correlations that vary around a null correlation In this procedure the proportion of redrawn samples that lead to a statistically significant estimator at a given alpha level is used to assess the Type I error rate Possible applications We have already seen that standard bootstrap resampling can be use to obtain various measures of sampling variability such as nonparametric confidence intervals The ability to alter the bootstrap sample size and to replicate the condition of the null hypothesis makes possible numerous new applications The following paragraphs give some examples of such applications Research planning Power estimation The possibility of comparin
254. slope INTERCEPT Regression intercept S T Student s T S F Student s F W Mann Whitney WILCOXON Wilcoxon W value SIGN Sign test Z value K W Kruskal Wallis EDIAN Median test Z value AGREE Percentage of agreement KAPPA Cohen s kappa SCOTT Scott s pi NFREE Free marginal nominal scale KRBAR Krippendorff s r bar KR Krippendorff s R OFREE Free marginal ordinal scale 184 SIMSTAT for WINDOWS DESC INTERVAL real PTILES Percentile tabl HISTOGRAM NBAR integer NORMAL POWER real PANEL BOX ENDBOX Syntax BOX Options ENDBOX Description Descriptive statistics Confidence interval Histogram Number of bars intervals Normal curve Statistical power analysis Displays the dialog box The BOX command displays a window with textual information By default the window is positioned in the middle of the screen The TOP and LEFT options can also be used to specify the position of the window s upper left corner The parameter for these two options is an integer value between 0 a 100 expressing a percentage of the screen height and width The window stays on screen until the user presses Enter or clicks on the OK button If the NOBUTTON option is used the window stays on screen until a key is pressed The DELAY option allows you to insert the minimum length of time the box should be displayed on screen before the user can proceed The colors of the text in the box can also be altered using th
255. standardized residuals along the vertical axis Residual normal plot This option displays a normal probability plot of residual values that allows you to evaluate whether those residuals are normally distributed If the residuals follow a normal distribution the data points will fall approximately along a straight line going from the lower left corner of the graph to the upper right corner STATISTICAL ANALYSIS 121 Save residuals This option instructs SIMSTAT to save the predicted and residual values as new variables in the current data file Creating those variables allows you to perform further analyses on them such as displaying scatterplots between each independent variable and the predicted and residual values The new variables are named PREDxxx and RESIDxxx where xxx stands for a serial number between 001 and 999 that corresponds to the number of regression analyses performed during a single command This number is automatically reset to 001 after each command To prevent overwriting those variables you must rename them using the DEFINE VARIABLE command CRITERIA PAGE P to enter P to remove The P to Enter and P to Remove fields allow you to set criteria to be used in statistical regression techniques i e stepwise forward and backward methods A variable will be entered in the model if its probability is less than the P to Enter criteria while a variable in the equation that has a probability above the P to Remove criteria
256. stered trademark of Corel Corporation Lotus 1 2 3 and Symphony are registered trademarks of Lotus Development Corporation Other product names mentioned in this manual may be trademarks or registered trademarks of their respective companies and are hereby acknowledged Acknowledgments Special thanks to Marc Aras Jean B langer Jacques P Beaugrand DeWitt Kay Warren L Kovach Mark Thomas Lindemann Ian D Livingstone Rashid Nassar Ben Riga Roel van Schaik George Schwartz Mark Von Tress Todd Woodward and several others for their invaluable support comment feedback and advice during the development of this program TABLE OF CONTENTS Introd ction rS Ps teat ELEME LEE LIED UL 1 Using this manual soora cenie ra eek ere bete b ibe bated Gon ei 1 Manual conventions 0 0 eee cette m n ene 2 1 Getting started sg LN A GE Au de Se BY 3 The SIMSTAT package I RR eon the ow Fa EASE eee ee 3 System requirements eps d sp detta athena eats Tte Lee Ra elas 3 Installing SIMSTAT eS EAE at e a EE 3 Making a backup copy seseeeeeeee m Ie 3 Starting the program eee eae 4 Startup Optom sss ss ch rts vuoi crie ett d ec D n sete qp Ete dep ts Ge ed 4 The working environment 0 0 e eee eee eens 4 Changing the active window 0 cece cee eee eens 6 Working with pull down menus 0 c eee eee nee 6 Working with dialog boxes 0 cee eee tee eee ence 6 Loo bal 2s ioe Gea oS E E
257. stimators by allowing the evaluation of the Type I and Type II statistical power error rate of a test While Monte Carlo simulations usually analyze data generated by assumed mathematical functions bootstrap simulation provides a direct assessment of sample distributions from data supplied by the researcher By performing simulations on data distributions more representative of real world data bootstrap may be a more appropriate evaluation of statistical robustness The dialog boxes ofthe ONE VARIABLE TWO VARIABLES and RANDOM SAMPLING are very similar and have comparable options The following section presents those options and provides an indication when an option is specific to an analysis OPTIONS RESAMPLING PAGE Estimator This option displays a drop down list from which you can choose a specific measure that will be computed on each bootstrap sample When the ONE VARIABLE command is chosen you can select from a list of seven estimators see above while the TWO VARIABLES and the RANDOM SAMPLING command offer a choice between 28 estimators Values Some estimators performed on independent samples require the specification of two values of the grouping independent variable Those two values will be used to define two groups or will be treated as minimum and maximum values of the grouping under consideration Option only available for ONE VARIABLE and RANDOM SAMPLING commands Number of samples This option allows you to choose
