lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) A MODEL FOR USER'S SERVICE: PROVIDING FOR INFORMATION AND DATA RETRIEVAL FROM AN ARCHIVAL, USER, AND DEVELOPMENT LIBRARY Harriet A. Dhanak Christopher D. Brown Michigan State University Although the Newsletter has carred several articles dealing with the various operations of specific archives (The Roper Center, NORC, etc.), the organizations reviewed were establishments devoted primarily to pro- viding archival and library services. Many of the members of lASSIST, in contrast, work alone or with limited assistance, within the context of an academic (or other) department. The article which follows was first pre- sented at the 1979 lASSIST Annual meeting in Ottawa and is a description of the way in which a depar tmentally organized archive is confronting its work . --Ed i tor . INTRODUCTION characteristics of the user commu- nity and the Michigan State Univer- This paper will deal with the sity computing system plus our problems of information storage and tasks and needs and those of our retrieval associated with a machine user community, readable data archive located within a teaching/research depart- ment, namely the Political Science Department of Michigan State Univ- LOCAL SITUATION ersity. A model, in the process of development and implementation, is Because the Political Data described that hopefully will Archive is located within a teach- alleviate the problems we have ing and research department, our encountered. The mode of delivery contact is with researchers, of copies of machine readable data faculty and students, during the files will be discussed as well as period that their work is in pro- modes of user information storage gress. Through the Pol itometr ics and retrieval. Laboratory also located within the department, we will either conduct Some subjects germane to the the computer runs or assist the topic, such as cataloging machine users in all aspects of the compu- readable data files in a tradi- ter applications. Therefore, we tional library, have been dealt find it necessary to address their with extensively elsewhere and will needs as well as needs of the not be discussed here. While there archive, are generic problems facing the archives that deal with machine The present situation on our readable data files, this paper campus is: will focus on the ones that are central to our users and functions of the archive. Points to be developed and discussed are the - 30 - lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) 1. We are the only archive of machine readable data files of cross discipli- nary material on the cam- pus . 2. The main library is not in the immediate future planning to hold our types of files. At pre- sent the code books are not located in the main library, but are in a library located within the Political Science Department. 3. The archive also serves the College of Social Science and the univer- sity community as the supplier of data sets from the Inter-University Consortium for Political and Social Research. 4. The Michigan State Univ- ersity Computer Labora- tory maintains a Control Data Corporation 6500, with an Inter Data that handles the terminal I/O. A CDC 6400 is available on a limited basis for batch work. A Hewlett Packard 2000 is main- tained primarily for instructional purposes. 5. The Computer Laboratory offers consultation on system problems, computer language (Fortran, Cobol, etc.) execution problems and for analysis packages such as Statistical Pack- age for Social Sciences and the MSU STAT package. 6. An interactive multipur- pose computing package is not maintained by the Computer Laboratory. Within the department, we need the use of special analytic programs that must be updated with com- puter system changes and modified for research needs. RESEARCH PROBLEMS One problem both for researchers and others using a computer system for information manipulation that differs from a traditional library use is that the functioning unit or system is under constant change, that is, the enviornment within which one works is constantly changing. The computer system itself may be changing with a new model or even worse a new vendor. The existing computer system is altered, hopefully upgraded, and is not necessarily upward compatible. New programs/methods are introduced on an existig system, and the changes may prevent previously exe- cutable program from running. This situation presents compu- ter-oriented researchers with an important cost in personal and pro- fessional time just to keep up with the changes. This time is above the time consumed learning to use computers as a research tool. For those who constantly use a computer the information is easily recalled and current. However, most researchers will use computers epi- sodically and the development of our model is partially dictated by the characteristics and needs of this type of user. Those using a computer must familiarize them- selves with some aspects of their use, at least analysis packages such as SPSS. We would hope to assist such users by enabling them to obtain a data set with a minimal knowledge of tape