DOCUMENT RESUME ED 189 999 HE 012 952 AUTHOR Strenglein, Denise TITLE Choosing a Computer Language for Institutional Research. The AIR Professional File No. 6. INSTITUTION Association for Institutional Research. PUB DATE BO NOTE 5p. AVAILABLE FROM Association for Institutional Research, 314 Stone Building, Florida State University, Tallahassee, FL 32306 JOURNAL CIT AIR Professional File; n6 Sum 1980 EDRS PRICE MF01/PC01 Plus Postage. DESCRIPTORS *Computer Oriented Programs: Higher Education: *Information Retrieval: Information Storage: Information Systems: Information Utilization: Institutional Characteristics: *Institutional Research: Management Systems: Programing; *Programing Languages; *Record Keeping; *Statistical Analysis ABSTRACT It is suggested that much thought should begiven to choosing an ,,,loropriate computer language for an institutional research offi.-e, considering the sophistication of thestaff, types of planned application, size and type cf computer, andavailability of central programming support in theinstitution. For offices that prepare straight reports andinferential statistics a statistical language that provides report features is recommended:straight report language is suggested for offices notdoing inferential statistics. Offices doing their own programming, in part,should keep in mind that:(1) programs producing reports should be clearly documented in English: (2) there should be no more than twolanguages used in an office, and no more than one person shouldknow each language; (3) new staff members should learn the languagesused, rather than introduce a new language the new staffer knows:and (4) existing official data files should be used whenever possiblerather than creating separate new ones. (MSE) *********************************************************************** Reproductions supplied by EDRS are the best that can be made from the original document. *********************************************************************** The AIR Professional File Summer 1980 I XThe Association for ingiMjimat RPsearch No. 6 EDUCATION &WELFARE NATIONAL INSTITUTr OF EDUCATION CHOOSING A COMPUTER LANGUAGETHiS DOCuMENT HAS BEEN REPRO- Denise Strenglein DUCED EXAC To, AS RECEIVED FROM Data Base Coordinator FOR INSTITUTIONAL RESEARCH THE PERSON OR oRGANtiATION ORIGIN- ATING IT POiNTS OF vIEW OR OPINIONSUniversity of South Florida STATED 00 NOT NECESSARtLY REPRE- SENT OFF iC[AL NATIONAL INSTITUTEOF EDUEATO,ns POSITION OR POuCY A strongly felt need for a multipurpose and simple-to-use success. An efficient language will do both sets of applications. computer language emerges as more and more institutional Both types of application require file handling, and the research offices acquire computer termina! ;. A veritable Babel file-handling ability of the language chosen is of utmost impor- -4 of computer languages exists (Ryland, 1979), and a good deal of tance Institutional research offices use data from many sources thought should be given *o choosing the right language. Learn- and rarely can control file formats. While the easiest files to use ing a computer language represents a significant investment in arc fixed length sequential, mom often than not the nature of time and effort, and because of the problem of training new staff institutional research projects requires reading variable-length, when turnover occurs, only one or, at most, two languages COBOL-generated files or matching information from a number should be introduced into an office. of different files. It is desirable to be abk to euract small sets of The state of the art in languages has progressed far enough information from large files in order to reduce run time and cost. that, in most instances, institutional researchers need not be- There are some automatic features which simplify report come full-tledged computer programmers in order to access writing: opening and closing of files, reading and writing of computer data. Many of the high-level languages have taken the records, movement of data from input to output, initializing drudgery out of programming, and having the computer do much variables, accumulating totals, sorting, and report formatting of the data manipulation takes the drudgery out of institutional with overrides available. Clear and understandable selection research. logic and the ability to manipulate character data as easily as The choice of the best computer language depends on the numeric data are also important. sophistication of the staff, the types of planned application, the It is possible to get usable, if not pretty, results almost size and type of computer, and the availability of central pro- immediately with automatic features, providing both a move gramming support. rapid return on invested time and an incentive to continue learning. Automatic features, however, also tend to remove Training Coasiderations control from the programmer, so overrides are necessary if The characteristics that make learning and using a language anything complex or unusual is to be produced. easier for the institutional researcher who has little computer Descriptive statistics such as percentages, frequency distri- background are these: syntax that looks like ordinary language; a butions, and means are, by far. the Inost common applications of well-written manual; availability of training programs; a lan- statistical analysis; some of the nonstatistical packages offer guage that is somewhat forgiving (that is, one that does not come these features. The next most frequently used statistics are regression and analysis of variance; one of the statistical pack- to a dead stop for every little infraction of the rules); a minimum ages will have to be selected for these uses. It should be noted of required coding; clear, understandable error messages; and job contoal language grouped at one end or the other of a that some of the statistical packages are beginning to offer both multistep job. On-campus assistance from someone who knows statistics and report features. the language is essential for the neophyte, especially in learning the system-specific job control language for the institution's Size and 'IYpe of Computer installation; a good place to start is with an undergraduate While some of the more common languages---for example, computer course in any language or a beginner's program Fortran and COBOLare machine independent, being avail- offered by the computer center. Certain fundamental concepts able on almost any brand of computer, otherslike IBM's are common to all languages. These include file itructure and PL/1have been developed for only one type. IBM and IBM- input mediums, flow charting or other planning techniques, type machines such as Amdahl. ce'n Itel, CDC. Magnuson, language coding, and job contml language. Once these are and Ryad models, comprise much o computer market. For learned, it is possible to gain a useful amount of skill in almost this reason, languages designed for BM computers tend to any of the report or statistical languages with a manual, rela- dominate. It should be noted that even the nominally stan- tively unimpeded access to a computer, and a lot of patience. Of dardized languages, COBOL among them, are highly machine course, a training program helps, and access to someone who dependent since each manufacturer tends to introduce features knows the language is even better. which take advantage of that particular computer's unique capabilities. Applications Size is also a factor. The amount of computer memory The types of projects done by the institutional irsearch available sets limits on the capabilities of the language and on the office should be considered next. There are two basic classes of size of the data sets and number of variables a given job can computer applications in institutional research. The first and handle. Some packagesSPSS and MarkIV, for example most common is the sort-accumulate-list type of repot. This come in versions to fit progressively larger computers. Other includes such things as count of students by race, by sex, by languages, such as Basic, are designed to operate best on small classification; aggregation of student credit hours by discipline, computers. As distributed data processing using microcomput- by level; cost figtqes by department; and government equal ers becomes more prevalent, available memory becomes more opportainity reports, among many others...The second type of of a limiting factor, at least for modes of operation which are application involves descriptive and inferential statistical independent of a large central mainframe. analysis at vatying levels of sophistication. This includes, for Normally, an institutional research office will simply adopt example, enrollment projection, ahalysis of questionnaires, sal- a language that is already available locally. However, if cir- ary regressiza analysis, and analysis of predictors of student cumstances arise in which an institutional research office "goes shopping" for a report or statistical language, it would be wise to par excellance. It is, however, very poor at handling character szt advice from someone who is familiar with hardware. data; under normal operation, the maximum number of letters a character variable can contain is four. Fortran, like COBOL, has Availability of Central Computing Support no automatic functions but does permit a great deal of flexibility and control. It is symbol oriented rather than language oriented, The motivation for acquiring an office terminal, in many as expected of a scientific