Home

User Guide CIS EXTENDED DATABASE

1. 3 5 Changes in Future Editions The Database Records 4 1 Field 1 Location 0 0 0 00200 AD Field Di Titler heer rece se eevee Ses dele AIDA 4 3 Field 3 Authors 0 000000 000s 4 4 Field 4 Key words and phrases 4 5 Field 5 Searchable author information 4 6 Field 6 Additional Searchable Information 4 7 Field 7 Administrative codes 0 Hints for Searching CIS ED Source Abbreviations Special Indexes 7 1 7 2 7 3 7 4 7 5 7 6 Subject Index nan a a sa na i we ee ea Author Index Index of Key Words and Phrases Index of Additional Searchable Terms Inverted Indexes 0 000000 00 eae Thesaurus sia a d Ako A Rem E geie ae D ot Abstracts and Book Reviews ii 35 39 40 40 42 45 46 47 48 IV Coverage 9 The Gap Project 9 1 Editorial policy s saanen A A eS 9 2 Contents tai ane ates ae he DR RR AA Kee E es 9 3 General Acknowledgments 10 License Information and Legal Matters 11 History of the CIS ED 11 1 Information from printed volumes of CIS 11 2 Acknowledgements 12 Errors Omissions and Corrections iii 50 50 50 51 53 54 56 56 58 iv Part I Quick Start 1 Quick Guide to Installation Read this chapter of the User Guide before installing the database 1 1 All Installations Last minute Changes Before beginning inst
2. James Stein in Field 4 A search of the database shows that in addition to these records an additional 41 records contain James Stein in the title field Field 2 7 4 Index of Additional Searchable Terms This short file named F6 V97 is a tabulation of the semicolon separated keywords phrases in Field 6 of the database The two fields count and 45 term are separated by a blank the count is right justified in a field of width 4 In 1994 thousands of database records were reT Xed with regard to math mode For example L2 originally entered as L2 was changed to Lo entered as L_2 Similarly t test t test became t test t test To simplify searches common expressions such as t test and F statistic were placed in Field 6 stripped of signs and with blanks replaced by hyphens Searching on the stripped version of one of these math fragments is guaranteed to retrieve all records containing the TE Xed version s For example the following entries from F6 V97 1996 Edition 239 t test 1 Two sample t test shows that there are 240 database records with the term t test in Field 6 and accordingly either the term t test t test or t test t test in Fields 2 or 4 7 5 Inverted Indexes The fastest search software makes use of inverted indexes to the database records Most users will not need to know details of the structure
3. cis 2 26 Fine Print The software and the documentation contained in this section are copyright Copyright 1994 7 Ronald A Thisted All Rights Reserved Portions of the program for the search command are Copyright 1993 by Douglas M Bates No warranty of any sort is provided with this software Authorized users of the Current Index to Statistics Extended Database CIS ED are granted a nonexclusive royalty free license to use this software in conjunction with the CIS ED ClScd is provided to you at no cost You may make as many copies of CIScd as you or individuals in your organization if you have an institutional CIS Database license can personally use You may not sell or redistribute the program in any other manner except to recover the actual cost of the diskettes you may use to provide the program to others 23 Part III User Reference 3 Overview The Current Index to Statistics is an organization sponsored jointly by the American Statistical Association ASA and the Institute of Mathemati cal Statistics IMS Under the direction of a Management Committee the organization produces an annual printed index to the statistical literature appearing in the previous year and a cumulative database issued in elec tronic form The 1997 edition of the CIS Extended Database CIS ED contains approximately 28 Mb of bibliographic information including en tries from Volumes 1 22 of the Current Index to Statistics covering the
4. Best Alvin M III Young Forrest W Hall Robert G Multidimensional scaling Data analysis etc 93JRSS B 55 39 52 J Modelling complexity Applications of Gibbs sampling in medicine Disc p53 102 Gilks W R Clayton D G Spiegelhalter D J Best N G McNeil A J Sharples L D Kirby A J gt gt Near Success gt A better way to accomplish the last search is to conduct a gt search on the string of characters Best A in the gt author field gt swap 61 main records 8 backup records gt author Best A 12 7 of 64 records meet criteria 7 main records 64 backup records gt gt Real Success gt gt We can save these results to a file in exactly the format above gt gt save best cis 7 records written to file best cis in SHORT format gt quit If we suddenly were to discover that we forgot something we can start CIScd again and the main and backup results will be exactly as we left them since the temporary files are not erased at the end of a session If you would like to erase them if there are some big search results for instance that you won t need again you can add the word clear to the end of the quit request 2 7 Commands A list of commands can be obtained at any time with the help request This produces a list like this help Prints this text swap Interchange main and backup searches count S
5. and Technical Reports Considerable interest has been expressed in title author and key word information for theses and technical reports In order for this effort to be financially feasible for university departments and research groups must submit records electronically to the editors Current information submission standards and instructions can be found at our Web site http www stat uchicago edu cis 27 If any thesis or technical report records are submitted they will be included on the CD ROM in separate files Records with such information would not be intermingled with records on journal articles books etc 3 5 2 Reviews Beginning with the 1998 edition book reviews will be distinguished from journal articles in Field 1 using a new code for reviews in column 28 of Field 1 3 5 3 Non English Titles In future editions we hope to include the title of articles in the original published language where possible as well as an English translation The keyword field s would contain English language keywords that may over lap words and phrases that would appear in an English translation of the title 28 4 The Database Records This chapter describes the information that may be contained in each of the seven fields of a record The first four fields are used in creating the printed version of the Current Index The remaining fields are included in CIS ED to make searches easier and for administrative purposes Thr
6. database that match the first six case insensitive characters of each of the keywords listed on the request Records that match all of the specified keywords are retained as the main search result If at least one record was found the previous main search result becomes the current backup search result 2 8 2 Refinements The syntax for the requests that refine searches is more flexible Each request has the form REQUEST Tokeni Token2 TokenK where REQUEST represents a command such as NOT AND ANDOR AUTHOR or ALLAUTH If the first four characters of REQUEST match a valid request the corresponding request will be honored Thus AUTH is an abreviation for AUTHOR Each token consists either of a string of non blank characters or a string of characters that include blanks enclosed in double quotation marks Neither the double quotation mark nor the tilde 7 is used in the CIS and these characters are not valid in search tokens The number of tokens on a single request is not limited for practical purposes After each successful refinement request the previous master search result becomes the current backup and the results of the request become the current master Here are the common refinement commands and a brief description of their function 14 and Finds records in current search that have ALL search terms andor Finds records in current search that have ANY search terms not Finds records in current search th
7. its lower case equivalent The CIS ED Technical Reference Manual describes the formats in detail This document can be obtained from the CD ROM and from 29 http www stat uchicago edu cis A brief synopsis of the contents of Field 1 follows Z 4 2 Table 1 Publication Types Book Article in journal or other periodical not electronic Article in proceedings or edited book Computer readable databases software etc Article or entry in electronic journal or periodical Administrative record no bibliographic information Dissertations not used in 1997 Technical Reports not used in 1997 HUONWADAD Columns 3 18 contain a publisher abbreviation 19 20 spaces 21 25 number of pages 26 the letter p 27 an optional character used if necessary for uniqueness Columns 3 10 Journal abbreviation 11 13 volume 14 16 issue num ber 17 21 beginning page 22 hyphen 23 26 ending page 27 an optional character used for special circumstances Columns 3 12 Abbreviated title of edited book 13 16 volume number if any 17 21 beginning page 22 hyphen 23 26 ending page 27 optional character Columns 3 18 Publisher producer or distributor abbreviation 19 20 spaces 21 24 size of file 25 26 kb mb or gb to denote size units 27 optional character Columns 3 10 Journal abbreviation 11 13 volume number 14 16 issue number 17 25 publication specific item locating information 26 code indicating fo
8. one edition to another for exam ple they may carry editorial records and copyright notices It is unlikely that ordinary searches of the database will retrieve any of them if they are retrieved their nature should be obvious and they are not likely to be confused with bibliographic records 11 History of the CIS ED 11 1 Information from printed volumes of CIS The usefulness of this edition of CIS ED depends in part on the efforts of past editors of the Current Index to Statistics printed volumes from whose files major portions of it were developed Brian Joiner Volumes 1 5 covering years 1975 1979 James Gentle Volumes 6 10 1980 1984 Richard Burdick Volumes 11 15 1985 1989 and Edward Gbur Vol umes 16 20 1990 1994 The work of their contributing editors editorial collaborators and editorial assistants acknowledged in each printed vol ume of the Current Index to Statistics was essential to preparing the original CIS ED databases Beginning with the 1996 edition of CIS ED and Volume 21 of the print version data collection and editorial efforts for the electronic and print editions have been integrated The new records on the CD ROM and in the print volume are now produced from a single set of edited records Items considered for indexing in CIS include journal articles books and edited books articles from edited books and conference proceedings provided that the proceedings are available to non conference partici pant
9. or Van Der Velt and these variations would be re flected in Field 3 for the corresponding records Similarly it is possible to find entries under Mendez Gonzalez Mendez y Gonzalez Dong Ho Park and SPSS Inc note the missing comma between the two words Even if you know which form is correct or which form the author prefers you cannot be sure that either the original journal or CIS ED 36 has it right It is best to scan the author index before doing any compre hensive searches for an author whose name might have several variants see Section 7 2 It may be helpful to search on a single word in the name unless that is likely to yield too many unwanted records or to use inter sections for firm names don t include Inc Ltd etc with or without a preceding comma in a search string since there is little consistency in usage Correction notes addenda etc Authors may submit a correction note addendum or acknowledgment of priority months or years after the original paper appeared The location of a correction note is indicated in parentheses after the title of the original paper to which it refers Discussions Some papers are immediately followed by comments and often the au thors replies Such remarks can usually be found by searching on a phrase in the title of the main paper Many entries indicate the presence of such discussion in parentheses after the title of the main paper espe cially if the
