Upload
xiaotian-chen
View
215
Download
3
Embed Size (px)
Citation preview
Overlap between Traditional Periodical Indexes
and Newer Mega Indexes
Xiaotian Chen
Available online 7 September 2006
0098-7913/$–see fro
doi:10.1016/j.serrev.
Chen is ElectronPeoria, IL 61625,
The author is graUniversity for edit
This article studies the overlap between the traditional periodical indexes and thenewer mega indexes or databases. Some traditional general indexes have 100%overlap with the mega index from the same vendor and about 90% overlap withmega indexes from other vendors. Other traditional and more specialized subjectindexes have considerable uniqueness when compared with the newer megadatabases. Serials Review 2006; 32:233–237.D 2006 Elsevier Inc. All rights reserved.
Introduction
Traditional periodical indexes, such as Art Index,Humanities Index, MLA International Bibliography,PsycINFO, Readers’ Guide to Periodical Literature,PAIS International, and Social Sciences Index, have beenessential to library services for nearly a century. It isprobably impossible to find a single American academicor public library that has never provided access toReaders’ Guide to Periodical Literature. In fact, manylibraries, including the author’s home institution libraryand hometown public library, have current subscriptions,print or online or both, to Readers’ Guide.Since the 1990s,manymega general index services have
come into being. Some of these are indexes only (no fulltext), such as OCLC’s ArticleFirst; some have full text orpartial full text, such as EBSCO’s Academic Search andMasterFILE databases, Gale’s Expanded AcademicASAP, and H. W. Wilson’s OmniFile databases; others,such as Ingenta, facilitate document delivery and otherservices. These databases vary in cost. Some are fairlyexpensive with prices of over $10,000 per year; some arebundled (e.g., ArticleFirst is part of the FirstSearch basepackage subscribed to by most academic libraries); andsome are free (e.g., the index part of Ingenta). EBSCOdatabases are also available in a large number of academiclibraries across America. Those academic libraries that donot have mega indexes from EBSCO probably have some
nt matter D 2006 Elsevier Inc. All rights reserved.
2006.07.002
ic Services Librarian, Bradley University,USA; e-mail: [email protected].
teful to Karen McClaskey of Truman Stateing this paper.
233
from H. W. Wilson, from the Gale Group, or ProQuest.Probably all, or a preponderance of, American academiclibraries have at least one of these mega indexes.
This article examines whether the newer mega indexeshave significant overlap when compared with the period-ical coverage of traditional indexes. If there is significantoverlap, the authorwill quantify the overlap and show thelist of unique titles from the traditional indexes, so thatlibrarians can decide if it is worthwhile to continuesubscribing to the traditional indexes that libraries haveacquired for decades or perhaps evenmore than a century.
Literature Review
Because the mega indexes discussed in this article did notcome into existence until the mid or late 1990s,publications before the 1990s are not relevant to thisstudy. No recent literature comparing traditionalindexes with newer mega indexes at the periodical titlelevel has been found. Myke Gluck summaries the historyof journal coverage overlap studies as of 1988, examinesthe definition of journal coverage overlap in abstractingand indexing services, and expands the definition usinga matrix of dissimilarity values.1 William Hood andConcepcion Wilson published some general studies onduplicate records from bibliographic databases, with themost recent one studying record overlap by using aspecific topic in 2003. The 2003 study compares therecords retrieved from various databases on a certaintopic for database overlap and has a very comprehensivebibliography at the end of the article.2 Peter Jacsoanalyzes journal coverage of six databases during 1966and 1996 by focusing on forty-two journals in libraryand information science.3 Johannes Voigt et al compare
Table 1. Social Sciences Index unique active titles
Indexes Compared withUniqueTitles
Percentage(out of 490)
Wilson’s OmniFile Mega 0 0EBSCO’s Academic
Search Premier98 20
Gale’s Expanded AcademicASAP
27 5.5
OCLC’s ArticleFirst 31 6.3
Xiaotian Chen Serials Review
the NCI (National Cancer Institute) open database withseven chemical structure databases.4 Janet Hughesstudies the unique titles in Zoological Record comparedwith BIOSIS and summaries the characterization of theunique titles in Zoological Record.5 Xiaotian Chencompares full-text overlap between subject indexes andcomprehensive indexes from one vendor at the full-textlevel rather than at index level.6 Joan Parker evaluatesbibliographic database overlap for marine scienceliterature using an ecological concept.7