258. t From the Current Set drop down list box select the set you want to delete e Click on the DELETE button and enter the new set name 48 SIMSTAT for WINDOWS 4 DATA TRANSFORMATION Often you may need to create new variables to represent global scores based on several responses given by a subject or to create composite indices based on several indicators Preliminary analysis of existing variables may also reveal either coding errors or an inadequate data distribution for the kind of analysis you want to perform SIMSTAT provides several powerful commands to transform values of existing variables or create new variables by performing mathematical operations on existing ones This chapter presents data transformation features currently available in SIMSTAT TheTransformation submenu contains various commands to perform transformation of existing variables or to create new variables from computation on existing ones Compute Provides transformations using various functions on one or more variables Quick transform Performs immediate transformation of the current variable using various functions Recode Changes the values of numeric variables or collapses values of a continuous variable into categories Rank Replaces the values of the current variable with their rank Dummy coding Automatically creates as many dummy variables as needed to represent the values of a nominal variable Numeric coding Automatically creates a numeric
259. tability of the test and the consistency of response to different item samples Another method to measure the reliability of a test known as SPLIT HALF reliability requires only a single administration of a test In this procedure the items in the test are divided into two halves comparable in terms of content difficulty etc The most common splitting procedure consists of comparing the scores on the odd and even items of the test The correlation between the two scores obtained on the same subjects is then computed The split half reliability coefficient provides a measure of the internal consistency of the scale The INTERNAL CONSISTENCY method also requires only a single administration of a test It is based on the correlations obtained between all items of the scale It provides an evaluation of the homogeneity of the scale also called internal consistency The internal consistency coefficient has been shown to be mathematically equivalent to the mean of all split half coefficients obtained from different splits of a scale The RELIABILITY command also provides a method of assessing the quality of multiple item additive scales through the computation of split half reliability and internal consistency statistics Test retest reliability can be assessed using the REGRESSION or CORRELATION procedures Each selected variable is considered as a single item of the scale The Xs and Ys are used in the split half method to specify how the various it
260. tate Mes Score MOS Analysis Log Analysis began 5 20 1996 at 20258731 The raw data file has 12 variables and 132 observations Correlation matrix created Factor extraction complete NOTE Trace 12 000 with 63 46 of the total trace extracted by 3 factors Varimax rotation complete Unrot ted tator s res calculated Rotated factor scores calculated Analysis ended 5 20 1996 at 20 58 32 STATISTICAL ANALYSIS 83 84 SIMSTAT for WINDOWS STATISTICAL ANALYSIS 85 86 SIMSTAT for WINDOWS Frequencies FREQUENCY performs frequency and descriptive analysis on all selected variables dependent or independent The dialog box allows you to display a table with frequency counts for each value of a variable the percentage of the count over all cases and over valid cases only and the cumulative percentage over all valid cases It allows you to obtain a bar chart a Pareto or a pie chart for numeric and string variables When used with numeric variables you can also obtain percentile tables and descriptive statistics mean median mode standard deviation variance skewness kurtosis minimum maximum and range for each variable as well as histograms box amp whisker plots cumulative distribution charts and normal probability plots OPTIONS ANALYSIS TAB Frequency table This option requests the output of a frequency table This table includes the frequency counts for each value of a variable
261. tative variables covariates SIMSTAT s implementation of analysis of variance and covariance is based on the General Linear Model Using a hierarchical regression analysis technique allows much greater flexibility than standard ANOVA ANCOVA procedures by allowing one to freely combine quantitative and categorical factors and to statistically control for covariates which are either categorical or quantitative The procedure can also be used to perform standard multiple regression problems that involve interaction terms The GLM ANOVA ANCOVA procedure also handles balanced and unbalanced ANOVA designs by providing automatic adjustment for unequal cell size The dialog box allows you to display standard ANOVA ANCOVA tables as well as various outputs usually found in ANOVA ANCOVA or multiple regression analysis Various adjustment methods for unequal cell sizes are provided including a hierarchical strategy where the user can set the order of entry of each variable in the model OPTIONS ANALYSIS PAGE Show steps This option requests the printing of various statistics at each step of the analysis The information output at each step includes an ANOVA table and a choice of statistics such as multiple regression coefficients regression equation for the variables entered in the model as well as cell adjusted means for nominal variables Multiple regression This option displays the multiple regression coefficients R square adjusted R square and t