10. or con tent of these indexes but this section provides the details for those who wish to use the inverted indexes in their own programs to access the database Douglas Bates and Paul Tukey defined the format used for the Inverted Indexes which were originally designed as part of their client server search software for CIS ED on Unix based computers That search package and its successors provided on the CD ROM and available from Statlib and the CIS web page http www stat uchicago edu cis includes a program to create an inverted index from the actual database files The index for the entire CIS ED is approximately 60 as large as the database itself The inverted index makes it possible to look up words and to retrieve a list of all records in CIS ED containing those words In CIS ED there are two sets of inverted indexes which reside in the directories INDEX and INDEX2 respectively These index sets are identical in structure and similar in content and size 46 7 5 1 Inverted Indexes in INDEX These indexes are used by several search programs e g the Bates Tukey software and ClScd There is one index file for each of the data files in the database Each such index file contains a list of the words contained in Fields 2 through 6 of the corresponding database file together with a coded list of positions at which the word can be found Here a word is defined to be a maximal length string of letters digits and und
11. preparation of Trumbo Gbur and Suess 2 ASA provided financial support for data entry and editorial assistance in covering Journal of the American Statistical Association 1955 92 3 The Technometrics Management Committee provided financial sup port for data entry and full editorial cooperation in preparing records for Technometrics 1959 74 51 William Q Meeker organized a team of 27 associate editors and frequent contributors to Technometrics to assist with the selection of key words for these volumes 4 The International Biometric Society provided financial support and Klaus Hinkelmann and Alexandra Kapatou provided editorial sup port including assistance in selecting key words in preparing records for Biometrics 1965 74 5 The Biometrika Trust supplied back issues to assist in the prepara tion of records for Biometrika 1965 74 6 Using CIS funds draft coverage in 1993 and full coverage 1994 was included in CIS ED for Applied Statistics 1965 74 For the entire period 1965 92 the numbers associated with compu tational algorithms and remarks on algorithms have been included and appropriate cross referencing of algorithms and associated re marks has been completed 7 Gap coverage for the journals listed below was completed in 1994 Advances in Applied Probability 1969 74 The American Statistician 1947 74 Australian Journal of Statistics 1959 74 Canadian Journal of Statist
12. short forms are similar in style to those used in Mathematical Reviews published by the American Mathematical Society There are some differences because the contractions used in CIS ED were chosen to be consistent across a much broader range of statistical publications than is found in Mathematical Reviews AB V97 This file is similar to ABB V97 It is easier for humans to look at but harder for computers to parse ABAMS V97 This file is related to ABB V97 and AB V97 For those publica tions and publishers that are represented both in CIS ED and Math ematical Reviews this file gives the short form e g Ann Statist of the name used by Mathematical Reviews ABBRAMS V97 This file is similar to ABAMS V97 but gives the long form e g The Annals of Statistics of the name used by Mathematical Reviews 39 7 Special Indexes The 1997 edition of CIS ED contains a Subject Index a Keyword Index a list of additional searchable terms an Author Index two sets of Inverted Indexes and a Thesaurus of related phrases The indexes and Thesaurus are located in the platform specific directories of the CD ROM in the subdirectories INDEX and INDEX2 7 1 Subject Index The subject index SUBJECT V97 is an ASCII file approximately 0 8 Mb in length tabulating the number of records in which various words occur somewhere in Fields 2 titles 4 key words phrases or 6 additional searchable information The subject index can b
13. should contain only jointly authored papers gt author Switzer P 8 of 13 records meet criteria 8 main records 13 backup records gt These papers contain an author with one of the variants above gt list 82JIAMaGeo 14 433 444 J A prior probability method for smoothing discriminant analysis classification maps Switzer Paul Kowalik William S Lyon Ronald J P etc gt whatis JIAMaGeo Abbreviation JIAMaGeo corresponds to Journal of the International Association for Mathematical Geology 2 9 Output formats The search results main and backup are stored using the same cryptic format that the Current Index Database itself uses we call this format the raw format By default requests to list or display the current search will produce screen output in an abbreviated human readable format called short format A longer format long lists the contents of each fields on a separate line Refer format outputs each record in a form that can be used by the UNIX bibliographic utility Refer as well as by some other bibliographic management software such as EndNote on the Macintosh 18 Tex format produces a file that can be run through the TEX typeset ting language to produce reasonably readable bibliography lists These may require some hand editing 2 10 Setting output format There are two ways to change the output format from the default value of short One option is to permanently change the default usin
14. the UCcis search software 7 6 Thesaurus Current Index to Statistics printed volumes contain lines suggesting items to See or See Also Some of these have to do with how the printed volumes are organized e g time series is found under time but not under series and some offer suggestions as to alternative terminology e g mark recapture is related to capture recapture LISREL stands for 47 linear structural relations clustering and classification are related ideas The latter kind of information can also be helpful in searching CIS ED This edition contains the file SEEALSO V97 which incorporates some cross listings taken from printed CIS volumes with others that we thought would be particularly useful for computer searches This file contains variable length records each containing a list of related words The two or more items within a record are separated by various symbols indi cating transition from a general to a specific concept gt full version of an acronym various levels of similarity or parallelism or an antonym To improve the thesaurus CIS solicits comments from users and con tributed entries from experts in various fields 8 Abstracts and Book Reviews Over the long term it is our hope to be able to include abstracts for all refereed journals affiliated with ASA and IMS as well as abstracts for some other important publications Abstracts accompany entries for the foll
15. the project Thanks are also due to former students at California State University Hayward who have assisted in a variety of ways includ ing proofreading and the attachment of correction notes to referent pa pers Rima Bannerjee Jeffrey Hill Huu Manh Hoang Pamela D Ochoa Jason H Stover Eric A Suess Douglas Whitehouse References 1 Trumbo Bruce E and Burdick Richard 1991 Index of IMS Sci 2 entific Journals 1960 1989 Institute of Mathematical Statistics Trumbo Bruce E Gbur Edward E Jr and Suess Eric A 1993 Journal of the American Statistical Association Subject and Author Index 1955 1991 American Statistical Association Trumbo Bruce E Munneke Brian M and Wichura Michael J 1996 Index of The American Statistician 1947 1994 Computer searchable database with search software American Statistical As sociation In preparation expected early 1996 53 10 License Information and Legal Matters CIS ED is distributed for the Current Index to Statistics by the IMS Busi ness Office 3401 Investment Blvd Suite 7 Hayward CA 94545 USA and the American Statistical Association 1429 Duke Street Alexandria VA 22314 USA Information about ordering subsequent editions of CIS ED will be announced in ASA and IMS journals and may also be obtained from the offices above Search software supplied on the CD ROM version is copyright by the authors of the software who make it available to users of
16. this save the current search in raw format Any file that is saved in raw format can be loaded into the main search results by using the retrieve command Thus the contents of the main results will be the same both before and after the following sequence of commands is given gt save best raw raw gt retrieve best raw 2 13 More more Records that scroll off the top of the screen cannot be looked at on MS DOS machines without doing another search For this reason the more command can be used to cause screen output to stop after roughly every 24 lines of output The next screen of output is produced by typing the Enter or Return keys on the keyboard This paging feature can be toggled on and off by repeated uses of the more command Alternatively one can use the following commands to control this fea ture gt set more enables paging gt set nomore disables paging gt more 60 stops after every 60 lines or so 2 14 Screen display list display The list and display commands show the results of the current main search at the screen These names are synonyms The output format can be selected as described above under Output format 2 15 Showing the next k records next A convenient way of looking at a set of search results is to display from two to four records at a time depending on the display format currently in effect The command gt next 4 20 will cause the next four records to be display
17. user from having to know anything about the specific formats for the various databases by automatically re trieving all relevant information relating to a record and exhibiting it in a standard bibliographic format such as Refer or BIBI px Descriptions of the abbreviation files abstracts and special indexes are contained in the CIS ED Technical Reference Manual 25 3 2 Search Software Search software contributed by several different authors for Unix DOS Windows and Macintosh computers are included on the CD ROM Al though the Current Index to Statistics Management Committee does not itself support license recommend or warrant search software it does en courage software development makes space available on the CD ROM for software contributed from a variety of sources and provides pointers to software on its Web page The CD ROM contains platform specific directories that contain con tributed software for searching the database All of the search software distributed on the CD ROM or a more current version of that software is also available on the CIS Web site http www stat uchicago edu cis and on Statlib Please consult the CIS Web page for updates revisions and the latest version of search programs for CIS ED The directories in which software can be found are DOSSFT for DOS specific programs WINSFT for Windows specific programs and UNIXSFT for programs intended for the Unix environment Macintosh use
18. variants are gt config to restart using the default configuration gt config alt cfg to restart using alt cfg 2 24 Quit quit This command will exit the program On the Macintosh the window from the search session will remain active so that it can be printed and or saved to a file Cut and paste are also active and after the quit command is 22 issued the contents of the window can be edited or formatted to delete dead ends or errors in the search for instance To completely quit on the Macintosh the Quit option must be selected from the File menu or COMMAND Q must be typed from the keyboard CIScd creates temporary files to hold the results of searches and re finement requests These temporary files are not deleted on quitting the program unless you specifically ask for them to be removed This makes it possible to exit from the program and then to start the program again later taking up exactly where you left a search project If you are certain that you will not need the current intermediate results you can give the command gt quit clear which will delete the work files If you do not specify clear you will be asked whether you wish to delete the files In this case the files will be deleted only if you explicitly request them to be 2 25 Modifying the cis cfg when Files are on Hard Disk Documentation for this feature is currently available from the CIS ED website at http www stat uchicago edu
