Table Of Content

EDITORIAL BOARD Editoy and Chair of the Editorial Board. . . . . . IENNTFER YouNGER EditorialAssistant. .... DevrpH. THoMAS Assistant Editors: CsnrsrraNB orssoNNas.. .....fbrAcquisitions Section Canr-rt Ruscnonr .., ., fbr Cataloging and Classification Section Jlur Tn'naowell for Collection Management and Development Section LEEDIRKS. ......lbrPreservationandRelbrmattingSection SYLVIAO . MenrrN {br Serials Section PRUDENcEW . DALRyMpLE Special Editor MARcARETR oHDy . Book R-eviewE ditor Er-Offi.ci.o Members: Er,aINs YoNTz, Delegate of the Chair, Council of Regional Groups KaneN MuLLsn, Executive Director, ALCTS DALE SwENSENE. ditor. ALCTS Neuslefter Hesburgh Library Universityo{'Notre Dame, Notre Dame, IN 46556-5629;( 219) 631-7790;f ax: (zlg)631-6772:e-mail:[email protected]. Ad.oertising:BillCollee,c/oBenson,Co{I'ee&Asso- ciates, 1411 Peterson Ave., Park Ridge, IL 60068; (847) 692-4695; tax: (847) 692-3877. ALA Production Sen:ices: Troy D. Linker, Christine S. Taylor; Tina V Bryson, Angela Hanshaw, Kevin Heubusch, Tracy Malecki, and Donavan Vicha. Memiers: Addressc hanges an"di nquiries should be sent to Membership Department-Libranl Resources{ t Techni.calS eroices,S0 E Huron St., Chi- c-ago,I L 60611. Nonmember subscribers: Subscriptions, orders, changes ofaddress, and inquiries should be sentto Library Resources b Technical Sensices,S & S Computer Services, Inc., 434 W Downer. Aurora. IL 60506. Library Resources C: Technical Senfices isindexedinLibrary Literature, Library b lnformation Science Abstracts, Current Indcrto lountals in Erlucation, Science Citation Indzx, antflnforma- tion Science Abstracts. Contents are listed in CALL (Current American-Librant Literature). Its reviews are included in Book Reaieto Digest, Book Reoieu Inder, and Reoieu 6f Reoiews. Instructions lbr authors appear on p. 72-73 ol the January 1998 issue and on dre LR?s web page at www.ala.org/alctVlrts. CopGs of boo^kslb r review should be addressed to Margaret Rohcly, Book iteview Edttor, Library Resarrces Lt Technical Sercices,YanPelt-Dietrich Library Cinter, University of Penn- sylvaniaL ibraries, 3420 walnut st., Pldladelphia, PA 19104-6206( e-maii: [email protected]). @ American Library Association lggg All materials -in this journal subject to copy'ight by the American Library A.s.sociationm ay be photocopied fbr-the nonc_ommercialp urpo.seo f scientilic or educational advancement granted by sections 107 and 108 ofthe copyright Revision Act of 1976. For other reprinting, phoiocopying, or translating, address requests to the ALA oflice of Rights and Permiisiotrr, 50-E. Huron st., Chicago, IL 60611. The-paper used in this publication meets the minimum requirements of American National Stan- dard lbr Inlbrmation sciences-Permanence of Paperlbr Printed Library Materials, ANSI 239.48-1992. - Publication in Libranl Resnurces l: Techni.cal Seroices does not imply official endorsement by the Association for Library Collections & Technical Services nor by iLA, and the assumption of editorial respon-sibility is not to be construed as endorsement ofthe opinions expresse<ib y the editor or individual contributors. /67 TheC orrelotiono f LocolC itolion Dolo with CilotionD olo from Journol Citotion Reports Jonice Kreider forlocal citation datau;hen eoaluatingtheirjoum'als, uith certain cauti.ons. lV niversity librarians continue to global, librarians have questioned the rele- search {br data that helns them evaluate iance ol'/CR's data to their own institu- their collections, particularly their jour- tions, pref'erring citation data generated nal collecti<lns, as subscription costs rise liom their own users'publications.T hese approximately 107oe ach year and as addi- Iocal data are more difficult to obtain, par- tional lunds are needed to pay for access ticularly ifone wishes to restrict the data to to {ull-text electronic iournals. Use stud- one subject area, which is crucial because ies are time-consuming if one wants to ob- citation pattems by subiect lield and "ury tain enough data to mai<et he study mean- a{I'ect citation figures. In order to deter- ingful, and