General Methodology
Index sources (title lists) are retrieved from providers’Web sites. EBSCO’s title lists link can be found on theEBSCO homepage at http://www.epnet.com/; Gale’s titlelists link is on its homepage at http://www.gale.com/;OCLC FirstSearch’s title lists are at http://www.oclc.org/firstsearch/periodicals/default.htm; H. W. Wilson’s titlelists can be generated at http://vnweb.hwwilsonweb.com/hww/Journals/. In order to assure consistent andfair comparisons, all title lists were retrieved fromvendors’ Web sites in March 2006. Microsoft Excel isused to compare overlap by finding identical periodicaltitles and ISSNs. Title lists on vendors’ Web sites aretypically in Microsoft Excel file and Text file. If a certainvendor offers Text file but does not offer Excel file, theauthor converts the Text files into Excel files. Theflagship or the most popular general mega indexes fromthe major vendors (EBSCO, Gale, OCLC and H. W.Wilson) are used to compare with the traditional orclassical periodical indexes. Some indexes studied arefull text or partially full text, but the focus of this studyis on index overlap, not full text overlap. Periodicals thathave ceased to be indexed by those indexes are not takeninto consideration. In their title lists, some vendorsclearly indicate non-active titles. Non-active titles mayinclude journals that have ceased publication, journalsthat are still being published but have been dropped bythe indexes under discussion, or journals that have newnames and are currently being indexed under those newnames. From title lists of those vendors that clearlymarked non-active titles, it is very easy to sort active andnon-active titles with Microsoft Excel. Other vendorssimply provide the starting year and most recent year ofthe titles indexed. From these title lists, if periodical titlesdo not have 2005 or newer volumes and issues indexedas of March 2006, they are not considered as activetitles, and thus will not be included in overlap compar-isons as active titles. The index starting year for a certainperiodical may vary on different indexes, and this factoris not taken into consideration. The comparison resultsare lists of unique periodical titles in each indexcompared with one newer mega general index.
Social Sciences Index Overlap with MegaIndexes
Social Sciences Index is one of the traditional or classicalindexes from H.W.Wilson. It dated back as early as 1907with the name of International Index; then it changed to
234
Social Sciences & Humanities Index; later it split into twoseparate indexes, one being Social Sciences Index, theother Humanities Index. Social Sciences Index currentlyhas three versions offered by H. W. Wilson: SocialSciences Index, Social Sciences Abstracts, and SocialSciences Full Text. All three versions have the samenumber of periodical titles indexed or abstracted. As ofMarch 2006, it covers a total of 621 periodicals, of which490 are active. That means Social Sciences Index/Abstracts/Full Text currently indexes 490 titles plus 131ceased or dropped periodicals it no longer indexes. Table1 shows the unique active periodicals Social Sciencescovers when compared with four mega indexes.
Indexes with Similar Pattern to Social SciencesIndex
Three other traditional indexes from H. W. Wilsonfollow a similar pattern when compared with the megaindexes. They are General Science Index (or Abstracts,Full Text), Humanities Index (or Abstracts, Full Text),and Readers’ Guide Index (or Abstracts, Full Text).Table 2 shows their profiles. Tables 3 through 5 are thecomparisons with the mega indexes from Wilson,EBSCO, Gale, and OCLC.
Analysis of the Four Wilson Indexes
H. W. Wilson’s Humanities Index, Readers’ Guide, andSocial Sciences Index have been in libraries for about onehundred years, and General Science Index has been inlibraries for a quarter of a century. They are the fourclassical periodical indexes that are all somewhat non-subject specific, or they all have fairly broad subjectcoverage. For the sake of convenience, the term bWilsonFourQ will be used in the rest of this article to refer tothese indexes. The above comparisons show that theWilson Four have 100% overlap with H. W. Wilson’snewer mega index OmniFile Mega. The Wilson Fouralso have significant overlap with the mega indexes fromother vendors such as EBSCO, Gale, and OCLC. Forexample, General Science Index has only 21 (9.8%)unique active titles compared with OCLC’s ArticleFirst;Humanities Index has only 45 (9.2%) unique active titlescompared with Gale Expanded Academic ASAP; Read-ers’ Guide has only 28 (11.4%) unique active titles whencompared with EBSCO’s MasterFILE Premier; SocialSciences Index has only 27 (5.5%) unique active titlescompared with Gale’s Expanded Academic ASAP.