handling proce- 31 - lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) dures, assuming that the data set aspect that we will focus on is his is on tape. suggestion of a machine readable index of holdings. He details the SSDC method of keeping tape file information on a card image file. ARCHIVE PROBLEMS We have already used that method, and found that it was not adequate Several authors have addressed for our needs. The card image file some of the issues of special could be augmented when new files interest to us, with most noting were created for a study, but basi- common problems. White (1974) dis- cally it was laborious and could be cusses and cites literature that inaccurate, compares the varyng functions between libraries and archives. All archives, large or small. The discussions deal with acquiring with all our variations, are facing libraries and those concomitant a similar problem: The basic prob- problems, not with data management lem is the storage and retrieval of problems associated with research various levels and types of infor- or secondary analysis. He cites mation. We have probably all gone the difficulties facing researchers through the same process of first in gaining information on the producing a hard copy list of hold- machine readable data files availa- ings -- typed at first, then later ble. placed in machine-readable form. Many archival holdings and most The implementation of our system user data sets were on cards and will expand the information on were laboriously carried around available data-sets, and will also with notes and information written archive user library services to on the cards or across the top of the university community. Our aim the deck. When data sets were/are is to break the pattern of informa- stored on tape or disk, a user can tion access noted by White that one no longer visually inspect the data needs personal contacts to gain and an uneasy feeling sets in. knowledge of data set availability. This is the period that technologi- cal change in data storage out- Ferguson (1977) notes the types stripped the means used to document of questions data file users ask of files and studies, library and computing organiza- tions. Their interest is not in This detailing of developmental partitioned responsibility between problems in archiving does not even the library, computer center and begin to confront the enormous task archive, but in the availability, of indexing variables across stu- access and documentation of data dies. Since we do not have the files. resources to deal with these prob- lems, we will move on to those that Grandon (1978) discusses archive are presently manageable. We hold development stages in general and about two hundred and fifty stu- then details the system under dies, and have a library of about development at the Social Science three hundred tapes that probably Data Center/Roper Center. His contain about two thousand files, paper describes functions in which We also have the usual need for the all machine readable data file dissemination of information on the archives must participate to a studies available and have a pro- greater lesser extent. The one gram that lists them as card images - 32 - lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) by subject and author within subject. Archivists and research- ers are faced with the problems of increased complexity of machine readable data files, and with mul- tiple files associated with one study title. A survey might consist of one rectangular i zed file of one to two thousand card images, or may con- tain tens of thousands. The latter presents only problems of bulk. A survey may be updated (new edi- tions) by cleaning or for other reasons. Panel studies present additional difficulties by having the problems already noted as well as the addition of new waves. To this list of complexities and prob- lems is difficulty of hierarchical files. At present our biggest problem is to store, retrieve, and have available information on tape files and a machine retrievable document for each study with a description of each file. Turtle (1978) also notes that file documentation was deficient in the library at the Tennessee Valley Authority and pro- posed procedures to improve the existing documentation system. We differ from her in that she states "a computer tape library ... con- tains only one physical format for information". A serious problem for our archive and users is that we may have multiple formats for files of one study. These may include forms usable by a batch-or- iented analytic package while the file may also be stored diffecently for on-line terminal use. file tha dif f icul t by Carte that is e ence Quan Universit levels of tation pi cate name as each e give our t may be updated is not and a method is suggested r and Roistacker (1976) nforced by the Social Sci- titative Laboratory at the y of Illinois. However, information and documen- us a prohibition of dupli- (a search is conducted ntry point for dups) will users more flexibility. MODEL The model that we are develop- ing : 1. assumes minimal computer- associated knowledge by users. 