the compilation of the results mine how global citation data relates to lo- can be cumbersome. With the recent ap- cal citation data, I explored the correlation pearance oI loumal Citation Reports between global citation data liom the,/CR (JCR) on CD-ROM, quantitative citation and Iocal citation data for the University of data are now relatively simple to manipu- British Columbia (UBC) lbr 20 subject late. Because the citation data on,[CR are lields in the sciences and social sciences. Ja,NIcnK nntoEn ([email protected])is Coordinator ofCollections at the University of British Columbia Library Vancouver. The authorwishes to acknowledge the use ofthe British Li- brary In{brmation Sciences Service and the Oxfbrd University Libraries, William Ng of the UBC Library Systems Division who provided lists of UBC subscriptions by subject, and Kathy Li and Kelly Kwork who were students in STAT 551. Manuscript received June 1I, 1998;,a ccepted {br publication August 18, 1998 68/ LRTS . 43(2) . Kreider LrrEnmunE REvIEw reviews ofthe literature. Broadus (1985, 33) summarized previous studies and Previous studies on the relationship of stated that "counts based on the,fC.R can global and local citation data have been be almost as good as expensive local [use] f'ew in number and limited to only a few studies forpredicting use ofperiodicals in subject lields. Wiberley (1982) it,tdi"d a given library." But Broadus cautions the relationship oflocal and national cita- against automatically eliminating a jour- tion data fbr social work by using an ency- nal that has low global citation counts in clopedia and {bur periodicals as the the library, as there can be valid local rea- sources Ibr the national citations and local sons lbr retaining the title. citations {iom publications {rom lists of Bensman (1985) summarized re- {'aculty publications. He compared the search on the correspondence between method of using earlier national citations global citation data and use, and despite to predict Iater local citations with the some conflicting studies, states (24) that method of using earlier local citations to "ISI citation {iequency is measuring an predict later ones and concluded (358) extremely power{ul variable in academic that fbr social work, "national databaseso f library use, and. . . is undoubtedly one citations are almost asg ood asl ocal data in of the most important measurest hat can predicting luture citation oflournals by be utilized . . . [fbr] managing the jour- local authors." This conclusion therelbre nal collections of research libraries." suggests that national citation data are Bensman was not concerned that the cita- useful fbr jburnal selection. Wiberley's tion data were not local. study took place in a university known {br Todorov and Gliinzel (1988) reviewed its high productivity in publishing in the studies on the relationship between jour- field ofsocial work. but becauset he {ield nal citation measures and objective and was one ofthe applied socials ciences,t he subjective ratings of scientific journals; in {ield exhibits different citation patterns some of the studies there was a positive than does a subject in the pure sCiences. relationship, while in others there was foswick and Stierman (1997) showed not. Kelland andYoung (1994) presented that there was little relationship between a comprehensive review of the literature global citation data fiom/CR and local ci- on the relationship between citation data tation data gathered {rom the three cita- and library use. Because the correlation tion databa^sesfr om the Institute {br Sci- data vary, the authors conclude (86) that entilic Information (ISI) and organized "citations represent a form of literature using DIALOG's "rank" commanil. How- use, and to some extent that can be con- ever, all subjects were considered to- sidered library use. . . . Actually, the relationship between library use and citations is so complex that it should not be ex- pected to produce high correlations." Swigger and Wilkes (1991) compared techniques (including using local citation cial science titles cited by the pro{'essors