262. tatus line From now on almost every action you perform will be recorded by SIMSTAT To deactivate this feature follow the same steps a second time You can also press the lt Ctrl F10 gt key combination or click on the Record keyword on the status line to toggle the RECORD script feature on and off The extensive correspondence between the commands keywords and the options available through the dialog boxes greatly facilitates the learning ofthe script language syntax However the easiest way for a beginner to write but also to become familiar with this syntax and the various keywords is to use the RECORD SCRIPT feature You can experiment by performing some analysis and looking closely at the commands generated This feature is also an efficient method to write script files rapidly and easily Running a script To run an entire script e Click on the b button or select the RUN command from the SCRIPT menu To run a specific block of commands Select the commands that you want to execute using either the mouse or the keyboard Make sure that the beginning of the selected text is located on the first line of a valid command e Click on the button or select the RUN SELECTION command from the SCRIPT menu CS Toruna single command simply position the caret anywhere on the first line of the command you want to execute and activate the RUN SELECTION command To start the execution of a script at a specific line Position the cursor
263. ted as incorrect responses Response total correlations This option displays for each item in the scale the frequency and percentage of endorsement of each response as well as the biserial and point biserial correlation between the response and the scale total omitting that item The item is removed from the total score since keeping it would produce artificially inflated correlation coefficients The table also includes the value the Cronbach s alpha would take if the item was deleted from the scale as well as a discrimination index D index obtained by computing the difference between the percentage of correct responses in the highest and lowest groups The percentage of respondents used for this index can be set using the Hi Low discrimination index option that can take a value between 10 and 5096 To include similar information about each alternate response enable the Include alternate responses option Item total endorsement rate This option allows one to obtain a matrix that displays for each value of the scale total the percentage of cases who scored positively on each item Item characteristic curves This option displays curves allowing you to examine the relationship between the total score and the success level on an item to compare the difficulty level of several items and to estimate their discriminatory value By default the curves are drawn using the original scores The Smoothed curve option may be used to eliminate some irre
264. ted by the letter X for extreme Normal plot This option displays a normal probability plot that allows you to evaluate whether the data are normally distributed If the data follow a normal distribution the data points will fall approximately along a straight line going from the lower left corner of the graph to the upper right corner Cumulative distribution This option displays a cumulative distribution of frequencies Histogram The Histogram option graphically displays the distribution of a numeric variable When this option is activated the program first separates the values into non overlapping intervals of equal width then plots bars that represent the relative frequencies of each interval The Number option allows you to specify how many bars will be plotted You can also assess how close the distribution is to a normal distribution by superimposing a normal curve on the histogram To obtain such a curve enable the Normal curve option OPTIONS PAGE Suppress table This option allows you to suppress the printing of frequency tables and nominal charts such as barcharts piecharts and Pareto charts when a variable has more values the the supplied cut off value 88 SIMSTAT for WINDOWS Sample output of a frequency analysis STATISTICAL ANALYSIS 89 Friedman Test The FRIEDMAN TEST is a procedure for testing whether two or more related samples have been drawn from the same population The output displays the mean
265. teger integer Values for success failure NOCONSTANT Exclude the constant ITER Show iterations CTABLE Classification table LRATIO Likelihood ratio statistics CI real Confidence interval width TOLERANCE real Tolerance level INTERACTION var var Interaction hak gs oak feaa PANEL Display dialog box SCRIPT LANGUAGE REFERENCE 205 MANN Syntax MANN varlist BY varlist options Description The MANN command produces a Mann Whitney U test that evaluates the hypothesis that two independent samples have the same distribution The Mann Whitney U is the nonparametric version of the t test for independent samples This test is performed on the dependent variable divided into two groups as defined by values of the independent grouping variable The probability test performed can be either one or two tailed Options VALUES integer integer Values of independent variable 1TAIL 2TAIL Direction of test PANEL Display the dialog box McNEMAR Syntax McNEMAR varlist BY varlist options Description The McNEMAR test is a procedure applied to a pair of correlated dichotomous variables to test whether there is a significant difference in proportions of subjects that change from one category to another A binomial test is used to compute the significance level when the number of changes is less than 10 Otherwise a chi square statistic with the Yates correction for continuity is used Options VALUES integer intege
266. tg Select a transformation Log base 10 Log base e Log base 2 Square root Logratio Transpose data Computation algorithm Reciprocal averaging Cyclic Jacobi Output of graphics Text scattergrams Graphic scattergrams only to Reciprocal averaging Detrending procedure Number of segments default Number of cycles default 4 Downweight rare species 26 only to Cyclic Jacobi inimum eigenvalu Kaiser s rule Jolliffe s rule Specify a minimum eigenvalue default none Accuracy of solution Select a weighting strategy Adjust to percentage Weight on rare species Weight on common species 190 SIMSTAT for WINDOWS CORRELATION Syntax CORRELATION varlist BY varlist options Description The CORRELATION command produces a matrix of Pearson correlation coefficients for all pairs of dependent independent variables If only one list of variables is specified the procedure produces a square matrix where each variable is treated as dependent and independent You can request either exact probabilities for the coefficients or a display of asterisks that indicate the probability levels attained You can also tell the program to calculate probabilities using one or two tailed tests to compute confidence intervals of specified width and to display cross product deviation and covariance tables A graphic scatterplot can also be obtained with or without regression lines Options EXACT Exact