19. Current Index to Statistics User Guide CIS EXTENDED DATABASE 1997 Edition Sponsored by The American Statistical Association and The Institute of Mathematical Statistics Ronald A Thisted Department of Statistics The University of Chicago Michael J Wichura Department of Statistics The University of Chicago Klaus Hinkelmann Department of Statistics Virginia Tech Bruce E Trumbo Department of Statistics California State University Hayward November 1997 Editor Current Index to Statistics RONALD A THISTED 1995 Editors CIS Printed Volume KLAUS HINKELMANN Volumes 21 1995 Edward E Gbur Jr Volumes 16 20 1990 94 Richard K Burdick Volumes 11 15 1985 89 James E Gentle Volumes 6 10 1980 84 Brian L Joiner Volumes 1 5 1975 79 Editors CIS ED MICHAEL J WICHURA 1995 Ronald A Thisted 1994 95 Bruce E Trumbo 1990 93 CIS Management Committee 1997 Paul Shaman CHAIR Douglas M Bates Michael J Meyer Julia A Norton Alan M Zaslavsky ISBN 1 883276 52 7 ISSN 1094 7469 Copyright 1997 American Statistical Association amp Institute of Mathematical Statistics All rights reserved Licensees of the 1997 CIS ED are granted a non exclusive royalty free license to copy and to distribute this User Guide to authorized users of the database Each such copy must contain the title page and this page in its entirety No bibliography index or other work whether in computer readable printed or other for
20. ackup records The 13 remaining records now contain both William and Krasker somewhere in them This search was virtually instantaneous What were the characteristics of the 3 records omitted in the refine ment Two of them were by others commenting on a procedure of Krasker and Welsch The other was by W S Krasker and J W Pratt an article that we would like to have found The difficulty is that names of authors are rarely consistent across journals or other publications and often au thors names are misspelled The author index on the CIS CD ROM can be browsed to find similar names represented on the database 2 5 Restricting the Search Here is an alternative strategy for refining the initial search use a search restricted to the author field of the records After the initial search on Krasker we could proceed with gt search krasker gt author krasker 14 of 16 records meet criteria which finds the appropriate entries 2 6 A More Difficult Example How should we find articles written by Al Best Here is a transcript that illustrates several features of ClScd gt gt First do an initial SEARCH which uses the inverted index gt gt search best 836 records found 836 main records 13 backup records That took over a minute due to the number of records that had to be retrieved A moment s reflection suggests gt that many of these records will deal wi
21. allation please be sure to read the file README TXT at the top level of the CD ROM volume This file will contain information about any last minute changes in procedure that differ from those described here For More Information Information latest versions of search software corrections and announce ments concerning the Current Index to Statistics are posted regularly on the World Wide Web The CIS home page can be found by pointing your Web browser to http www stat uchicago edu cis Updating an Earlier Version of CIS ED Previous editions of the CIS ED must be removed before installing the 1997 edition Do not attempt to update the 1996 edition of the CIS Ex tended Database by just adding the additional years from the 1997 edition of CIS ED Doing so will corrupt the database and will produce misleading or erroneous results Your license agreement prohibits mixing editions in the same installa tion The main reason for this is that the additions that each new printed CIS volume make to the database extend far beyond publications in the most recent year Approximately 25 of the entries in an annual printed volume carry earlier nominal publication dates In fact CIS editors are constantly trying to pick up items overlooked in previous years As we prepare each edition of CIS ED these additions will be sorted according to the years of publication of items indexed In general there are other good reasons no
22. alled on individual computers and or on a network for on site use by employees students and visiting scientists There is no limit on the number of stations at the site or the number of simultaneous users at the site However off site access via modems or networks is not permitted Administrative hardware and or software controls should be in place to prevent unauthorized transfer of the CIS ED or parts of it off site 54 University Campus License CIS ED is licensed for use by faculty staff students and authorized visit ing scientists holding currently valid passwords for access to the computer system or network of a university campus CIS ED may be accessed from on or off campus Administrative controls on passwords their use and their renewal should be in place to prevent CIS ED or parts of it from being transferred to or used by those who are not currently part of the local campus community Available only to educational institutions Commercial License CIS ED is licensed for use on an unlimited number of computers at a single site as well as by staff holding currently valid passwords for access to a single computer system or network of a corporate entity CIS ED may be accessed either locally or remotely Administrative controls on passwords their use and their renewal must be instituted to prevent CIS ED or parts of it from being transferred to or used by those who are not currently employed by the licensee Four stat
23. and Clustering Names may be followed by a ROLE ABBREVIATION in parentheses as discussed in Section 4 3 3 4 3 2 Individual Authors Authors names are given in inverted order The syntax for an individual author s name is e SURNAME or family name e Tf necessary a comma followed by FIRST NAMES and or initials e If necessary another comma followed by a SUFFIX such as Jr II Esq etc e If necessary a ROLE ABBREVIATION in parentheses 31 Thus a person s name may have zero one or two commas Initials in names are always followed by exactly one period Initials are separated from subsequent text by a single blank space following the period unless the next character is a comma semicolon or character Here are some examples of names without role abbreviations Editors Bhagawandas Chan Fu Chen Sanathanan Lilly Blattberg Robert C Brown L D De ak Istvan Erd H fo s Paul Le Breton A van der Horst Charles M Mu noz P erez J Garcia Carrasco Pilar Harrell Frank E Jr Pickands J III Records originating with CIS Volume 14 1988 and later and all Gap records have TX codes for accent marks on authors names and some earlier entries have also been edited to include correct accent marks 4 3 3 Role abbreviations Parenthetical roles Auth Ed Rev Transl for author editor reviewer and translator respectively follow all other components of a name A few oth
24. at if the source used neither an umlaut nor the spelling with the extra e then the briefer spelling alone will appear in the CIS ED Although in the 1997 edition we have made an attempt to systematize usage the pre 1988 practice for non German umlauts diereses and cer tain Scandinavian letters does not appear to have been consistent so for these it is still safest to search on both forms Examples Noel and Noeel Bjorn and Bjoern Ake and Aake Also journals sometimes print Hun garian umlauts as German umlauts so you may see Csoergoe as well as Csoergo for Cs rg Transliterated Cyrillic Names Standardized spelling of many Russian and other names which have been transliterated from Cyrillic texts such as Chebyshev Gikhman Markov Prokhorov and Skorokhod have been used in Field 5 A list of the names for which standardized spelling has been introduced is contained in the CIS ED Technical Reference Manual Miscellaneous For important words that can be spelled either with or or with our use the U S spelling In other cases where U S and British spelling differs it is best to try both search strings The abbreviations U S U K U S S R U N and W H O do not contain spaces in Fields 2 3 and 4 they should always contain periods For the convenience of those using software based on inverted indexes which may treat periods as word breaks even if not followed by a space instances that contain periods have
25. at match NO search terms notboth Omits records having ALL search terms keeps all others merge Merges current and backup search REPLACING current search author Find records with ANY of keys in author field only allauth Find records with ALL of keys in author field only noauth Omits records with ANY of keys in author field only diff Find all records on backup but not on main search 2 8 3 Example Papers which cite refer to or are written by any Efron and which cite refer to or are written by any Stein gt search efron 138 records found 138 main records 7 backup records gt and stein 18 of 138 records meet criteria 18 main records 138 backup records Of these papers some are written by Efron and Stein ALLAUTH searches for records whose author fields contain all of the listed tokens gt allauth efron stein 1 of 18 records meet criteria 1 main records 18 backup records gt list 81AnlsStat 9 586 596 J The jackknife estimate of variance Efron B Stein C Analysis of variance Bootstrap U statistic 94x U statistic Of course some of the papers might have been written by Efron about Stein or his work after returning to the original search results using SWAP we have gt swap 18 main records 1 backup records 15 gt author Efron 12 of 18 records meet criteria 12 main records 18 backup records gt list 71JASA 66 807 815 J Limiting the risk of Bayes and empi
26. be quite costly with an on line system where you are paying for each search by the second or for each item retrieved In planning a search you may find it helpful to first use a text editor to browse the subject keyword or author indexes see Chapter 7 This will give an idea of the number of variants of a term as well as the relative frequency of each variant in the database If you are using contributed search software you should become fa miliar with the search strategies that it recommends Beware of hyphens versus spaces Hyphens and spaces are used quite inconsistently and often even in correctly in the titles of journal articles These inconsistencies are re flected in the Current Index to Statistics which records titles as they appear As a result you cannot predict for example whether the record you seek will have log linear model log linear model or even log linear model In such a case it is best to search for records which contain both log and linear If your search software uses an inverted index it can retrieve only the precise words that its authors have placed into the inverted index For example with such software you may need to retrieve the union of loglinear and log intersect linear Here is an example of such a search using the CIScd program search log linear search loglinear merge 35 Case sensitivity When the search program is capable of doing so it is usually best to do case sensitive sea
27. been repeated in Field 5 or 6 as USA UK USSR UN and WHO Before 1988 the printed volume of Current Index to Statistics did not use T X codes for umlauts but transliterated them in the customary fashion 38 6 Source Abbreviations The journal and publisher abbreviations used in Field 1 called keys below are those used in the annual printed Current Index to Statistics volumes Many keys are easy for statisticians to decode e g AnlsStat or JASA however others may be mysterious The database contains several files that give expanded forms of the keys These files are located in the ABBREV directory of the CD ROM and are described in detail in the CIS ED Technical Reference Manual Here is an overview of the files ABBRMST V97 This has the most complete information about each key However it is not well suited to human browsing because of the exten sive use of cross references to other keys ABBR V97 This is perhaps the most useful file It gives the full name asso ciated with each key e g The Annals of Statistics for AnlsStat The publishers of journals and edited books are identified as are the editors of most edited volumes ABBR V97 gives information similar to that in Parts 4 and 5 of each annual printed Current Index to Statistics volume but it is comprehensive for the entire span of CIS ED ABB V97 This file is similar to ABBR V97 but uses abbreviated forms of names e g Ann Statist for AnlsStat These