data generated by a DIALOG search of and ,/CR's top cited titles, which lean ISI) when evaluating journals at Texas heavily toward scientific titles. Woman's University and concluded (52): "There is only a weak correspondenceb e- tween use of serials as measured by reshelving data and by citation data, and no correspondence between citation data and the subjective judgments of faculty or libraries, there is continuing &scussion librarians." The Texas Womant Univer- and debate on the relations[ip between sity has graduate programs in allied citation data and other indicaiors of the health, education, and library science,s o use of journals in libraries. Since 1972, the results are limited to thoie fields. On there have been numerous studies and the other hand, Bensman (1996) showed . Correlation of Local Citation Data /69 in the U.K., stated( 318):" Citationr ank- ings seemedt oo general;w e would not hale t'elt lustifiedln cancelings ubscrip- tionsp urelyon the basiso fexternallygen- eratedl ists." total agreement that the relationship of STATEMENTO F THE PROBLEM citation data with other indicators ofuse AND PURPOSEO F THE STUDY in libraries is strong, it generally is ac- knowledged that such a relationship exists Citation data are commonly used along and that it is worthwhile to take citation with other criteria such as reshelving data into account with other measures statistics and ratings by faculty and librar- when evaluating journal titles. ians to determine core journal titles in a The question remains whether the ci- field or to lind candidates fbr cancella- tation data must be local, or whether tion. Librarians often rely on local cita- global data are just as use{ul. Despite the tion data rather than global citation data, dif{iculties inherent in generating local despite the easeo fgathering global cita- citation data, authors ofarticles published tion data and the ability to manipulate in the last several years reveal tliat librari- them electronically, using the CD-ROM ans are using local citation data in prefer- version oI lCR. Libraria-ns nevertheless ence to, or in addition to, global citation remain unconvinced of the value of global data due to a widespread reluctance to citation data, thanks partly to the lack of rely on global citation data to evaluate studies.T he purpose ofthis studyis to ex- journals. Haas and Kisling-U n(1iv9e9r4si)t yr e- plore the corielation between global cita- ported on a project at the of iion data and local citation data to deter- Florida to eualnite the relevanceo ftheir mine the potential of relying on global sciencei ournal collection using local cita- citation daia instead oflocil citation data, tion dafa produced from ISII Schmidt, fbr evaluating a journal collection. A sec- Davis, and Jahr (1994) used local citation ond purpose is to design and test a consis- tent method lbr obtaining correlations across a number of subject {ields at one specilic institution. SOURCESO F DATA theses and dissertations in a cost ratio to evaluate their psycholory journal collec- The local institution involved in this study tion at St. Mary's University. Hughes (Igg5) examined local citation and publication data along with global data when ranhng journals in molecular and cellular biology at Pennsylvania State University. Dole and Chang (1996) conducted {bur local citation analvses in the fields of marine sciences,c hemistry and sciences as awhole to produce local citation data to use along with f'aculty rankings and use studies in preparation {br a journal has ranged between 25 and 30 during the cancellation project at the State Univer- p^ ast 10 years. sity of New York at Stony Brook. The local citation data were provided Loughner (1996) produced local citation by the Local Journal Utilization Report data {iom ISI to evaluate use of science tiJUn), produced by ISI fiom its dlata- journals at the University of Georgia. baie specilically for UBC. This report, in a Lambert and Taylor (1996), who evalu- database fbrmat, contains counts for the ated journals at Staffbrdshire University number of umes authors from UBC cited 70/ LRTS . 