Peter's Digital Reference Shelf

Title: PsycINFO (WoK version)  

Publisher: American Psychological Association and Thomson ISI

URL: http://scientific.thomson.com/support/products/wok

Cost: To be negotiated

Tested: June 4-7, 2006

One of the most widely used indexing/abstracting database benefits from the synergy with the Web of Science (WoS) database, and reciprocates the advantages by enhancing the psychology and psychiatry sections of  WoS  through a broader, discipline-specific  journal base, and through the inclusion of  records for monographs, book chapters and dissertations.

 THE CONTEXT

Psychology is one of the most popular disciplines (especially at the pop culture level) in U.S. colleges. There is so much interest (and convenience) in taking Psychology 101 courses that some colleges could get away with offering them without any formal contact with qualified faculty members. Students pay the tuition,  buy the textbook, do the exam evaluated by computer, get their credits without ever having face time with faculty, or class meeting. This beats the most money-making online, or -more euphemistically- asynchronous courses in terms of investment of money by the university and of time by the students. It seems to be “good for the goose, good for the gander”, or so the ganders believe in the short run. 

It looks good and generous in the course descriptions to offer unlimited access to the PsycINFO database, as a digital student aid, so it is one of the most widely licensed databases. (It is another matter that undergraduate  students rarely read scholarly journal articles, let alone need to do comprehensive subject searches day in day out).

PsycINFO is available on the most online services from Dialog to DataStar, to SilverPlatter, OCLC, Ovid, ProQuest, EBSCO, CSA-IDS and Web of Knowledge. The latter two are far the best, but EBSCO’s version  also seems to get a second wind after its recent revision. I reviewed PsycINFO more than two years ago on the CSA platform which I found to be the best at the time. When ISI jumped on the bandwagon of licensing 3rd party content, with PsycINFO as their first choice, it was a good move, being the only 3rd party database enhanced (although selectively) by cited references, the home turf of ISI. The recent introduction of the new WoK platform early June, makes it an excellent PsycINFO implementation. In addition, the database content has been further enriched in the past two years by adding cited references retrospectively to pre-2001 records. 

This is good news especially as there has been only modest competition for PsycINFO from other content providers of discipline-oriented databases. These were marred by incompetence, illusion  and self-delusion by the providers. The e-psyche database which served a wake-up call and prompted APA to engage  into a very significant record enhancement project turned out to be a huge fiasco, with utterly naïve day-dreams presented as reality by two information industry veterans, fooling customers and themselves  for a good year, before folding. It took longer for the Mental Health Abstracts database to disappear, even though the provider who took over the database in the early 1980s systematically emaciated the database mha for 20 years, which once was a real competitor to PsycINFO, but became an embarrassment for the LIS profession. It represented the worst example of the endangered species of indexing and abstracting databases for reasons I discussed and illustrated in my talk at the UKSG conference earlier this year. The cheap but very expensive  database finally was removed form Dialog’s stable of databases in early 2006.

The only discipline-specific alternative to PsycINFO is the Psychological and Behavioral Science Collection of EBSCO which may have only less than a third of indexing/abstracting records of PSycINFO, a narrower journal base, and shorter retrospective time span, but for more than 524,000 it has the full text of the document in PDF format (less than 1% in HTML). However, this database does not have (yet) the cited reference searching and linking functions, which has represented the greatest improvement in PsycINFO  in the past 5 years.

On the front of citation-based and citation-enhanced databases the alternatives to PsycINFO are the psychology/psychiatry subsets of the multidisciplinary Scopus and Web of Science databases .

 THE CONTENT

The content provider, the American Psychological Association has an unusually detailed and  informative page about the content and composition of PsycINFO. It has a remarkably good mix of indexing/abstracting records for scholarly journal articles, monographs, textbooks, conference proceedings, conference papers, dissertations, reports, digital materials and even encyclopedia entries. Although 97% of the PsycINFO records are for English language materials, in such a large database, 3% is quite substantial. There are more than 22,000 records for Spanish language materials, and nearly 30,000 for German  language materials which is good news not only for the Germans, but also for researchers who prefer German in Switzerland and Austria where among others Jung, Rorschach, (the Hungarian-born) Szondi, Adler, Reich  and above all  Freud laid down many of the ground principles of psychology, psychiatry and psychoanalysis.