28. dexing truncated to six characters Thus the words regress re gression and regressor are all considered instances of the index word regres For purposes of indexing upper and lower case letters are considered equivalent Certain very common words such as of by and etc are also not indexed The author title keyword and alternative spelling fields of the records are indexed A search looks up one or more keywords in the index files and then uses the list of CIS ED entries found there to construct a new search result containing all of the entries that match the keyword criteria Once a search result is obtained the search can then be refined by adding additional constraints or by combining the current search result with one obtained earlier ClScd operates on a stack of search results in a fashion similar to calculators that use reverse Polish notation Each new search command creates a new search result which becomes the top most search result on the stack All previous search results are pushed down onto the stack maintaining the order in which they were created Refinement commands operate either on the most recent search top of the stack or the two most recent searches The contents of the stack after executing a command depend on the particular operation being carried out but generally the top most search result on the stack is the output from the last operation Always r
29. discussion appears immediately after the paper Look for notations beginning with Disc Com C R or Pkg Except for Pkg which denotes a package of related papers and discussion page numbers in these notations are not included in the span of pages for the main paper Some journals may publish comments often as letters to the editor several years after the original paper Often especially for more recently created records a parenthetical notation Disc Com C R after the title points ahead to such discussion However if you know that a par ticular journal has a policy of encouraging delayed discussion and if you have retrieved a paper by key words rather than by words in its title you may want to search by title for detached discussions The distinction between comments and a correction or addendum is authorship comments are contributed by persons other than the author s of the referent paper CIS ED uses the words Comments and Reply even if the journal calls these remarks Discussion and Rejoinder or gives them separate titles of their own 37 Accent marks Eliminate all accent marks in searches Similarly for transliterated names omit any apostrophes that is Silvestrov not Sil vestrov Fields 5 and 6 have been created to make this easy approach work well Exceptions For German umlauts it is best to use the spelling with ae oe or ue so that you will retrieve both pre 1988 and later records but be aware th
30. e used to browse the contents of the database to check for alternative spellings of words and to locate the occurrence of names included as parts of titles For instance the subject index can be used to identify occurrences of Stein James Stein and James Stein all of which potentially refer to the 1955 shrinkage estimator of Charles Stein Subject Index Format The format for the Subject Index is intended to make browsing with a word processor or screen oriented text editor relatively easy Because many words and phrases in titles and keywords contain TX codes the non mathematical entries in the Subject Index are sorted by a T X free version of the name the sort word obtained by stripping all T X codes and special characters from the word or phrase the actual word Records in the subject index have three fields count sort word and actual word The last field is empty if the actual word is the same as the sort word as is commonly the case The count field is of fixed length five characters the two remaining fields are of variable length and the fields are separated by the TAB character hexadecimal code 09 Each entry is terminated with a newline character The index should be searched in a case insensitive manner However for the sake of readability some words in the index appear in mixed case This is true of most acronyms for example ARMA SAS and EM and most 40 proper names for example Markov Fi
31. ed each time the command is issued If the last request issued was a next command either with or without a following integer an empty command line will be treated as equivalent to the next command Thus setting next 3 followed by an empty line will display the next three items in the current search Every time the Return key is pressed another three records will be displayed Any command other than next or Return will cause the Return key to revert to its usual function 2 16 Manipulating the search result stack The stack command gives a somewhat cryptic status report of the con tents of the stack of search results The push command and pop command move the stack contents by one position 2 17 Displaying the contents of a disk file type The type command will copy a named disk file to the screen 2 18 Displaying settings The current values of all options that can be set are displayed with the show command 2 19 Version The version command displays both the current version of the software and a brief description of the database being examined as told by the configuration file 2 20 Comment It is sometimes useful to put comments between commands Any com mand line beginning with is merely printed and ignored Blank lines are also ignored except immediately after a next command 2 21 Looking up an abbreviation whatis Not every abbreviation used in the CIS ED is self explanatory The whatis command will l
32. ee illustrative records taken from CIS87 V97 are shown below just as they appear in the database except for the line breaks needed to print them here 87Academic NY Lnd 442p B Probabilistic methods in mathematical physics It o K Ikeda N Ed Ito K PV15 87AnisStat 15 1155 1173 J Convergence rates for the bootstrapped product limit process Horv ath Lajos Yandell Brian S Lorenz curve Censoring Product limit estimator Horvath Lajos PV13 IMSC 87Ap1MaCmp 24 47 63 J Regression estimates of inputs to an M t G infty service system Patterson Richard L Queueing PV13 94m 4 1 Field 1 Location This field contains information on the year of publication source and location or size of the item indexed The order of records within each file is determined by a case insensitive alpha sort on this field That is all letters are temporarily changed to upper case before sorting The file C1IS87 V97 contains records showing a 1987 publication date Elements within Field 1 appear in fixed column locations Several different formats are used depending on the type of publication being referenced Field 1 is always 28 columns in length Columns 1 2 always contain a two digit year of publication and column 28 always contains an indication of the publication type using the codes in Table 1 If an abstract or other short description or summary is available for the record within CIS ED the code in column 28 is replaced by
33. emember Search first then refine 2 3 Tutorial Getting Started To start a search open or double click on the ClScd icon the one on your hard disk not the CD ROM DOS give the command CISCD Some thing like the following will appear on the screen The examples in this chapter use the 1996 Edition of the database Output using later versions of either the database or of ClScd will of course be slightly different Current Index to Statistics Extended Database Search Software Version 1 7 2 8 Mar 97 Ronald Thisted 94 95 file format Copyright 1995 7 Ronald Thisted Portions of Search Engine Copyright 1993 Douglas M Bates All rights reserved CIS Extended Data Base Publication or other use of more than 250 records requires permission of the CIS Type help for help gt To search for all papers by William Krasker enter search Krasker at the prompt This will produce gt search Krasker 16 records found 16 main records 0 backup records The dots are produced sequentially as the CD ROM is searched The time required for this query to find these 16 of the 150 000 records on a Power Macintosh was about one second An alternative would be to enter search William Krasker at the prompt In that case all records that contain both William and Krasker will be produced Note that the software automatically converts William to willia and Krasker to kraske before doing the sea
34. er roles are also used a full list can be obtained from the editors Role entries have been standardized so that the role abbreviations are now consistent across all records In addition the order in which roles are listed is standardized so that authors precede editors who precede reviewers Translators appear last The role Auth is used to indicate the author s of a reviewed work to distinguish them from the author s of the review itself The latter are indicated by the Rev notation For conciseness of presentation the role is included in the database entry only after the last person who plays each role For example a book 32 by Efron and Tibshirani reviewed by Jones and Smith would have the following entry in Field 2 Efron B Tibshirani R Auth Jones J Smith S Rev 4 4 Field 4 Key words and phrases These may be edited versions of key words published in the source or they may have been prepared by a CIS editor Beginning in 1997 Field 4 for some journals also contains authors key words which were not selected by the CIS editors These words and phrases are enclosed in square brackets and they follow the editors key words Some articles appear in a named department such as the Teacher s Corner for the American Statistician These department names may also be included in Field 4 if so they are contained in square brackets and they preced the editors key words The tendency but far from a f
35. erscore characters _ upper case letters are folded to lower case prior to index ing Words that don t begin with a letter e g numbers and dates are not indexed and neither are one character words and very common words and the of etc Words exceeding six characters in length are treated as if they consisted of the first six characters only Thus statistic statistics statistician and statistical would share the index en try statis Linking the separate index files together is a master index which has an entry for each word appearing in at least one index file These entries consist of the word followed by a coded list This list identifies each index file containing the word the number of times the word appears in that index file and the location of that word s entry in the index file If a word appears only once in an index file the master index records the position in the data file rather than the position of a list in the index file 7 5 2 Inverted Indexes in INDEX2 The second set of inverted indexes is created in the same way as for INDEX except that in the second set words exceeding six characters in length are not collapsed to their six character root Thus in INDEX2 statistic statistics statistician and statistical each have separate index entries By inspecting the keys in the master index you can easily tell which variants of a word you want to retrieve and which you do not The indexes in INDEX2 are used by
36. g the file with a text editor or word processing program Each entry is encoded in ASCIT and consists of two fields separated by a TAB character hexadec imal 09 The first field is exactly six characters in length and contains a count four characters a single blank space 1 character and the role code 1 character The second field consists of the stripped version of the name being indexed and is followed in parentheses by the TEX encoded version from which it came if appropriate Each entry is terminated by a newline character Examples 1 The following example illustrates several uses for the Author Index To investigate the works of Murray Aitkin one might start by perus ing the following section of the Author tabulation The examples in this section are from the 1996 Edition of the CIS ED Aitken M Aitken Murray Aitkin M Aitkin M Aitkin M Aitkin M A Aitkin Murray Aitkin Murray Aitkin Murray Aitkin Murray A Aitkin Murray A Aitkin Murry A P R00DOorerR gt se EE Fy D 33 RRE 43 It is immediately apparent that some entries which probably repre sent works by Aitkin are listed under Aitken or Murry It is not obvious whether this state of affairs is due to a precise transcription by CIS of an error appearing in the original journal due to a ty pographical error introduced by CIS or due to a second or third individual having a remarkably similar name to the perso