43(2) . Kreider specific journals during a 12.5-year period ation, corresponding to departments at lrom January 1981 to June 1993. The the university. The subjects assigned by L|UR was generated using all author ad- the/CR are finer categories than those as- dresses,n ot solely the {irst author. signed by UBC. In order to have them The 1994 edifion o{,fCR on CD-ROM correspond as much as possible, the JCR produced by ISI was the source of the was filtered to obtain similar subiect global citation data {br this study. The spe- groupings. For example, to correspond cific global citation data used were lhe with the subject of "Chemistry" as de- '1994 Total Cites," which is the number fined by UBC, the following subject cate- of times a given journal title was noted as goriesw ere chosenf rom the/CR: Chem- beinq cited-inI Si'.sS cienceC itation Index istry; Chemistry, analltical; Chemistry, and Social Sciences Citation Index during inorganic and nuclear; Chemistry, or- 1994. ganic; Chemistry, physical; Electrochem- The time periods of the two setso f data istry; and Spectroscopy. do not overlap; the sources were chosen The following databases were loaded because they were available in electronic into a local databaseo n a personal com- fbrmat, which facilitated analpis using puter: the LIUR, the list of'titles lbr each spreadsheetsa nd databases.I SI produced ol'the 20 subjects {iom the,fCR, and the the Lf UR for UBC in lgg3, bu[ the /CR ,/CR lists ofall science titles and social sci- was not available on CD-ROM until the ence titles. Using the UBC subject list- 1994 edition was produced in 1995. ings ol'subscriptions, subject codes were Becauset his proiect was based on ISI keyed into the LJUR. If ;UBC subscrip- data, it has all the limitations of that data- tion was not in the LJUR, the title was base: added, with zero as the number of cita- o errors in the data themselves tions. The databasew as used to create a . the limited number of journals cov- list fbr each subject, with the lbllowing in- ered (an example is the field of law, fbr Ibrmation fbr each lournal: (l) its title as which only g6 titles are listed in JCR, abbreviatedb y ISI;i2) the global citation which led to the exclusion of tho sub- count (i.e., the 1994 total citations from ject oflaw in this study) /CR); and (3) the local citation count (i.e., c JCR is based only on journals; hence the number of times UBC authors cited citations to journal articles from other the title over the 72.5 year period from types ofpublicatiens are not included, the LJUR). The beginning o1'oneo f these and {br sublect areas that are lessj our- lists ii shown in t;ble r lbr the tield of nal-centric, these might be significant zoology. o f'ewf oreigfr languaget itles are included The Pearson correlation coe{Iicient was chosen to measure the correlation between the qlobal citation data and the local citation"data. The Pearson correlation coeflicient reflects the extent ofa linear relationshirr between two sets of data: ther global or local citation data com- it ranges between -1 (indicating a perf'eci parable to those fiom well-established negative relationshlp) and +l (indicating a titles perf'ect positive correlation). A correlation close to +I indicates a stronq correlation. Other correlation stutlies hlve used the METHoD Spearman correlation coeflicient, which In-orde-r to identi{yjournal titles {br a spe- measures the correlation of ranked data. cific subject field, two sources were usbd. However, if the raw data are available (not The lirst was the/CR, which assignse ach just the rankings), it is pre{'erable to use journal title to one or more subie-t fields. the Pearson correlation becauset he actual The secondw as UBCi existing local cod- data give more information, such as the ing ol' all active subscriptioni to one or varying size ol'gaps between the ranked more ofthe 20 subiects under consider- data. Belbre applying the Pearson correla- LRTS . a3Q) ' Correlationof LocalCi.tationData /7I TABLE 1 Zoor-ocr fouRNALT ITLESw ttH Gloner- (rnou/CR) nrupL ocer. (FRoML TUR)C ITATIoNC ouurs Abbreviated