Table 2. Profiles of General Science Index, HumanitiesIndex, and Readers’ Guide
Index nameTotal periodicals
coveredTotal activeperiodicals
General Science Index 290 215Humanities Index 588 491Readers Guide 401 246
able 4. Humanities Index unique active titles
Indexes compared withUniquetitles
Percentage(out of 491)
ilson’s OmniFile Mega 0 0BSCO’s Academic Search Premier 79 16.1ale’s Expanded Academic ASAP 45 9.2CLC’s ArticleFirst 90 18.3
Volume 32, Number 4, 2006 Overlap between Traditional Periodical Indexes and Newer Mega Indexes
Wilson’s OmniFile Mega contains 100% of allWilson Four titles. Therefore, it seems reasonable toconclude that if a library hasOmniFile Mega, there is noneed to subscribe to any one of the Wilson Four. Forthose libraries that do not have Wilson’s mega index butdo have one or more mega indexes from other vendors,such as EBSCO, Gale, or OCLC, the question arises: Is itnecessary to keep the Wilson Four that have about 10%unique titles, or will the library be able to provideadequate service without having an index to thoseunique titles? For guidance in decision making, theunique titles are presented in Table 6 and 7.Table 6 lists the twenty-one General Science Index
unique titles as compared with ArticleFirst. Many ofthose twenty-one titles can be found in other commonlyheld indexes and databases. For example, The NewYork Times is indexed by LexisNexis and many otherdatabases; Horticulture is indexed by many EBSCO andGale databases; Nursing Research is indexed by MED-LINE and CINAHL.When more than one mega index/database is factored
in, the unique titles from the Wilson Four indexes arereduced to single digits or to zero, even if Wilson’sOmniFile Mega is not taken into consideration. Each ofthe Wilson Four indexes has zero unique titles comparedwith Wilson’s OmniFile Mega. Table 7 shows a total ofonly two General Science Index unique titles comparedwith three non-Wilson mega general indexes (Article-First, Academic Search Premier, and Expanded Aca-demic ASAP).
Beyond the Wilson Four
The Wilson Four indexes, General Science Index,Humanities Index, Readers’ Guide, and Social SciencesIndex, are neither very general nor very specializedindexes. Do the more specialized subject indexes, such asArt Index and PAIS International, also significantlyoverlap with the newer mega indexes? Similar to theWilson Four, these subject-specific indexes have beenlibrary research tools for decades.Comparisons reveal that subject-specific indexes do
show much less overlap with the newer mega indexes.
Table 3. General Science Index unique active titles
Indexes compared withUniquetitles
Percentage(out of 215)
Wilson’s OmniFile Mega 0 0EBSCO’s Academic Search Premier 24 11.2Gale’s Expanded Academic ASAP 26 12.1OCLC’s ArticleFirst 21 9.8
235
T
WEGO
Art Index, which is owned by H. W. Wilson, has 187(56.5%) unique titles among its 331 active titles whencompared with Wilson’s OmniFile Mega. This is verysignificant when considering the fact that the WilsonFour have zero unique (0%) titles against the sameOmniFile Mega. The number of Art Index’s uniquetitles is even higher when compared with non-Wilsonmega general indexes. When compared with EBSCO’sAcademic Search Premier, Art Index has 255 (77%)unique titles among its 331 active titles. PAIS Interna-tional, which is owned by CSA and indexes 621 activeperiodicals, follows a similar pattern to Art Index: 255(58%) unique titles compared with Academic SearchPremier. The number of PAIS International’s uniquetitles would rise if the non-periodicals (such as booksand book chapters) it covers are taken into consider-ation. While relatively small subject indexes, such as ArtIndex and PAIS International, have considerable uniquetitles compared with newer mega indexes, by extrap-olation, larger subject indexes, such as MLA Interna-tional Bibliography (about 17,000 total titles indexed,over 1,000 active titles) and PsycINFO (about 13,000total indexed, over 2,500 active), would probably have ahigher percentage of unique titles when compared withthe newer mega general indexes, since MLA andPsycINFO are about the same size as the mega generalindexes. Indexes with more specialized contents such asSciFinder Scholar would also have unique contents,since SciFinder Scholar covers millions of substancesand provides additional features which general indexescannot handle.