267. the examinees is useless since it will not allow one to differentiate between individuals The item difficulty index obtained by computing the percentage of subjects who responded correctly to the item represents a crude indication of the item s capability to discriminate between examinees From the perspective of the latent trait model if we assume a perfect relationship between the ability of the examinees and the success on an item then an item with a difficulty index of 5096 would differentiate between those subjects who are positioned on the higher half of the ability scale from those in the lower half In this ideal situation a scale with all of its items at the same difficulty level of 50 would not be adequate since it would only differentiate between subjects in two groups and would not allow one to differentiate further among respondents in either of these groups It would then be important to choose items with various levels of difficulty in order to be able to differentiate between all levels However such a perfect relationship between ability level and item response is seldom achieved partly because items are seldom homogeneous According to Henryssen 1971 when the level of correspondence between the item response and ability level as measured by the item total correlation is moderate between 30 and 40 then a scale composed of items with difficulty levels between 40 and 60 should provide an adequate test that will permit reli
268. the intervention smoothed data split middle trend control bars 224 SIMSTAT for WINDOWS Options BRIEF DETAIL CUMUL LOG EGRESSION MEDIAN MAVG int int RMED int int RBAR PCT integer IN integer AX integer PANEL SCREEN Syntax SCREEN color Description Descriptive statistics Cumulative frequency Log transformation Display mean lines Display least square regression lines Display split middle trend lines Smoothing technique Moving average Running median Control bars Width of interval Lower value Higher value Display the dialog box This command allows you to display a background screen on which textual information pictures movies menus and dialog boxes will be displayed The color option lets you specify the background color To remove the screen background use the HIDE option Option color HIDE Valid Colors Black Maroon Green Gray Silver Red Examples SCREEN NAVY SCREEN HIDE See below for valid background colors Remove the screen Olive Navy Yellow Purple Teal Lime Blue Fuchsia Aqua White SCRIPT LANGUAGE REFERENCE 225 SENSITIVITY Syntax SENSITIVITY varlist BY varlist options Description The SENSITIVITY analysis allows one to assess the ability of a quantitative measure X to differentiate a dichotomous criterion condition Y and provide guidelines to c
269. the same group of subjects at different occasions under different conditions or using different sets of equivalent items that are supposed to measure the same underlying variable By using this type of procedure the difference or fluctuation observed between the various administrations of the test provided that it cannot be attributed to real changes in the subject can be used to estimate the proportion of the total variance of the test score that can be attributed to error of measurement Various methods are used to estimate the reliability of a test The TEST RETEST reliability method is obtained by administering the same test to the same subjects on two different occasions usually no more than 2 months apart The correlation between the scores obtained by the same subjects on the two administrations of the test is then computed to obtain a measure of temporal stability or test retest reliability Provided that the length of the interval between the two administrations is short enough to preclude any real change in the variable being measured any fluctuation between the two scores is attributed to random error of measurement The higher the reliability the less susceptible the scores are to the random daily changes in the condition of the subject or the testing environment A test retest reliability method can also be applied using alternative forms of the test The reliability coefficient obtained with such a method measures both the temporal s
270. tile percentiles STATISTICAL ANALYSIS 87 CHARTS PAGE Bar chart If this option is activated the program produces a bar chart that provides a graphical representation showing the relative frequencies of every value of a variable Pie chart This option displays a pie chart where the relative frequency of each value is represented by a slice Pie charts are appropriate when you want to compare individual values to other values and to the whole Pareto chart This option allows you to obtain a bar graph that displays the categories of a variable sorted in descending order of frequency with a line chart above the bars to represent the cumulative percentages of the cases Box amp Whisker This option allows you to obtain a box amp whisker plot that can be used to examine the distribution of the variable It is especially useful to detect the presence of outliers and asymmetry in the data distribution The box includes values that fall between the first and the third quartiles about 5096 of the values The line in the middle ofthe box represents the median value while the whiskers extend to the farthest observations within 1 5 times the interquartile range measured from the nearest quartiles Values that are situated further than 1 5 times the interquartile range but within 3 times this distance are represented by the letter O for outliers Values farther than 3 times the interquartile range from the nearest quartile are represen