37. g the set command as in gt set refer All subsequent requests for displaying records will be in the newly requested format If you would like to display the current results in an alternative format without changing the default format add the format name at the end of the list request The following command will display the current results in long format for instance but will continue to use the default format for any subsequent requests gt list long 2 11 Saving to a file save If it is desired to save the results of a search on a disk the save command will prompt you for a file name and then ClScd will copy the contents of the main search results to the file with that name If the file name contains spaces or special characters or if it specifies a full path name then the file name should be enclosed in double quotation marks This makes it possible to save results to a floppy disk for instance Here are two examples gt save ridge regression biblio gt save a papers searches ridge cis The save file is written using the current default format You can specify a different format for saving the file by adding the the format name to the end of the save request as in this example gt save best cis refer 19 2 12 Loading a saved file retrieve It may be particularly useful to save the results of a long search for further refinement or perhaps for using as a fresh starting point later on To do
38. gram works 0 0 7 2 3 Tutorial Getting Started o 8 2 4 An Initial Refinement 9 2 5 Restricting the Search 9 2 6 A More Difficult Example 10 2 Commands ea a ee RRA A A De 13 2 8 Search Commands 00 14 2 9 Output formats 002000008 18 2 10 Setting output format 19 2 11 Saving to a file save o o 19 2 12 Loading a saved file retrieve 20 2 13 More more ira dr 20 2 14 Screen display list display 20 2 15 Showing the next k records next 20 2 16 Manipulating the search result stack 21 2 17 Displaying the contents of a disk file type 21 2 18 Displaying settings 21 2 19 Version LA A so ft he Pa 21 2 20 Comment P conan en eee eed 21 2 21 Looking up an abbreviation whatis 21 2 22 Using ClScd with a printer DOS WINDOWS only o o o a 2 23 Restarting or changing configurations 2 24 Quit quit co iaa BRAG Dele ee REO eS 2 25 Modifying the cis cfg when Files are on Hard Disk 2 20 Fine Print a goss oy ee ww aie a ie St ae ed ee III User Reference 3 Overview 3 1 Organization and Contents 3 2 Search Software 3 3 User Guide ill ee ira DP bee we ols 4 3 4 Technical Reference Manual
39. he previous section Windows 95 1 In the Windows Explorer drag the DOSSFT folder from the CDROM to your hard disk This will copy that folder and its contents to your hard disk 2 While still in Windows Explorer double click on the DOSSFT folder 3 While still in Windows Explorer double click on the INSTALL ap plication 4 When prompted give the drive letter that corresponds to your CD ROM drive This program creates a file called CIS CFG in the DOSSFT program on the hard disk This file tells the search program how to find the information on the CD ROM on your particular computer When the INSTALL application finished it will print All Done on the screen in an MS DOS window You will need to close the MS DOS window that the program opened yourself 5 Read the instructions for ClScd found in Chapter 2 6 To use the CIScd search program make sure that the CDROM is in the drive and then double click on the CIScd application in the DOSSFT folder on your hard disk Note the program will not run properly if you run the copy of CISCD EXE directly from the CD ROM itself When you quite the ClScd application you will need to close the MS DOS window that it opened manually 1 3 Hard Drive Installation on Desktop Computers If you have the space to spare on your hard disk about 40Mb access to the database can be speeded considerably by copying the contents of the CD ROM to your hard disk Consult the documenta
40. how current contents of main and backup searches config Initialize and restart with optional config file and Finds records in current search with ALL keys andor Finds records in current search with ANY keys not Finds records in current search matching NO keys notboth Omits only records that have ALL specified keys merge Merges current and backup search REPLACING current save Saves results of current search to named file retrieve Retrieves a named file of CIS database records display Show results of main search in current format quit Quit current program set Set options For more info type HELP SET show Show options For more info type HELP SET batch Read search commands from named file search Extract records from CISED using inverted index version Display version number of software and database more Set pager options 13 Comment author Find records with ANY of keys in author field only allauth Find records with ALL of keys in author field only noauth Omits records with ANY of keys in author field only type Display the contents of a disk file to the screen next Display the next k records Use NEXT k to set k diff Find all records on backup but not on main search whatis Looks up an abbreviation in the abbreviation file 2 8 Search Commands 2 8 1 The preliminary SEARCH Every CIS search will start with a broad search using the inverted in dex to the database The SEARCH command looks for all entries in the
41. ics 1973 74 Journal of Applied Probability 1964 74 Journal of the Royal Statistical Society Series A 1965 74 Journal of the Royal Statistical Society Series B 1965 74 Sankhya Series A 1965 74 Sankhya Series B 1965 74 Zeitschrift fiir Wahrscheinlichkeitstheorie 1962 75 Data entry for the The American Statistician was financially sup ported by ASA in preparation for a forthcoming index on diskette of that journal 3 all other data entry was done using CIS funds 52 An excellent index prepared by the Applied Probability Trust cover ing the years to be added to CIS ED greatly assisted in preparing records for Journal of Applied Probability and Advances in Ap plied Probability We thank Imke Janssen for her assistance with the translations of the many German titles in ZfW and G P H Styan for supplying back issues of the Canadian Journal of Statistics and for other assistance Coverage for the journals listed below was completed in 1995 Econometrika 1965 74 Theory of Probability and its Applications 1956 79 For the latter journal CIS has indexed the English translation How ever the Russian original was used to correct the large number of faulty transliterations of Russian names that appeared in the En glish edition 9 3 General Acknowledgments Almost all of the data entry for the Gap Project has been done by Bill Brent whose accuracy efficiency and patience have made a major contri bution to
42. imations for Pearson s chi squared test Best D J Rayner J C W Turnbull A P Goodness of fit 89Astrl1JSt 31 491 492 J Review of Goodness of fit statistics for discrete multivariate data Read Timothy R C Cressie Noel A C Auth Best D J Rev 91Biomtrcs 47 788 789 J Review of Smooth tests of goodness of fit Rayner J C W Best D J Auth Kemp A W Rev 93JRSS B 55 39 52 J Modelling complexity Applications of Gibbs sampling in medicine Disc p53 102 Gilks W R Clayton D G Spiegelhalter D J Best N G McNeil A 11 J Sharples L D Kirby A J gt gt Well that was positively unhelpful none were the targets gt of our search We can back out of the last result by gt interchanging SWAPping the main and backup files gt swap 64 main records 5 backup records gt gt The ALLAUTHOR command finds records that contain ALL of the gt listed search terms in the author field We can also request gt records that meet ANY of a set of criteria using the AUTHOR gt command Think of AUTHOR AUTH OR gt author Al Alvin Theodore Simon 8 of 64 records meet criteria 8 main records 64 backup records gt list 79ASAProStCp 153 153 P Discussion of LINMOD A flexible system for multivariate linear models computations Best Alvin M III 79Psymtrka 44 395 408 J On the precision of a Euclidean structure
43. ion License Access to the CIS ED must be limited to four designated stations either stand alone or in a local area network You may NOT make the CIS ED available over a network that serves more than four stations even if net work software limits the number of simultaneous users Remote access is not permitted Not available to national government agencies Personal License An individual is licensed as the sole user of CIS ED which may be in stalled on up to three computers under the licensee s control Users of the CIS ED should have easy access to this User Guide and should be encouraged to read it Except for Personal Licenses a copy of the applicable License Agreement must be available to any station that can access CIS ED In particular all users must be informed that indexes bibliographies and other works based on more than 250 records from CIS ED may not be published or distributed without written permission of the CIS Manage ment Committee and that published works prepared with the substantial assistance of CIS ED should acknowledge its use and cite the 1997 edition as a reference The following is a suggested citation Michael J Wichura Editor 1997 Current Index to Statistics Extended Database American Statistical Association Alexandria VA and Institute of Mathematical Statistics Hayward CA 55 Administrative records appear in many of the database files These have various uses which may differ from
44. items are in the subdirectories corresponding to the supported com puter platforms Macintosh MS DOS and Unix They are located in directories ABBREV LICENSE USRGUIDE CONTRIB INDEX and DOCS re spectively CIS ED consists of several flat databases The main database con sists of individual records separated by end of line characters There are separate versions for Macintosh MS DOS and Unix computers with the line separators appropriate to the respective computers Each record pertains to one indexed paper book database or computer program There are about 200 000 records altogether Each record is divided by eight characters into seven fields of variable length except for the first some of which may be empty Multiple entries or subfields within fields are separated by semicolons The contents of these fields are described in detail in Chapter 4 To use the database well it is important to understand how the information in each field is organized Other databases included in the distribution also with a flat struc ture are data files whose records contain information about e Abbreviations used for publishers and edited volumes e Abstracts of selected articles appearing in the main database e Indexes to the main database These flat files can in principle be linked using keys in common across databases so that they can be configured into a relational database Con tributed search software may free the
45. ixed rule has been to use singular and non possessive forms e g Chebyshev polynomial rather than Cheby shev s polynomials Words and phrases appearing in the title are not repeated in Field 4 4 5 Field 5 Searchable author information Since 1988 T X codes for accents on authors names have been included in Field 3 These codes should in most cases produce the appropriately accented version of the name for printing purposes Unfortunately authors names can be accented transliterated and even spelled in very different ways These variations in presentation make comprehensive searching difficult For this reason Field 5 contains the following kinds of author entries to make searches more likely to be suc cessful e Names with accent marks and other symbols removed These are called STRIPPED names and consist of the original name from Field 3 with all TEX codes and apostrophes removed For example In Field 3 In Field 5 Mu oz Munoz O Brien OBrien 33 e Names RESPELLED using standard roman transliteration For ex ample In Field 3 In Field 5 Chadeuf Chadoeuf Muller Mueller Bj rk Bjoerk ke Aake e Standardized TRANSLITERATION Many names originally spelled in Cyrillic or similar alphabets have been repeated in Field 5 with standardized transliterations For example In Field 3 In Field 5 Hasminski Khasminskii Has minsky Khasminskii Kha sminskiI Khasminskii See the CIS ED Technical Reference Manua
46. ks well A floppy disk will work but the diskette will be much slower and will be unable to accomodate the results of large searches In order to operate properly both the application itself CISCD EXE on DOS systems and the file CIS CFG must be present in the same folder directory on the hard drive The version of CIS CFG supplied on the dis tribution diskette will read data and indexes from the CD ROM On DOS machines an installation program must be run to create the configura tion file as described in section 1 of this manual Tf you search databases located on different drives you may have a separate configuration file for each setup and then switch between them as needed 2 2 How the program works To use the search program to its best advantage it is essential to learn how the program works The Current Index Database on CD ROM actually consists of a set of files containing the CIS entries themselves as well as a set of index files which are nearly as large These index files are tables of keywords together with the location of records in the CIS file that contain those keywords The keywords used to construct the index files on the CD ROM were all words that 1 begin with an alphabetic character 2 are followed by a numeral punctuation mark or blank space 3 contain at least two characters and 4 are no more than six characters in length Words in the database that exceed six characters are for purposes of in