PsycINFO has more than 2.25 million records as of early June, 2006, quite an impressive number. The actual number of records in the various implementations by the online hosts may differ depending on the most recent update date of a specific implementation.

As it turns out, with the June 4th update, the WoK implementation of PsycINFO is the largest and most current, with 2,275,076 records. Ovid’s version may seem to offer a higher number of records when you check its size, but it is misleading, because Ovid keeps insisting on retaining the corrected records when it adds the correction records. I keep insisting that it is not the way to do the updating. I keep saying that the previous version of the records should be replaced. Ovid’s practice produces a few thousand duplicates in Ovid’s version of PsycINFO like this one. Luckily, Ovid has the best duplicate removal utility (for multiple database searching), so with a little trick these can be removed as I show in an excerpt here.    

For comparison, the Psychology, Psychoanalysis and  Psychiatry sections of the Social Science Citation Index database of ISI (which starts from 1972) has close to 1 million records. (The Science Citation Index database may have a few thousand records in psychiatry which are not included in the Social Science Index database). PsycINFO for the same 1972-2006 period has about 1.79 million records. The Psychology section of Scopus has close to half a million records, and 465,328 records of this subject domain has records for documents published from 1972 onward.

PsycINFO has the largest number of abstracts, close to 1,992,000, i.e. 89% of records have abstracts in PsycINFO. For the 1972-2006 period this proportion goes even higher, to 91%. From the 1990s practically all records have abstracts. This is valuable as it helps in selecting the most promising items from a result list. Again, for comparison, for the 1972-2006 period PsycINFO has 1,625,000 abstracts, the Social Science Citation index psychology, psychoanalysis and psychiatry sub-domain 297,000 and Scopus psychology domain about 360,000.

As for the number of records enhanced by cited references, APA provides yet another very informative and current page of background information about the cited references and statistics which rhymed with my own test results, and seem reliable. It tells you among others, that more than half a million records have been enhanced with about 20 million cited references, yielding an average of 42 citations per record. (It should be noted that this very high average is mostly due to the inclusion of records about books ( a good idea in psychology), which tend to have orders of magnitude more cited references than the average  articles).

APA provides year by year details of the  number of records enhanced and the total number of citations added. The total number of records are not presented, so one may not have a good feel for the proportion of the records which have been enhanced. My test results indicate that between 2001 and 2006 about 70-98% of the records have been enhanced. There is an expected but still  sharp drop for 1999 and 2000 (to about 20% and 22%, respectively), and from there on the proportion fluctuates between 3-6% with some whopping outliers on both ends, such as the 0.07% rate for publications of 1986 and the above 14% rate for 1959 and 1965.

In this regard the psychology, psychoanalysis and psychiatry subject categories of the  Social Science Citation Index database (from 1972 onward) have 822,000 such records which is way above of what PsycINFO offers. In the entire WoS database (from 1945) the  number of citation enhanced records about articles in the psychology and related subject category is much higher but cannot be determined as the subject categories are not used in WoS, and search sets are limited to 100,000 records]. For the 1972-2006 period PsycINFO has about 485,000 records enhanced with cited references.  Scopus which enhanced its records with cited references from 1996 has about 236,000 such records. 

THE SOFTWARE

There are significant differences in the software aspects of PsycINFO implementations in terms of browsing, searching and displaying/printing/saving records. WoK offers right from the search template browse options for most of the data elements of the master records (author, corporate author, author affiliation, source titles, classification codes, and descriptors). This integrative approach increases the chance that some users would look up, say, the author index to find variations for the author names with different middle initials, and typos. Of course, this does not help if the misspelled name is not in close vicinity of the correctly spelled version as is the case with jasco  instead of jacso. Several data elements which are used to limit the search (publication type, language, age group) are browsable directly on the search template in form of pull-down menus.