271. ting and using variable sets When working with large data files involving several hundred variables it may become difficult to locate the variables of interest The SETS OF VARIABLES command facilitates this task by allowing you to define several groups of variables in the data file and later restrict the variables displayed in dialog boxes involving selection of variables to those associated with a specific set To define a set of variables Select the SETS OF VARIABLES command from the DATA menu The following dialog box will appear Define set of variables x Current set Impact variable E Add Delete Rename Variable list Sorted OK x Cancel Help Click on the ADD button and enter the new set name You can enter a string of up to 20 characters to describe this new set Move all the variables that will make up the new set to the list box on the right of the screen DATA FILE OPERATIONS 47 Click on the OK button to exit from the dialog box or click on the ADD button again to create another set To modify a set definition From the Current Set drop down list box select the set you want to modify e Add or remove the variables from the set To rename a set of variables From the Current Set drop down list box select the set you want to rename Click the RENAME button and enter the new set name e Click on the OK button to confirm the new name To delete a se
272. tive statistics for any variable assigned as a dependent or independent variable Display includes mean standard deviation minimum and maximum values for each variable To obtain other descriptive statistics such as the skewness kurtosis mode median etc use the FREQUENCY command Sample output of descriptive analysis DESCRIPTIVE Variable SEX AGE SIBLING HOURSTV AGGRESS Mean Jump 9 54 Ton 1302 Dulko 4 Std Dev 5 90 1 48 67 du iG 5 Sis Minimum LOO 6 00 00 De D0 00 Maximum 200 HOT O10 2 00 225 00 66 00 59 59 59 59 59 Label Sex of the child Age of the child Number of siblings Hours per week spen Nb of agressive beh STATISTICAL ANALYSIS 81 Factor analysis EASY FACTOR ANALYSIS v3 0 performs two common types of factor analysis Principal components analysis and image covariance factor analysis The program has a good selection of features such as variable criteria to limit factoring varimax rotation factor scores for both rotated and unrotated solutions and flexible output The program can also handle up to 100 variables and an extremely large number of cases For more detailed information on factor analysis and the various statistics computed by EFA see EFA user s manual NOTE EFA v3 0 is an addon module written by Dr Darren Fuerst and sold separately To get more information on this module or to order a copy contact Provalis Research OPTIO
273. tively you may use the various commands in the HELP menu to display a table of contents of available help topics or search for a specific topic From the help window you can copy paste annotate and print help text using commands in the FILE and EDIT menu TUTORIAL 13 2 TUTORIAL PERFORMING STATISTICAL ANALYSES The following section provides a tutorial for beginning users of SIMSTAT It contains step by step instructions for computing descriptive statistics and performing a regression analysis on a sample data file The data is from a fictitious study about the influence of various factors on children s aggressive behavior during school recreational activities This tutorial uses data stored in the dBASE file SAMPLE DBF that comes with your program disk Step 1 Opening a data file for analysis The first step to performing a statistical analysis is to open a data file that you want to analyze After calling up SIMSTAT select the DATA OPEN command from the FILE menu This command pulls down an Open File dialog box that lists all valid files and subdirectories File name Folders c NsimstatwNdata Cancel cars dbf Cy c sample dbf Cy simstatw y data di Network Read only List files of type Drives xBase DBF gt amp c diskl voll Double click on the SAMPLE DBF file to open it If the file is not on the current drive or in the default directory you may use the following method
274. to rectify situations where there is too much imbalance in the proportion of bootstrap estimates falling on each side of the observed statistic median biased The Width option allows you to define a fourth interval Percentile table This option displays a percentile table whereby the number of cases in the sample is divided into equal categories These categories indicate the percentage of cases that fall below the corresponding value of the variable The number of categories computed in this table is determined in the Number option and is displayed with the corresponding variable values For example if you set this option to 4 the program will rank all valid cases divide them into four equal groups and then display the values that delimit the 25th lower quartile 50th median and 75th upper quartile percentiles Histogram This option produces a graphic display of the distribution of a numeric variable When this option is activated the program first separates the values into non overlapping intervals of equal width and then plots bars that represent the relative frequencies of each interval The Number of bars option allows you to specify how many bars or intervals to be plotted You can also superimpose a Normal curve on the histogram and visually assess how close your data distribution is to normal Type l error rate This option gives the proportion of samples that have reached a specified alpha level The standard display in
275. to specify how many decimal places to display in the data grid This option is used exclusively to control how numeric values are displayed in the data grid and does not affect the internal precision of the variable Missing values In SIMSTAT any blank cell is treated as a missing value also called a system missing value However you may also want to specify the reason for the missing value For example in a survey some respondents may not respond to a specific question because it does not apply to them They may also refuse to answer or may have simply forgotten to answer this question For each numeric variable it is possible to define up to 3 numeric values that will be treated as missing When performing statistical analyses or data transformations all cases containing a blank field or any one of these numeric values will be ignored By default these numeric values are treated as discrete values However It is possible to exclude a range of values by treating the second and third missing values as lower or upper limits For example if you have a variable containing the ages of respondents you may choose to exclude all subjects under 18 years by setting the second missing value to 17 or 17 99 if the variable is measured on a continuous scale with up to 2 decimal places and selecting the less than or equal radio button You may also exclude cases equal or above a specific value by specifying the upper limit in the third missing valu