2. will be a method for information storage and retrieval of documenta- tion and/or numeric data. 3. will be a process that will allow modification of the stored indexes* and the programs execut- ing number two above without having to redo the stored sets of archi- val information. This model can be viewed as a multipurpose system that will serve users who: 1. are browsing for informa- tion by author, title subject (general) and Clerical procedures, for tracing files and problems with tape, (such as parity errors or the need for cleaning), require an inordinate amount of time and become insuffi- cient and redundant. File documen- tation for studies with only one *This refers to archi- val documentation records as opposed to codebooks and other hard copy docu- mentation describing a specific data set. 33 - lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) geography; 2. are searching for a spe- cific author, subject, or geographical area; 3. may want to retrieve a file that they specify f r om t a pe or disc; 4. have the need to create their own library; 5. wish to hold information and files in the develop- mental section of the library for creating tem- porary files and the documentation associated with them. The system will enable archival maintenance and updating of all material, and logging and account- ing records of all activities. After trying approaches previous became obvious that prehensive approach ment of information their associated fi enable us to actua procedures. One n ency of only doing multiple times, and dent help one needs tern that does no explanatory time time. some of the ly mentioned , i t we need a com- to the manage- on studies and les, that would 1 ly simpl ify our eeds the effici- a job once, not when using stu- an ongoing sys- t require more than executing Frequently not enough time is spent in designing a system, which leads to implementation problems. We have spent much time, thought, and planning for a model that will serve well in the context within which it is anticipated it will operate. The development or use of a sys- tem such as RIQS (1975) with tuto- rials would be very helpful to new users, but at the present time it is beyond the scope of the system we will be implementing. A CAI is under development at our institu- tion, and will be used if suitable to our application. We plan to develop a simple querying system, not a sophisticated one. Our interest in a developmental library and in allowing users to create their own libraries comes from working with researchers and classes and attempting to keep track of all the tape and file activity. As with many aspects of research work, documentation of work in progress is one of our big- gest problems. Although the prob- lem of classifying, cataloging and documenting studies has been dealt with in the appropriate literature, there is not, to my knowledge, any (or much) literature on systems designed to deal with the informa- tion problems associated with developmental file systems for users in an academic setting. Those who have large grants may be able to hire their own staff to keep records, but most university researchers must keep their own records of their work as it pro- gresses to completion through vari- ous files. At a later period the information on files could remain on a permanant record and accessi- ble to them or it could be scat- tered . A user crea 1 ibrary wi 11 be g ity of entering a documentation tha the system and wi of limiting or access to any o information. Fe noted the relucta to enter informat dies into libr records. Our s them the privacy ting thei iven the fl 11 informat t is stand 11 have the pe rmi tting r all 1 ev rguson (197 nee of rese ion on the ary or ystem will they requir r own exibil- ion and ard for cho ice public els of 7) has archers ir stu- archive allow e and a - 34 - lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) method of documentation they need and a history of their studies pre- viously entered . The information to construct the indexes of information resides on several files. The files contain information in a hierarchical structure with imbedded fields to link them together. This allows us flexibility both with the files (used as data) and programming, that is, one or both can be modi- fied independently. By using the hierarchical file structure of the holding files, we plan to include a list of studies availabe to us, but not located in the archive. This will enable the university community to know all the ICPSR studies listed in the User's Guide and could be easily updated when we receive periodic notice of new holdings. If we obtain a membership in the Roper Center, that information would also be entered . At the on-line querying stage any user will be given the informa- tion needed to submit a batch job to obtain a copy of the required data. Tape numbers and their necessary information (tracks, den- sity, character mode, etc) and file position will not be needed by the user to request a file. File loca- tion, whether disc or tape, will be retained in the index and the file retrieved with a simple command followed by pertinent