The Full-Text Factor
Since this study compares the overlap or unique titlesfrom various periodical indexes in terms of indexing andnot with regard to full-text coverage, full-text access isnot a factor; however, one full-text phenomenon meritsconsideration. Vendors handle full-text coverage differ-ently in their own general full-text indexes and subjectfull-text indexes. The full-text titles in all H. W. Wilson’ssubject indexes are 100% included in Wilson’s generalfull-text indexes (OmniFile Mega or its variants),
Table 5. Readers’ Guide unique active titles
Indexes compared withUniquetitles
Percentage(out of 246)
Wilson’s OmniFile Mega 0 0EBSCO’s MasterFILE Premier 28 11.4Gale’s Expanded Academic ASAP 94 38.2OCLC’s ArticleFirst 113 45.9
Table 6. General Science Index unique titles compared withArticleFirst: 21
Title ISSN Starting year
Ambio 0044-7447 1997/07American Gardener 1087-9978 1996/05Biomechanics (Rockville, Md.) 1075-9662 2002/07Bulletin of the American
Mathematical Society0273-0979 1992/06
Chemical Heritage 0736-4555 2002/03Conservation Ecology (Online) 1195-5449 2002/01Earthwatch 8750-0183 1997/07Electronic Green Journal 1076-7975 2002/04Evolution 0014-3820 1982/11HealthInform’s Resource
Guide to Alternative Health1090-6649 1997/01
Horticulture 0018-5329 1984/04Integrative and
Comparative Biology1540-7063 2002/02
Journal of Mammalogy 0022-2372 1984/08Journal of Undergraduate
Mathematics (Online)0022-5339 N/A
Nature Conservancy 0028-5200 1997/07New York Times (Late New
York Edition)0362-4331 1993/06
Nursing Research (New York) 0029-6562 2002/01Practice Nurse 0953-6612 2002/06Skeptic (Altadena, Calif.) 1063-9330 2001/12Weekly Epidemiological
Record (Online)N/A 2006/01
Wilson Bulletin (Lawrence, Kans.) 0043-5643 1984/06
able 7. General Science Index unique titles compared withmega indexes: 2
Title ISSN Starting year
onservation Ecology (Online) 1195-5449 2002/01ournal of UndergraduateMathematics (Online)
0022-5339 N/A
Xiaotian Chen Serials Review
according to sources provided by H. W. Wilson at http://vnweb.hwwilsonweb.com/hww/Journals/. EBSCO’ssubject full-text indexes have unique titles comparedwith EBSCO’s general full-text indexes, according tohttp://www.epnet.com/titleLists.php?topicID=380&tabForward=titleLists&marketID=. One well-knownEBSCO example is that Harvard Business Reviewfull text is available on EBSCO’s Business SourceElite and Business Source Premier, but it is notavailable on any of EBSCO’s general full-text indexes.Since many newer general indexes are full text orpartially full text (e.g., EBSCO’s Academic Searchand MasterFILE databases, Gale’s Expanded Aca-demic ASAP, and Wilson’s OmniFile Mega), replac-ing the Wilson Four indexes with mega generalindexes would be more cost effective. In this way,libraries could not only continue to provide access tothe contents covered in the Wilson Four, but wouldalso gain access to all full-text periodicals availablefrom the mega indexes. If libraries chose Wilson’sOmniFile to replace the Wilson Four, they wouldlose nothing and gain all full-text periodicals avail-able from H. W. Wilson.