276. u C Somers D symmetric Somers D X dependent Somers D Y dependent Gamma Spearman s Rho Pearson s r Regression slope Regression intercept Student s T Student s F ann Whitney Wilcoxon W value Sign test Z value Kruskal Wallis edian test Z value Percentage of agreement Cohen s kappa Scott s pd Free marginal nominal scale Krippendorff s r bar Krippendorff s R Free marginal ordinal scale Descriptive statistics Confidence interval Percentile tabl Histogram Orientation Minimum value Increment value Nb of bars intervals Normal curve Type I error rate Displays the dialog box RANK Syntax RANK varname Description SCRIPT LANGUAGE REFERENCE 219 Replaces the values of the current variable by their rank If ties occur the mean rank of the tied values is used Missing values are excluded Examples RANK FINALNOTE Tostore the rank in another variable use this command in combination with the COMPUTE command as in the following example COMPUTE FINALRANK FINALNOTE RANK FINALRANK RECODE Syntax RECODE value value varname Description value I The RECODE command provides an easy way to make multiple changes to the values of numeric variables or to collapse values of a continuous variable into categories The recode expression can consist of several transformations enclosed in parentheses each including one or mor
277. u define the number of decimal places for numeric variables Note that the length of a numeric variable with decimals includes the decimal point a leading zero and an optional minus sign The minimum length for a numeric variable that contains one decimal position is therefore 3 unsigned or 4 signed The maximum number of decimal places permitted by SIMSTAT is 17 Description The description column allows the entry of a variable label up to 60 characters long that can be used to describe in more detail the content of the variable You may leave this column blank if you wish since it is always possible to later add or edit those labels by using the DEFINE VARIABLE command Use the Up and Down arrow keys to move around the variable list and the Left and Right arrow keys or the Tab and lt Shift Tab gt keys to position yourself on the field that you want to modify To insert a new variable Position the cursor on the last row and press on the Down arrow key 26 SIMSTAT for WINDOWS To modify the position of a variable in the list Click on the left border of the row you want to move While holding down the mouse button drag the row toward its new location Release the mouse button to drop the variable at its new location When you have finished defining the structure information click on the OK button You will be asked to specify the name of the data file you want to create If a file with a simila
278. uency for each category of the independent variable are placed side by side When the overlayed barchart is chosen the bars are displayed on a 3 axis plane where bars representing the frequency of each category of the independent variable are place on different rows This chart type is available only in 3 D view While this kind of chart is very popular we strongly recommend against its use since it is virtually impossible to determine the exact heights of the bars or compare the heights of bars located on different rows In a stacked barchart the bars representing the frequency of each category of the dependent variable are stacked on top of each other The 100 bars type is similar to the stacked barchart in that the bars for each category of the dependent variable are stacked on top of each other However like a pie chart each bar represents the proportion of a category of the independent variable from the total number of observations in a specific category of the dependent variable This type of barchart is especially useful if you want to STATISTICAL ANALYSIS 79 compare the proportions between different categories of the dependent variables rather than the absolute frequency Perspective This option allows you to specify whether the barchart should be displayed in 2 or 3 dimensions A sample output of crosstabulation analysis 80 SIMSTAT for WINDOWS Descriptives DESCRIPTIVES immediately computes univariate descrip
279. ues with the next statement When used with the GOTO command this command can increase the flexibility of the program by allowing SIMSTAT to switch to different parts of the program or perform specific actions under certain conditions Valid expressions A valid expression consists of an existing memory or a data file variable name and a valid relational operator followed by a string or numeric expression This numeric expression may consist of a numeric constant another numeric variable or an equation with arithmetic operations and mathematical functions see page 176 for a list of available functions NUMERIC lt gt lt gt lt gt numeric expression STRING lt gt string Examples IF SANSWER1 gt 10 THEN GOSUB AgeError ELSE DESC all IF SNAME John THEN GOTO MonthlyAnalysis IF DB SALARY gt TRUNC SMAXS 1 1 THEN LET DB SALARY SMAXS 1 1 SCRIPT LANGUAGE REFERENCE 199 IMPORT Syntax IMPORT filename Description This command reads a data file produced by other applications and creates a new data file that may be used by SIMSTAT The extension of the supplied filename is used to determine which kind of file is to be imported DB Paradox data file v3 0 v5 0 WK Lotus 123 v1 0 v5 0 XLS Excel v2 1 v5 0 WB or WQ Quattro Pro v1 0 v6 0 WR Symphony v1 0 v1 1 SIM SIMSTAT for DOS v1 0 v3 5 SYS SPSS P