Title Global Local ACAROLOGIA r42 0 38 ACTAANAT L,DIZ ACTA BIOL CRACOVZ OO 3 0 ACTAPHYTOPATHOLH UN 97 0 0 ACTAPROTOZOOL lvD ACTATHERIOL 360 I3 0 ACTA ZOOL HUNG JJ ACTA ZOOL.STOCKHOLM 397 J ADV INSECT PHYSIOL 425 34 ADV PARASIT 481 12 ADV STUD BEHAV 336 13 AFR J ECOL 189 D AM BEE I IDD 0 AM ENTOMOLOGIST 0 4 AM J PHYSA NTHROPOL 2,642 41 AM J PHYSIOL 78,546 2,634 AM J PRIMATOL 826 0 AM MALACOL BULL .to 4 AM ZOOL 2,652 202 AMPHIPACIFICA 0 0 tion, the citation ligures were transfbrmed using the logarithmic trans{brmation. The decision to transfbrm the data was based on an examination of the histograms of the data {br the field of microbiolory. For that Iield, both the global and the local citation tent. The LJUR required extensive editing data sets were dominated by a small num- in order to combine titles with di{I'erent ber of highly cited journals; in statistical abbreviations, to combine tides that had terminology, the variance increases with changed, and to correct errors. When the the mean, resulting in a negative binomial editing was completed, a great many titles distribution. The other subject fields ap- listed ivith only one citatioi remained, and peared to have similar distributions. Fig- many of these were dif{icult to identifr by ure I illustrates the dominance in the field full title, either because they were eso- of zoology by one title, in this case the teric, ambiguous, incorrectly ibbreviated, Am.erican Joumal of Physiology, for both or ref'erreito monographs. Examples of local and global citations. These observa- such titles include PHYSIS, PLENARY tions agree with those of Bensman (1996), LECTURE, and fiB PSYCHIATRY. In who has worked with similar citation order to eliminate having to spend undue counts and who noted that the logarithmic time identilying these titles of little im- transformation is an appropriate way of 'pboarstaen ce and in order to make the data- dealing with such data. smaller and easiert o manipulate, all 72/ LRTS . 43(2) . Kreider = Go o 2,000 o o 1,500 E o o 1,000 @ f 40,000 60,000 80,000 100,000 JCFCites( Global) Figure 1. Scatter Plot of Raw Data for Zoolog;r Titles, needed, because it was alreadv several years old and numerous tiiles had changed or split into parts, so data for the various versions of a title had to be gathered underthe latestversiono fthe nime. CANADIAN, BC, BRITISH COLUM- The Lf UR ended up with 10,60I titles, BIA, VANCOLIVER, etc., because we with locai citations ran'ging from a high oi wishedto retain as much Canadian data as 5,350 fbr the Proceedligs-of the Nat{onal was feasible. Although not necessary {br Acuderny of Sciencest olheiSS titles with this correlation stuay, obvious mbno- graphs; titles indicated as INpRESS, UNPUB, and THESIS; and titles beginning with numbers or abbreviationi of months were eliminated. Another modification to the Lf UR was its expansion to include 755 titles in the 20 1,402 titles, with global citations ranginq subject fields to which UBC actively sub- liom a high of 20,638 for Archioes of Ceri scribes. These titles had a local cilation eral Psychiatry to a low of one citation for Discussion of Results citations at UBC, foreign titles, and titles perhaps a bit outside aitrict delinition of journal such as the Yearbook of Astron- orny. Modiftcations to the /CR were also LRTS . 43(2) ' Corcelation of Local Citati,on Data ns TABLE 2 Connn,rrroN or GLoseL enn Local CImtloN Dn.Ie Pearson PearsonC orrelation Correlation of IngarithmicallY Srrbier:tF ield of RawD ata Trmsformed Data Microbiolory 89 0.953 0.814 Forestry 140 0.817 0.810 Astronomy 42 0 963 0.755 Economics 260 0.916 0.731 Businessa nd Management 279 0.849 0.695 Biochemistry 272 0.959 0.687 Pharmacy 176 0.832 0.682 Computer Science 254 0.696 0.681 Mining 39 0.617 0.677 Biolory 346 0.968 0.675 Botany 154 0.783 0.661 Mathematics 307 0.800 0.634 Zoology 220 0.927 0.634 Physics 343 0 903 0.625 MechanicalE ngineering r77 0.816 0.623 Librarianship I35 0.613 0.622 Geology 208 0.818 0.607 Chemistry 40r 0.871 0.567 Metalsa nd Materials 2r2 0.579 0.547 Chemical Engineering 148 0.477 0.530 N = number ofjournal tides