Summary and Conclusion
Just as traditional indexes were essential resources forfinding periodical articles before the 1990’s, the mega
236
general indexes from various vendors have becomeessential resources for libraries in recent years. Mostindex/database vendors offer at least one mega generalindex. As a result, most libraries typically subscribe tomore than one mega index. For example, the author’shome library has one mega index from each of the majorproviders: EBSCO, Gale, OCLC, ProQuest, and H. W.Wilson, with some being paid for by consortia and someby the library. With the role of the newer mega indexesincreasing in importance, librarians may now need toconsider whether they should keep all the bold playersQ-the traditional or classical indexes.The study described in this article finds that some
traditional periodical indexes are 100% covered by themega indexes from the same vendor. When comparedwith a single mega index from a different vendor, sometraditional indexes have 10% unique titles and that10% will decrease when more than one mega index isfactored in. The study finds these traditional indexesinclude General Science Index (or Abstracts, or FullText), Humanities Index (or Abstracts, or Full Text),Readers’ Guide Index (or Abstracts, or Full Text), andSocial Sciences Index (or Abstracts, or Full Text). Theyhave some common features: all have been around fordecades and can be considered as classical indexes; allwere vital library research tools during the 20th century;all are fairly small in regard to the number of periodicalscovered (a few hundred); all are somewhat general orcover a broad range of subjects.The study also finds that traditional indexes that are
more subject-specific or more specialized (such as ArtIndex) tend to have a higher percentage of uniquecontents when compared with the newer megaindexes. Wilson’s Art Index has well above 50%unique titles when compared with Wilson’s OmnifileMega, and the percentage of the unique titles is evenhigher when compared with non-Wilson megaindexes. Similarly, PAIS International also has wellabove 50% unique titles compared with a mega index.The fact that these fairly small indexes covering a fewhundred active titles still have over 50% unique titlesis very significant. Larger subject indexes, such asPsycINFO, which is as large as the mega generalindexes in terms of number of journals covered, wouldcertainly have more unique contents when comparedwith the newer mega indexes.Therefore, for the libraries that still have current
subscriptions to those traditional indexes that arecovered 90%–100% by the newer mega players in theindex field, it would be cost effective to consider sayinggoodbye to the classical players, especially whenlibraries face tight budgets. Even when cost is not an
T3
CJ
Volume 32, Number 4, 2006 Overlap between Traditional Periodical Indexes and Newer Mega Indexes
issue, the trend of one-stop-shopping, bGooglizing,Q andthe federated (or meta) search in the informationretrieval field also favors newer mega players oversmaller traditional players that have no unique contents.
Notes
1. Myke Gluck, bReview of Journal Coverage Overlap with anExtension to the Definition of Overlap,Q Journal of the AmericanSociety for Information Science 41, no. 1 (1990): 43–60.
2. William W. Hood and Concepcion S. Wilson, bOverlap in Biblio-graphic Databases,Q Journal of the American Society for Informa-tion Science and Technology 54, no. 12 (2003): 1091–1103.
3. Peter Jasco, bAnalyzing the Journal Coverage of Abstracting/Indexing Databases,Q Library and Information Science Research20, no. 2 (1998): 133–151.
237
4. Johannes H. Voigt, Bruno Bienfait, Shaomeng Wang and Marc C.Nicklaus, bComparison of the NCI Open Database with SevenLarge Chemical Structural 5 Databases,Q Journal of ChemicalInformation and Computer Sciences 41, no. 3 (2001): 702–713.
5. Janet Hughes, bCharacterization of Unique Serials Indexed in theZoological Record,Q Issues in Science and Technology Librarian-ship no. 30 (2001): http://www.istl.org/01-spring/refereed.html(accessed July 6, 2006).
6. Xiaotian Chen, bThe Secrets of Full-Text Databases: The OverlapBetween a Same Vendor’s Subject Database and General Data-base, and the Differences Between Different Vendors in Embargo,QBrick and Click Libraries Conference (2002).
7. Joan Parker, bEvaluating Bibliographic Database Overlap forMarine Science Literature Using an Ecological Concept,Q Issues inScience and Technology Librarianship no. 42, (Spring 2005),http://www.istl.org/05-spring/refereed-2.html (accessed Feb 28,2006).