280. uld mean administering a test with a Cronbach s alpha value of zero prior to the training and with a undetermined value after the training Yet even in this situation item analysis may still be useful since information about response patterns may allow test developers to identify flawed items unintentional clues misleading formulation etc and ineffective or needless instruction or provide useful information about the kind of errors made by individuals their misconceptions etc 104 SIMSTAT for WINDOWS OPTIONS Keys location This option allows you to specify whether the key responses are stored in the first record of the database or in a key file If the key file option is selected the program will look in the data file directory for a file with the same file name as the data file but with a KEY extension This file is a plain text file where each line contains the name of a variable followed by the value of the correct response If a key file is not found or if a variable has no key response the analysis will automatically stop NOTE Even if the responses to the item have been previously dichotomized as correct or incorrect you still need to specify the key value for each variable by providing the value used to represent a correct response Exclude case with missing This option allows you to exclude cases containing missing data either system or user missing values from the analysis If disabled all missing data will be trea
281. up The Table Content group box allows you to request other statistics in addition to frequencies to be included in the cells of the table To obtain the desired statistics enable the corresponding check boxes Row percentages Column percentages Total percentages Expected frequencies Chi square residuals e Standardized chi square residuals When performing multiple response crosstab analyses an additional option allows you to specify whether the percentage will be based on the total number of responses or on the number of respondents STATISTICS PAGE The Statistics page allows you to specify which statistics should be computed on the table You can specify as many statistics as needed in a single analysis 78 SIMSTAT for WINDOWS Nominal level statistics Chi Square and likelihood ratio statistics Phi for 2 x 2 tables or Cramer s V for larger tables Contingency coefficient Ordinal level statistics Spearman s Rho Kendall s tau b Kendall Stuart s tau c Goodman Kruskal s gamma Somers d Continuous or interval level statistic Pearson s R CHART PAGE Barchart If this option is activated the program displays a 2 dimensions barchart that provides a graphical presentation of the relationship between the dependent and the independent variable Type The Type option offers a choice between 4 different bivariate barcharts to display n a clustered barchart the bars representing the response freq
282. utomatically adjusted to the size of this original sample When disabled you can use the Size option to specify how large each bootstrap sample will be The sample size can range from 1 to 100 000 cases Number of samples This option allows you to choose the number of times resampling is carried out on the data This number can range from 1 to 32 000 Seed SIMSTAT automatically provides a seed value for the random number generator that drives the simulation analysis Alternatively the Initial Seed Value option can be used to specify a seed value To instruct SIMSTAT to randomly choose a new seed value click on the Shuffle button SIMSTAT outputs the seed value with the simulation results This value can be used to regenerate the same data at a later time or to compare various estimators using the same bootstrap samples 74 SIMSTAT for WINDOWS Breakdown analysis The BREAKDOWN procedure computes descriptive statistics for sub groups ofthe sample This procedure can display a single line of statistics including the means standard deviations minimum and maximum values for the dependent variable Y within groups defined by the values of the independent variables X More detailed statistics can also be obtained for each group and for the entire sample The dialog box also allows you to restrict the number of groups to a specified range and to obtain a multiple box amp whisker plot that can be used to compare the distribution of th
283. variable to express the values of a string variable DATA TRANSFORMATION 49 Examining data distributions SIMSTAT provides a special feature that allows you to inspect the distribution of a variable and assess the effect of common transformations on the data distribution To use this feature position the data grid cursor on the numeric or date variable you want to examine and select the VARIABLE STATISTICS command from the DATA menu A dialog box will appear displaying various statistics including mean median standard error of the mean variance standard deviation skewness kurtosis minimum and maximum values etc as well as 3 graphs i e a box plot a histogram and a normal probability plot Descriptive statistics for HRSPWR Mean 105 0 Variance 1503 Kurtosis 519 Median 95 00 Standard dev 38 76 S E Kurtosis 245 Mode 150 0 Standard error 1 938 Skewness 1 034 Minimum 45 00 Sum 42033 S E Skewness 122 Maximum 230 0 Range 184 0 Valid 400 0 Missing 6 000 LTT es riginal data M By default the distribution statistics and the graphs displayed depict the untransformed values of the variable A list box allows one to temporarily transform those values using common transformation formula and examine the resulting distribution Navigation keys at the lower right side of the window allow you to quickly move from one variable to another To perform a transformation on your data se
284. vides a short non technical introduction to the bootstrap technique followed by a description of SIMSTAT s particular implementation of bootstrapping methodology Potential applications for researchers statistical consultants and for students and teachers of statistics are also presented For further information about bootstrap methods and its applications you can read the articles of Efron and his colleagues Diaconis amp Efron 1983 Efron 1981 Efron amp Gong 1983 Efron amp Tibshirani 1993 Wasserman and Bockenhold 1989 also provide an excellent introduction to bootstrap methodology while Stine 1989 offers a comprehensive presentation of its potential applications What is bootstrap simulation Bootstrap simulation is a resampling technique whereby initial sample subjects are treated as if they constitute the population under study By replicating those data an infinite number of times we then draw at random from that population a large number of samples each the same size as the original sample By computing a statistical estimator of interest such as a mean or a correlation between two variables for every bootstrap sample this resampling procedure recreates an empirical sampling distribution of this estimator The main advantage of such a procedure is that the sampling distribution is not mathematically estimated but empirically reconstructed based on all the original characteristics of the data So it automatically takes