in each subject of journal titles in each subject cohort. ject, a relationship that gets decidedly The correlations between the global and weaker for the smaller c'itation values. local citation counts using the raw data Figure 2 shows the relationship lbr zoology. Note the cone shape of the data p6ints that gets wider for the smaller lrulu"r, indic*ating the increasing weak- nesso fthe relationship as the citation values decrease.T he {act that the linear rela- bution ofthe data (large clusters ofboth tionship between the global and local global and local citation data dominating citation transfbrmed dat"ag ets weaker lbr the low end) indicates that more realistic the smaller values has implications for correlations can be obtained from the libraries. lt means that a low global cita- transformed data. This led to the ratio- tion count does not always imply a corre- nale {br arranging the table in descending order by correlation ofthe data after a logarithmic transformation was performed. The actual ranking of the subiects is less important than the {'act that the correlation ligures are moderate to moderately mit any general conclusions about the high fbr all the subiects. The scatter plots strength of a subjectt correlation based of the transfbrmed data show a linear rela- on whether the subject is a pure or an ap- tionship of varying strength fbr each sub- plied science, a science or a social sci- 74 LRTS . 43(2) . Krei.der - 810 -98 oG o- o2 o0 D o JCB Cites (Global) Figure 2. Scatter Plot of Translbrmed Data fbr Zoolog;r Titles. titles; recalculating the correlation with- out the Canadian t-itles gave a correlation oI'. 794 instead of the .81-0w hen the Cana- dian titles were included. The /CR data (1994) and the LJUR An interesting phenomenon is that data (1981-93) obviously do not come local citation data-a ie zero lbr numerous liom the same time periods. At the time joumals in all 20 subject fields. There of this study, there was only one year of .seemt o be three reasons for this. First of JCR data available on CD-ROM, and it alI, 755 titles were imported into the sub- might have seemed more worthwhile to have used only one year oflocal data (and liom the r"tn" y""i) to correspond pre- cisely with the one year oI global data. However, fbr a given year, local citation data are considerably sparser than global citation data. Just as use studies based on reshelving counts require a long time reason is that most of the titles from the liame to be representative, a citation LJUR that had only a single citation were study using merely one year of local cita- culled. Finally, some of these titles are sim- tions would not be asi ndicative of local ci- ply of little rele'nr.r"" to research at UBC tation practices as more years would be, (e.g., American BeeJ ournal). both in terms of the variety ol' titles cov- It might be assumedt hat researchers ered and citation counts themselves. at UBC would cite Canadian iournals making a reliable correlation {igure dif{i- more heavily than would be the cise glob- cult to obtain. Whether a{'ull 12.5 years of ally, which would produce relativelv Iocal citation datawas necessaryi s open to lower correlations in this study th# debate.C ertainlywith onlyone yeai ol'lo- would be obtained lbr an institution in the cal data, the co.r'"lution. *onld have been U.S. However, each subiect {ieldincludes only a {'ew Canadian titl-es, and olten only one is truly important, so the effect is minimal. The subject list {br business and management includes 25 Canadian titles, more than the other disciplines. Exclud- Line (1985) lbund a 92Vo overlap f'or the ing the 25 Canadian titles and then recal- top 500 journals cited the most fre- culating resulted in a correlation oI .702, quently in ISI'.s Science Citation lndex compared with the correlation of .6g5 (SCf) in 1979 and in 1982. For the Social when the Canadian titles were included. ScienceC itation Index (SSCf), there was However, the opposite ell'ect occurred in an877oo verlap for the same time period. lbrestry, whicli lncluded l4 Canadian Hence, using/C.R data (which are drawn