information. For various policy reasons, the computer laboratory at MSU will not allow a request for a tape mount to be executed when one is on-line at a terminal. Although it is limit- ing when one needs a small file, it is not as restrictive as it appears because when one is working with a large data set, batch mode is gen- erally preferred due to both time and cost factors. Users superv i sio mitted for of a tape a po o 1 of ing their indexing i the job rel ieving entering m index at time (1979 informatio cal manner author izat file. runni n coul them file, tapes anal nfo rma is e them inimal a late ) we n 1 ink , plus ion f i ng und d have to ret We wi ava ilab ysis da tion e xecut io of t inform r date have f ed in a sub j le and er archive a job sub- rieve a copy 11 also have le for stor- ta and the ntered while n, thereby he problem ation to the At this our files of a hierarchi- ect file, an a documents The holdings file contains information on the type of system (Archive, User Library, Develop- ment) . A twenty character acronym will be the index reference for a study. Authors, title, subjects, geography, and date are information that will be used for searching and printout. Read and alter passwords can be set at this level. The mapping file contains infor- mation on the version (edition, set or subset) , and as many as neces- sary can be established for a study. This will cover multiple files, updates, etc. This level will also offer a restriction code if needed. Date, document, and other user information will be held here . Th on t binar etc) . numbe tion file) There versi users wi th archi forma Other archi e access fil he form of y, card deck The file 1 r and file of reel in , and num will be as on as is ne will be r permi ssion) ve files tha t that mate form inform val informat e has a fi , book, ocat ion po s i t i o a mul ber of many cessary estr ict to ace t are h the ation ion onl inform le (c SPSS (tape n and t ipl e acce fo rms Ge ed (e essi ng in sta codeb will b y ation oded , file, reel posi- reel sses . of a neral xcept only ndard ooks . e f o r - 35 - lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) The tape file will contain the Center, and main library choose to standard tape information, e.g. build an infrastructure similar to tracks, density, label, mode, those units found at Stanford, Wi s- owner, plus a tape history of consin and Northwestern among oth- accesses, cleaning date, and number ers, we believe that the existing of mounts. files of information in combination with the executing program struc- The authorization file will con- ture will serve as a base for tain access level, accounting expanded holdings and services, information, logging information and scheduling action if necessary. One is left with a frustrating This file will provide information feeling that there are developments for report generation. The sche- occuring elsewhere that would be of duling algorithms will be particu- use to us and coversely, others larly helpful for instructional might use our efforts to their files. benefit if we could surmount the major problem of program transport- The document files contain the ability to other vendor systems, standard information on each study: We seem to be "reinventing the number of cases and variables, sam- wheel" at many institutions, pling design, abstract, issuing archive, published material, etc. References We are still formulating a design for a version document section. We Carter, M. C. and Roistacher, R. C. plan to have documentation accessa- Statistical and Data Support to ble in several modes and are still a Heterogenous user Community, designing this phase. First International S. A. S. Users Meeting, 1976. Those studies under archive entry and control will have all of Ferguson, D. Social Science Data the specified information entered. Files, The Research Library and To store a file a user may enter the Computing Center. Drexel all of the above information for a Library Quarterly 1 3 (1977) , permenant file, or enter only the 70-79. acronym, version and form for tem- porary files. In this case default Grandon, G. M. An Archive Develop- information will be entered by the ment System: Specification and archive system. Progress. 1978 lASSIST Annual Conference. ittman, B. and Borman, L. Person- CONCLUSION alized Data Base Systems. RIQS—Remote Information Query This paper has described an System (Los Angeles: Melville information system still under Publ ishing Co., 1975), Chapter development that will solve some of 2. our problems of maintaining a machine readable data file archive Turtle, M. and Smart, C. W. The in an academic setting. Much plan- Role of the TVA Libraries in ning and programmming effort has Digital Data File Documenta- been expended to date on this rela- tion: A Demonstration Project, tively small problem. If the Col- 1978 lASSIST Annual Conference, lege of Social Science, Computer - 36 - lASSIST Newsletter, Vol. 3, No. 2 (Spring 1979) White, H. Sets : Ph .D. Dissertation D. Social Science Data California: University of A Study for Librarians . California, 1974), Chapters I, (Berkeley, II, and III. 37