47. l for more information e Suspected CORRECTIONS In a few instances where we suspect that an author s name may have been misspelled in Field 3 we have put a suggested correction into Field 5 usually followed by a question mark in parentheses It is important to note that the entries in Field 5 are provided to en hance searchability and are often not displayed by search software Most entries in Field 5 are automatically generated by algorithm to enhance both human and computer generated search terms 4 6 Field 6 Additional Searchable Information Key words phrases in addition to those in Field 4 are placed in Field 6 These include unaccented stripped versions of words and phrases from Field 4 noun phrases giving unaccented information from Field 2 title phrases with Americanized spelling of words or phrases that are likely to be the objects of searches etc 4 7 Field 7 Administrative codes Field 7 is used by the CIS ED editorial staff Search software ordinarily suppresses Field 7 34 5 Hints for Searching CIS ED Some general principles Consider starting with a search that will retrieve too much and finding ways to trim the result to what you want This sometimes requires less imagination and skill than retrieving too little at first and then trying to figure out how to expand the inadequate result Note that this principle can be implemented cheaply with a locally installed database such as this one but it might
48. m derived from more than 250 Database records contained in the Database may be distributed without the written permission of the authorized representative of the CIS ED Any publication prepared with the substantial assistance of the Database should acknowledge its use and cite it as a reference Michael J Wichura Editor 1997 Current Index to Statistics Extended Database American Statistical Associ ation Alexandria VA and Institute of Mathematical Statistics Hayward CA Certain programs files and documents on CD ROM distribution of the CIS ED are copyright by their individual authors Each such item contains a notice of copyright and may also contain other restrictions on redistribution described herein Information concerning current prices licensing and availability can be obtained from our page on the World Wide Web URL http www stat uchicago edu cis or by contacting either the ASA 1 703 684 1221 or IMS 1 510 783 8141 offices Contents I Quick Start 1 1 Quick Guide to Installation 1 1 1 All Installations o 1 1 2 Installing ClScd Search Software 2 1 3 Hard Drive Installation on Desktop Computers 4 1 4 Unix CD ROM Installation 2 2 4 1 5 Unix Hard Drive Installation 5 1 6 Technical Support 000 5 II Search Software 6 2 CIScd Version 1 7 2 6 2 1 Anstallation mg aii a kik Ae ho e e ee ee A 6 2 2 How the pro
49. m unixsft 1 6 Technical Support Additional information on special installations and corrections additions may be found at the CIS Web page http www stat uchicago edu cis Part II Search Software 2 CIScd Version 1 7 2 ClScd is a program for searching the Current Index to Statistics Extended Database CIS ED for entries that contain combinations of keywords specified by the user Versions of ClScd that run under MSDOS Windows and MacOS operating systems are included on the CD ROM This program is one of several that have been contributed by the CIS ED user community other programs which run under Unix and other operating systems can be found on the CD ROM or via the CIS web site ClScd is not fancy but it can be used to conduct quite intricate searches Although it does not currently have a menu driven interface much of its power can be realized by learning a handful of commands 2 1 Installation The program is distributed in both Macintosh and MSDOS formats The descriptions below are written for the Macintosh version except where DOS specific instructions are required in which case the DOS versions appear in square brackets The first step is to copy ClScd to a writable storage medium such as a hard disk There must be enough free space on this disk to accomodate the results of searches and intermediate results Except for the most extensive searches 1Mb of space will suffice and for most purposes 500 Kb wor
50. n in whom we are interested A careful scholar will investigate these entries by using standard CIS search methods to identify the records and then by checking with original sources for additional information The following excerpt from the index illustrates the treatment of names which have been transliterated from the Cyrillic alphabet using different transliteration methods 1 A Aleshkyavichene A 1 A Aleshkyavichene A Ale v stkevi v c iene A 1 A Aleshkyavichene A D 7 A Aleshkyavichene A K 1 A Aleshkyavichene A K ee EA K 2 A Aleshkyavichene A K K 2 A Aleshkyavichene A K Aleskeviciene K 3 A Aleshkyavichene A K Aleskjavicene i 1 A Aleshkyavichene A K Alyeshkyavichene A K 1 A Aleshkyavichene A K Alyeshkyavichenye A K Here the entry in parentheses is the actual form of the name as found in Field 3 The main entry is a canonical transliteration based on the American Mathematical Society system no TEX codes are used Since the canonical transliteration is also inserted into Field 5 a search for Aleshkyavichene A would recover all of the records indicated above The final example illustrates the manner in which T X ed names are converted to a common stripped key 1 A Gonzalez W Gonz alez W 1 A Gonzalez Wenceslao 1 A Gonzalez Andujar J L 1 A Gonzalez Andujar J L Gonz alez Andujar J L 2 A Gonzalez Farias Graciela 44 1 A Gonzale
51. of the early papers appearing in the journal availability of financial or editorial support from the publisher the lack of systematic coverage in other bibliographic sources and availability of the journal in libraries Abstracts book reviews news notices and officer reports are gen erally not included in this project Discussions of papers are sometimes indicated only by including the expression with discussion after the title of the main paper rather than by making a separate record for the contribution of each discussant in this case the pages given for the main paper include the discussion As a general practice records created under the Gap Project include the titles of non English papers both in English translation and in the original language inside parentheses Especially for papers published before 1965 key words may be present only for papers with particularly uninformative titles Unless otherwise indicated below key words were chosen by Bruce Trumbo Since the pub lication of author s key words and abstracts with papers was not common before 1975 many of these key words are based on introductions section headings and summaries of papers Some of the key words use terminol ogy that came into use after the paper was published Some of the Gap Project records included in this edition of CIS ED are still in draft form These have been entered carefully from a variety of 50 sources but have not yet been fully ve
52. older MACSFT contains versions of ClScd which are PPC or 68k specific These versions can simply replace the copy of ClScd in the folder on your hard disk MS DOS Installation 1 QO a FP WwW N Insert the CIS ED CD ROM into the CD ROM drive The drive letter assigned to the CD ROM differs from one computer to the next In this example we shall assume that the CD ROM is in drive F and that your hard disk is drive C Give the command mkdir C DOSSFT Give the command XCOPY F DOSSFT C DOSSFT S Give the command CD C DOSSFT Give the command INSTALL When prompted give the drive letters that correspond to your CD ROM drive This program will create a file called CIS CFG in the DOSSFT folder that tells the search program how to find the infor mation on the CD ROM in your computer Read the instructions for ClScd found in Chapter 2 To use the ClScd search program make certain that you are in the DOSSFT directory on your hard drive and give the command CIScd Note the program will not run properly if you run the copy of CISCD EXE directly from the CD ROM itself Windows Installation The easiest way to use the CIS ED on computer running Microsoft Win dows is to use the search program CISCD EXE from an MS DOS window and to access the database directly from the CD ROM Windows 3 1 From the File Manager open the icon MS DOS Prompt then follow the instructions given under MS DOS installation in t
53. olved with CIS since its inception several more people have made contributions of time energy and expertise to improve various aspects of the CIS ED and deserve acknowledgement and thanks Bruce Trumbo has been indefatigable in expanding coverage and he has also made major contributions to the planning and implementation of the many changes and improvements from the inception of the database to the present Michael Wichura has devoted many late nights and early mornings to the task of making the CIS ED a cohesive work of the high est quality and consistency Mike is responsible for literally thousands of changes corrections improvements and innovations Ed Gbur has made major improvements to the CIS abbreviation files and he has generously provided valuable advice to the current editors Bill Meeker wins first prize for bringing to our attention the largest number of typos inaccu racies and other infelicities in our early editions Correcting these has often led to identifying and systematically correcting large numbers of similar errors Finally Paul Shaman and the other members of the Man agement Committee have done an excellent job of making the publication responsive to the evolving needs of the statistics community 57 12 Errors Omissions and Corrections Thousands of errors have been found and corrected since the first edition of this database Please report any errors you find No error is too trivial to report To enable us
54. ook up the abbreviation in the abbreviation mas 21 ter list provided that ClScd has been configured with an abbreviation file gt whatis CmpBiomd Abbreviation CmpBiomd corresponds to Computer Methods and Programs in Biomedicine The command may be abbreviated to its first four letters 2 22 Using ClScd with a printer DOS WINDOWS only Portions of a search session can be printed to an attached printer using the built in capabilities of your computer To print a copy of the current page displayed on your screen press the CONTROL and PRINT SCREEN keys simultaneously If you would like to capture a transcript of your session or portions of it to the printer typing CONTROL SHIFT PRINT SCREEN will send a copy of each subsequent line typed or displayed on the screen to the printer This will continue until CONTROL SHIFT PRINT SCREEN is pressed again to disable the feature Many printers do not begin printing a physical page until the page s entire contents have been sent to it by the program Hewlett Packard inkjet printers are an example In this case pressing the reset button on the printer once you have finished sending it the material that you wish to print will initiate actual printing 2 23 Restarting or changing configurations Although this is rarely needed or desirable it is possible to start the program again from the beginning optionally reading a specified config uration file The commands in these two
55. owing journals beginning with the publication years given in parentheses e The American Statistician 1996 Journal of Agricultural Biological and Environmental Statistics 1996 e Journal of the American Statistical Association 1996 Journal of Business and Economic Statistics 1996 Journal of Computational and Graphical Statistics 1992 Journal of Statistics Education 1993 e Technometrics 1996 Book reviews from those journals in the list above that review books are also included with the corresponding entries in the database Entries in the data base for which abstracts or book reviews are avail able are denoted by a lower case letter j or e in column 28 of Field 1 The detailed file structure for abstracts is described in the CIS ED Technical Reference Manual 48 Editors and publishers of journals who would like to have abstracts of their articles included in CIS ED should contact Michael J Wichura for instructions on preparing submissions 49 Part IV Coverage 9 The Gap Project Bruce E Trumbo Associate Database Editor California State University Hayward Brian M Munneke Assistant Database Editor California State University Hayward and Purdue University 9 1 Editorial policy The Gap Project seeks to include in CIS ED information on scientific papers that appeared in selected statistics and probability journals before 1975 The selection criteria for this project include scholarly importance