285. w variables Enabling this option causes SIMSTAT to prompt you to confirm the creation of new variables The current data file format used by SIMSTAT requires you to determine in advance the physical size of a new variable in the data file This section allows you to specify the default width and number of decimal places of new numeric variables created during a transformation or an analysis Width The width of a numeric variable should be set to the maximum width that a value can have The maximum width for the numeric type is 19 Number of decimals The number of decimal positions if the field type is numeric Note that the length of a numeric field that contains decimals includes the decimal point a leading zero and an optional sign The minimum length for a numeric field that contains one decimal position is therefore 3 unsigned or 4 signed The maximum number of decimals permitted is 16 The DATA grid section provides two options that allow you to specify how the datasheet will display the content of your data file Display font This option allows you to change the font used to display the data in the data grid By choosing a different font size it is possible to increase or decrease the number of rows and columns displayed on a single screen Minimum cell display width While it is possible to individually adjust the display width of each variable in a data file this option allows you to override these values by setting a m
286. with Template characters L or Y The Function string consists of a leading character followed by one or more formatting characters If the Function string is present the character must be the first character in the picture string with its formatting characters immediately following and it may not contain spaces If a Template string exists as well it follows the Function string A single space separates the Function string and the Template string Function string characters allowed for numeric values are B left justify C display CR after positive numbers X display DR after negative numbers Z blank a zero value enclose negative numbers in parentheses Function string characters allowed for strings are R inserts unassigned template characters converts all alpha characters to upper case The R Function requires a Template the Function does not The Template string describes the format on a character by character basis The Template string is made up of special characters which have specific results and optional unassigned characters which either replace characters or are inserted in the formatted string depending upon the absence or presence of the 9 R Function string APPENDIX A XBASE SYNTAX AND FUNCTIONS 249 Template assigned characters are as follows A N X are place holders and are interchangeable L displays logical values as T or F Y displays logical values as Y or N converts the corresp
287. wn Move one screen up Move one screen down Move to the beginning of the current line Move to the end of the current line Move to the first line of the script Move to the last line of the script Move to the beginning of the next word Move to the beginning of the previous word You can also move to a specific string within the current script by choosing the FIND command in the EDIT menu or by pressing lt Ctrl F gt The followings editing commands are also available Key lt BackSpace gt lt Del gt lt Ctrl Ins gt or lt Ctrl C gt lt Shift Del gt or lt Ctrl X gt lt Shift Ins gt or lt Ctrl V gt Ctrl Z lt Ctrl Shift 0 gt to lt Ctrl Shift 9 gt lt Ctrl 0 gt to lt Ctrl 9 gt Action Delete the character to the left of the caret Delete the current character or selected text Copy the selected text to the clipboard Delete the selected text after copying it to the clipboard Paste the text from the clipboard Undo the last operation Set the position of marker 0 9 to the current caret position Move to the previously set marker 0 9 usin SRIPTS 371 Using the RECORD SCRIPT Feature This feature automatically generates proper commands corresponding to the action you undertake using the menus and dialog boxes and appends them to the end of the current script To activate this feature select the RECORD option from the SCRIPT menu The Record keyword will appear on the s
288. x of the child Number Standard Standard of Cases Mean Deviation Error SEX 1 29 367690 TS Z8 ZH SEX 2 SO 19 09 LA se AS Be ove Pooled Variance Estimate Separate Variance Estimate F 1 tail ic Degrees of 1 tail 5 Degrees of deer Value Prob Value Freedom Prob Value Freedom Prob 1 46 314 4 94 57 000 4 92 54 33 000 Effect size 90 0 Confidence Interval R Sra ale 6827 To STI I IDE MEINE QUAS No OO 8096 STATISTICAL ANALYSIS 151 Sample output of a paired sample t test 152 SIMSTAT for WINDOWS Wilcoxon test The WILCOXON matched pairs signed ranks test is a procedure used to test whether two related samples have been drawn from the same population Like the sign test it computes the difference between the values of the two variables but takes into account the magnitude as well as the direction of the differences The Wilcoxon signed ranks test is the nonparametric version of the t test for paired samples OPTION Direction This option allows you to request the program to select either a one tailed or two tailed probability assessment Sample output of a Mann Whitney U Wilcoxon rank sum W test analysis MANN WHITNEY U WILCOXON RANK SUM W TEST AGGRESS Nb of agressive behavior With SEX Sex of the child Mean Rank Cases 40 48 29 SEX FOTO 3 9 5197 SOME Six 200 5 neu EXACT Corrected for ties U W 1 tailed P Z 1 tailed P ISL 5 9 1174 0 0000 ZA 6325

SIMSTAT - Provalis Research

Contents

Download Pdf Manuals

Related Search

Related Contents