56. rch in the index This search takes a bit longer about 4 seconds since every record that contains William must be checked to see if it also contains Krasker 2 4 An Initial Refinement Having done the search above we can refine the results using any of a col lection of powerful commands These commands can do arbitrary Boolean searches on terms or even on strings of consecutive characters The 16 main records found in the original search are stored in a temporary file on your disk The search refinement commands do not operate on the database as a whole but only on this subset of records contained in the main records file Each command that operates on the main records file leaves its result in the main records and the previous result in the backup records In fact ClScd creates a stack of search results This stack has two visible elements the main results and backup results file This stack can be manipulated with commands such as push pop and swap The file operations can be thought of commands given to a reverse Polish calculator The stack command gives a status report of all files in the stack The push and pop commands are used to manipulate the stack These commands tend to be quite fast For instance starting with the 16 Krasker records we can issue the request and William which produces gt and William 13 of 16 records meet criteria 13 main records 16 b
57. rches for author names and case insensitive searches for key words and title words When doing a case sensitive search note that the first word of a title is always capitalized in CIS ED Subsequent words are capitalized only if they are or are derived from proper nouns e g Monte Carlo The inverted indexes supplied with CIS ED are not case sensitive Greek letters and mathematical notation Do not attempt to use TEX codes in search terms Because the TEX code for a mathematical notation is seldom unique and because some journals spell out Greek letters phi mixing while others use symbolic notation mixing or y mixing the database contains in Field 6 a version of such words stripped of T X notation The file F6 V97 can be scanned to confirm the standardized version used in Field 6 Some examples are phi mixing n by n Latin square Cp statistic and 2 by 2 table Author s names The names of people are listed in Field 3 in inverted form with the family name s appearing before the given name s see Section 4 3 2 Thus a search for Stein C will retrieve Stein C and Stein Charles and Stein Charles M but not Stein Michael or Stein William or Steinberg etc Multi part family names and firm names Some names are particularly likely to cause difficulty Consider for exam ple Willem van der Velt Carlos Mendez Gonzales Park Dong Ho and SPSS Inc Some sources may have van der Velt but others may have vander Velt
58. ric characters or TEX accents e g M G 1 SAS GRAPH AR p and Scheff e e the language of publication if different from English enclosed in parentheses e g French and German and e a few phrases involving numbers e g 0 1 and 2 dimensional Some words in the index occur in two forms the way the word actually appears in the database for instance Scheff e or i i d and the version of the word used for purposes of sorting e g Scheffe or iid 7 2 Author Index The author index AUTHOR V97 approximately 2 5 Mb in length is de signed to fill a similar role to the subject index For the author index entries in the author field Field 3 are extracted and sorted by their stripped versions that is the name with all TX codes and special characters removed Each author is tabulated separately in each of sev eral possible roles encoded by a single character as shown in the table below This makes it possible to distinguish indexed items written by an individual such as articles books or book reviews from indexed items written by someone else about a work for which the individual was author or editor such as a reviewed book or computer program The role codes are given in the table below 42 Author of indexed item Author of reviewed work Discussant Editor Reviewer Translator Hyp gt II Author Index Format The format of the Author Index is intended to facilitate browsin
59. rical Bayes estimators Part I The Bayes case Ref V67 p130 139 Efron Bradley Morris Carl Stein estimator Shrinkage estimator These 12 references also contain the Efron Stein paper cited earlier To find the papers written by Efron and Morris regarding Stein shrinkage estimation we use gt allauth efron morris 9 of 12 records meet criteria 9 main records 11 backup records Of the original 18 papers with two pop s from this point how many dealt with the Efron Stein inequality gt pop 12 main records 18 backup records gt pop 18 main records 1 backup records gt count 18 main records 0 backup records gt and Efron Stein 4 of 18 records meet criteria 4 main records 18 backup records gt list 84IneqlSt amp Pr 112 114 An expansion for symmetric statistics and the Efron Stein inequality Richard A Vitale U statistic 86AnlsStat14 753 758 An Efron Stein inequality for nonsymmetric statistics J Michael Steele Symmetric function Variance bound 16 87AnlsStati5 1317 1320 An application of the Efron Stein inequality in density estimation Luc Devroye Kernel estimator 88StPrLet 7 105 112 A differential version of the Efron Stein inequality Bounding the variance of a function of an infinitely divisible variable Richard A Vitale Note that the SEARCH command would not accept Efron Stein as a search token or more accurately it would take the hyphen as a
60. rified by comparison with title pages of the original articles For many Gap Project journals CIS records in the period 1975 79 which were previously reclaimed from deteriorated media have been compared with original sources and brought up to current CIS standards Users interested in the details of the editorial status editorial assistants and copyright holders of records for articles published before 1980 may want to browse the administrative records that precede the bibliographic records for each volume of the journals involved 9 2 Contents The GAP records comprise over 2Mb of material it is estimated that the Gap Project has now captured about a quarter of the information that might have been included had the Current Index to Statistics been in existence in the period from 1965 74 In view of our guidelines for inclu sion we hope that this is roughly speaking the part of the bibliographic record from that period that will be the most important for current users Previous editions of CIS ED contained information on pre 1975 pub lications in the journals listed in order of inclusion in items 1 6 below Dates in italics correspond to the initial volume of a journal 1 As part of the preparation of Trumbo and Burdick 1 IMS provided financial support for data entry and editorial assistance in covering Annals of Mathematical Statistics 1960 72 Annals of Probability 1973 74 Annals of Statistics 1973 74 2 As part of the
61. rmat used in columns 17 25 27 Optional char acter Columns 1 27 are reserved Field 2 Title This field shows the title of the indexed item as it appeared in the source When non English titles are translated into English the name of the orig inal language is given in parentheses In some records full titles are given 30 in the original language in parentheses after the English translation Par enthetical comments may also show the location of discussions correction notes addenda or acknowledgements of priority Mathematical notation is represented in T X codes Knuth Donald E 1986 The TpXbook Addison Wesley Reading Mass 4 3 Field 3 Authors Authors names are given in the spelling shown in the source except to correct confirmed typographical errors in the source Multiple authors are separated from one another by semicolons Different formats are used for corporate or institutional authors and for individuals 4 3 1 Corporate Authors Corporate or institutional authors are uniquely identified by a leading blank space in their names In contrast to the syntax for individual names any commas in the names are used merely for proper punctuation Here are some examples Note that each name begins with a single space which we have emphasized with the symbol uAkademie Verlag uASA Subsection on Teaching Stat in Health Sciences U S Bureau of the Census uPanel on Discriminant Analysis Classification
62. rs will find contributed software in the Current Index Software folder Pointers to other sources of search software are contained in the directory OTHSFT All of the software for DOS Windows Macintosh and Unix com puters contained with the CIS ED has been contributed by users in the hope that some CIS ED users will find it helpful Users who develop or can recommend helpful software are encouraged to contact the editor of the database All contributed software freeware or shareware submitted to and approved by the Editor will be included in the 1998 CD ROM dis tribution Consult the CIS Web page http www stat uchicago edu cis for information for software developers including schedules for sub mission Disclaimer Any software distributed with CIS ED is provided as a service only The Current Index to Statistics does not support or recommend any search software The software supplied on the CD ROM is copyright by the individual authors As a condition of use users of the software must comply with any licensing or other restrictions imposed by the software s authors 3 3 User Guide The User Guide can be found on the CD ROM in the directory USRGUIDE There are three versions WT X file USRGUIDE DVI Adobe Portable 26 Document Format file USRGUIDE PDF and ASCII file USRGUIDE TXT The TEX and Adobe Portable Document Format versions are identi cal in content to the bound volume Adobe Portable Document Format
63. s book and software reviews prepared discussions of articles and correction notes Items such as notices and advertisements editorials let ters to the editor and meeting announcements are not knowingly indexed Subject to the above proviso CIS includes all articles from approx imately 100 core journals and selected articles from approximately 400 non core journals and other sources The list of core journals and the ex tent of their coverage in CIS ED and Volume 22 for 1997 is in the file CORE V97 in the ABBREV directory on the CD ROM The Tukey indexes to the statistical literature available from the American Mathematical Society both in printed form and through some of the MathSci on line services offered by AMS cover various important 56 statistical journals for differing periods of time mainly ending in the mid 1960s Because the Current Index to Statistics was not established until 1975 there was a gap of about ten years in the systematic bibliographic coverage of the statistical literature Since its inception an important goal for the CIS computer readable database has been to provide useful coverage for the period of this gap This effort has had the official backing of CIS ASA and IMS Chapter 9 details the current status of the Gap Project under the editorship of Bruce Trumbo and acknowledges the efforts of those who have made particular contributions to it 11 2 Acknowledgements In addition to everyone inv
64. sher and Egdeworth Occasion ally a word appears in both upper and lower case for example BLUE an acronym and blue presumably the color Here are some examples taken from the 1996 Edition of the CIS ED 18 R times C 18 r times c 380 t 124 t distribution 136 t test 16 t tests 2745 French 19777 analysis 8 analysis of variance 684 ARMA 15 ARMA p q 15 ARMA p q 58 BLUE 10 blue 1 BLUES 6 BLUEs 37 French 366 goodness 942 goodness of fit 1 goodness of fit tests 10 iid 2 iid i i d 73 iid i i d 142 M G 1 249 meta analysis 60 Scheffe Scheff e Generating the Subject Index The computer algorithm used to generate indexed words is too compli cated to describe in detail here Roughly speaking the process involves 1 extracting textual fragments containing TRX math mode material for example r times c 2 breaking the remaining text on blanks and semicolons 41 3 stripping most leading and trailing parentheses and quote marks including apostrophes and single quotes and 4 discarding some items not likely to be used in searches such as most numbers including page ranges and volume numbers As a result the subject index contains e all ordinary words e g analysis estimation and French e all hyphenated phrases e g goodness of fit chi squared and PSP e all textual fragments containing T X math mode e g t and M estimator e words with imbedded non alphanume
65. statistical literature published in the years 1975 1996 This edition of the CIS ED also includes many pre 1975 journal articles not originally con tained in the printed Current Index to Statistics volumes called Gap records below more complete coverage from the early years of the printed volume and some electronic abstracts 3 1 Organization and Contents Information from each year is organized into a single file For example the file for 1987 is called CIS87 V97 The items in the files for each year are the ones with that year as their nominal publication date even if some of the items did not appear in the printed CIS volumes until later years In this edition the 1996 files are shorter than the files for other recent years because some 1996 publications appeared too late for inclusion in this edition Files containing records from publications prior to 1965 are contained in a file named CIS0064 V97 In addition to the bibliographic files organized by years each copy of CIS ED contains e Source abbreviation files e Copies of the license agreement e A computer readable version of this User Guide e A computer readable version of the CIS ED Technical Reference Manual 24 Abstracts for some referenced articles Text of book reviews from ASA journals published after 1995 Contributed software and e Approximately 17 Mb of pre compiled index files for use with search software e Other documentation These
66. t to mix versions The ongoing effort to attach correction notes and comments to their refer ent papers creates a subject matter incompatibility with earlier editions Changes and additions are made to the source abbreviations each year Ongoing structural changes may make editions incompatible with each other Finally the inverted indexes distributed with the database will not work with mixed versions 1 2 Installing ClScd Search Software The steps to follow for installing search software depend on the computer system that you are using For desktop computers the easiest way to use the database is to install the search program ClScd on your hard drive and to use it to access the database directly from the CD ROM Instructions for installing this program on Macintosh MS DOS and Windows computers follow Once the software is installed refer to Chap ter 2 for full documentation Macintosh Installation 1 Drag the Current Index Software for Mac folder onto your hard drive This folder may be left permanently on the hard drive 2 Open the folder on your hard disk that you just created The program will not run if it is launched from the CD ROM 3 Read the instructions for ClScd found in Chapter 2 4 Double click on the ClScd icon and start searching The CIScd application in the Current Index Software for Mac folder is a fat binary which will work on computers running the MacOS on either PowerPC or 680x0 platforms The f
67. th Best invariant tt tests or Best linear So we refine the search by looking only in the author field gt author best 64 of 836 records meet criteria 64 main records 836 backup records gt list 71Biomtrcs 27 895 901 J Descriptive functions in disease Best William R Becktel Jack M Johnson Arthur F Regression Discriminant analysis Public health Multivariate analysis 73Biomtrka 60 429 430 J Extended tables for Kendall s tau Best D J Rank correlation coefficient Nonparametric regression Ties in ranks 74ApplStat 23 98 100 J Algorithm AS 71 The upper tail probabilities of Kendall s tau 10 Best D J Gipps P G Edgeworth approximation 74Biomtrka 61 385 386 J The variance of the inverse binomial estimator Best D J Proportion defective Inverse binomial Mean squared error Sequential sampling 74Technmcs 16 621 624 J MORE ynq q gt gt Although Best is in the author field many of them are not gt the individual we are after We can search for records gt whose author fields contain A and Best gt allauthor A Best 5 of 64 records meet criteria 5 main records 64 backup records gt list 83Biomtrka 70 447 453 J A test for comparing large sets of tau values Best D J Cameron M A Eagleson G K Kendall s tau Multiple comparisons Poisson limit theorem 83STATCOMP83 41 49P Power approx
68. the database The Current Index itself does not provide support for search software distributed with the database CIS ED and this User Guide are protected under U S Copyright Law Copyright 1997 all rights reserved Use of the database and User Guide is subject to a License Agreement with each site or individual member of ASA or IMS authorized to use them In 1997 there is a single License Agreement with several licensing levels one of which applies to your usage of CIS ED Below we give brief descriptions of the different levels but you should carefully read the version of the License Agreement supplied in the LICENSES directories on the CD ROM that applies to you because the informal language used here does not substitute for or modify the legal language of the License Agreement Fees to current licensees for subsequent versions are set considerably below the license fee charged to new users Users are encouraged to license each consecutive version Licenses of non consecutive versions are treated as new licenses Each annual version contains a complete updated and corrected copy of the latest edition of the database which must be sub stituted for the earlier edition General Site License We require that a site consist of a single contiguous geographic location unless otherwise agreed in writing Examples of a site might be a single of fice of an industrial division or government agency or a university campus CIS ED may be inst
69. tion for ClScd see also Part II Search Software for information about setting up the configuration files in this case 1 4 Unix CD ROM Installation To use the CD ROM directly mount the CD ROM typically using the mount or cdmount commands On some SUN systems it is critical to inform the system that the CD ROM is in High Sierra format hsfs and is not a Network File System nfs If the device is say cdrom then all files and documents are contained in the directory cdrom unix Database records are in the directory cdrom unix files index records are in the directory cdrom unix index and abstracts and book reviews are in the directory cdrom unix abstract 1 5 Unix Hard Drive Installation To use a hard disk rather than the CD ROM mount the CD and copy the cdrom unix directory to the hard disk Only the subdirectories of unix that will actually be used need to be copied from the CD ROM For instance if the search software being used does not employ the inverted indexes you may copy just the author subject and keyword index files author v97 subject v97 keywd v97 6 v97 from the index subdi rectory or you may wish to omit this directory entirely Be certain to configure any search software to look for the database in the new hard disk directory Several options are available for search software on Unix platforms Several contributed packages are contained and documented in the direc tory cdro
70. to correct any errors you discover please provide us with a printout of the defective record with the error clearly marked We will try to compare suggested corrections against the original source Where possible please send a photocopy of the title page or first and last pages of the item in question marked with publisher journal name date volume inclusive pages etc as appropriate to show how the CIS ED record should read Corrections can be faxed to the CIS ED Editor at 773 702 9810 by e mail to cised galton uchicago edu or by postal mail to Current Index to Statistics Department of Statistics The University of Chicago 5734 University Avenue Chicago IL 60637 USA Photocopies of source material are required for suggested additions to CIS ED The Current Index to Statistics does not index all articles with statistical content We welcome lists of possible omissions However they will be included in future editions only if they meet the current criteria for indexing and the necessary documentation is included These materials can be sent to Klaus Hinkelmann Current Index to Statistics Department of Statistics Virginia Tech Blacksburg VA 24061 0439 USA 58
71. versions of documentation require Adobe Acrobat version 3 0 or higher This free document reader can be obtained from Adobe s web site at http www adobe com The ASCII version is unpaginated it can be used to browse the User Guide on line say with a text editor The ASCII version is necessarily only a rough approximation to the bound volume which should be consulted to settle any ambiguities Any of these versions may be printed and duplicated 3 4 Technical Reference Manual Some information previously contained in the User Guide has been moved to a new CIS ED Technical Reference Manual This document can be downloaded from http www stat uchicago edu cis in Adobe Portable Document Format It is also provided on the CD ROM The Manual contains technical specifications and technical informa tion on the CD ROM format the detailed specifications of formats used for fields 1 and 6 of the database records specifications for the abbrevia tion format files specifications for abstracts IATFX styles for printing CIS abstracts and information about Russian and other names transliterated from Cyrillic characters 3 5 Changes in Future Editions Your suggestions for improving CIS ED are welcome Also the correction of errors is an important continuing effort in which user cooperation is essential Several important changes will be implemented in future editions of CIS ED We solicit comments from users about these items 3 5 1 Theses
72. word boundary and search for the token efron which we know produces more than 100 matches Similarly we could have refined the search in the following way note the use of quotation marks gt and Efron Stein inequality 4 of 18 records meet criteria 2 8 4 Another Example Here is an example of commands that find all indexed works by Paul Switzer written without a co author and excluding reviews gt search switzer 39 records found 39 main records 4 backup records gt author Switzer 33 of 39 records meet criteria 33 main records 39 backup records gt gt That gets rid of Switzerland for example gt noauth Rev 27 of 33 records meet criteria 27 main records 33 backup records gt Now the reviews written by Switzer are gone gt noauth 14 of 27 records meet criteria 14 main records 27 backup records gt That deleted all multi authored works 17 gt author Switzer P 14 of 14 records meet criteria 14 main records 14 backup records gt The last request verifies that the author field of all of gt the selected records contain both Switzer as a last name gt with a given name that starts with P such as P or Paul gt pop 14 main records 27 backup records gt This command pops back up one step gt diff 13 records meet criteria 13 main records 14 backup records gt These
73. z Garcia C Gonz alez Garc i a C 1 A Gonzalez Guzman Jorge Gonz alez Guzm an Jorge 1 A Gonzalez Manteiga W 5 A Gonzalez Manteiga W Gonz alez Manteiga W 1 A Gonzalez Manteiga Wenceslao Here the parenthesized version denotes the author s name as found in Field 3 while the main entry indicates the stripped form placed into Field 5 By visually scanning the Author Index in the neigh borhood of an author of interest variants of the author s name can sometimes be found In this case it is likely that the five entries marked with an asterisk above were all written by the same author but no single search strategy would easily find both forms of the name Forearmed with this information an investigator can search in the CIS ED under both variants 7 3 Index of Key Words and Phrases The keyword index KEYWD V97 approximately 0 5 Mb in length is a tabulation of the semicolon separated key words phrases in Field 4 of the database The two fields count and term are separated by a blank the count is right justified in a field of width 4 The count does not include occurrences of the term in the title field Field 2 of database records For example the following entries from KEYWD V97 in the 1996 Edition 2 James Stein 2 James Stein estimation 58 James Stein estimator 1 James Stein predictor 1 Truncated James Stein estimator show that there are 64 database records with the expression

User Guide CIS EXTENDED DATABASE

Contents

Download Pdf Manuals

Related Search

Related Contents