The PsycINFO thesaurus is nicely implemented, making it easy to navigate the branches and sub-branches of a term. I would like to see the posting information next to the term to get a feel how often the terms were assigned to a record, as it is so well done in Ovid  which also offers two other powerful features. One is term explosion to include the narrower terms of a descriptor automatically. The other is the focus feature  to limit the search to records which have the descriptor in a major descriptor role, indicating that it is a key topic in or the focus of the document.

In the WoK implementation the explode and major descriptor options are not available.  Neither are browsable indexes available for cited author, cited journal, and cited book title which could help to realize how many ways there are to spell and misspell author names, abbreviate journal or book titles in cited references.      

Search options are fine for an abstracting/indexing database, i.e. the lack of adjacency operation is not critical. The key software issue is how the software allows the users to benefit from the citation enhanced records. Searching cited references is simple, you may combine in a single cell the cited author, and journal name as well as the cited year. It would be better if the former two would be also browsable. 

Of course, the uniqueness of the citation enhanced databases is that you can switch between keyword based and citation based searching, especially when you are lead or ushered from one to another, and when getting to a promising record of the keyword search, it can be used as a stepping stone for WoK to show the various citation-searching option from that record.   

Search results can be splendidly refined by the new Refine feature, which maps out from the result set the values of the descriptor, journal name, and several other data elements. Selecting  one or more descriptors will narrow the search. This is very similar to the Analyze function, but it is “in your face” in a very intuitive manner, making it impossible to overlook it. The Refine function is ideal for scanning several filtering criteria and values. The grid-like layout facilitates the process of  fishing for criteria to  refine the query and to see how it filtered the results. The Analyze function is more appropriate for the post-processing phase to see trends in the coverage of a topic year by year, or by various journals.

It is no accident that in the Refine function the values of the selected data elements are presented by term frequency  while in the Analyze function you can also sort the values alphabetically or numerically to see the emergence of a topic, such as PTSD throughout the years. You may also download the list into a spreadsheet  from the Analyze function. A link from the Refine result list to the Analyze function makes switching from the former to the latter very swift. The extra beauty of the Analyze function in the new version is that the former limit of 2,000 items was raised to 100,000. Scopus does not seem to have any limit in its very similar function (which is triggered automatically), and is amazingly fast, but it analyzes the frequency of values of  a predefined set of fields (journals, authors, publication years, document types and subject categories), while WoK lets the user choose the options, including for example, the creation of descriptor frequency list of a result set. Those who do bibliometric studies will appreciate this enhanced feature.

I wish the 500 item limit for the export process would have been also increased. True, it is not needed by the average users, but many researchers would like to download all the results of a search in one fell swoop for further analysis, instead of doing it in 500-item increments, emptying the export buffer, and going through the necessary steps in each cycle.

Then again, WoK is the only system which exports the so important citedness count (the “cited by n documents” value) in a tab-delimited format, which can be then imported into a spreadsheet. Scopus offers that option now through a special edition of the spiffy RefWorks software, which is a godsend for those who need to analyze Scopus results but not in its standard export version. Interestingly, the fairly new Citation Tracker option  in Scopus was born with the option to offer exporting its result in a spreadsheet compatible format – a nifty idea.

WoK also raised the limit of the sort operation from 500 records to 100,000 (although I have not seen it documented on the site). Beyond Scopus, WoK is the only software which allows sorting of the results by citedness score, which is so important in selecting the most cited and hence likely to be the most relevant documents in a topic. In PsycINFO the results must be interpreted carefully, as there are many book records, and books (especially ones dealing with methods and procedures) which are very heavily cited and may drown out highly cited journal and conference records.

No system offers an option which I have been pontificating about for years, apparently in vain. I hoped that the new version of WoK would offer the option to see the adjusted citedness count which takes into account the age of the document cited. It is obvious that 80 citations garnered in 4 years suggest a relatively more cited items than the one which was cited 80 times in 10 years. The adjusted citation count per year of 20 versus 8 would immediately show this. Of course one can calculate this in a spreadsheet created from the search results, but the instant display while searching and sorting results would be a far superior alternative. I make my call for this feature once again.

Maybe such a feature will be added to WoK in the next release, along with some other enhancements, such as the maximum number of records that can be exported in one step, or the display of the citedness count directly in the list of cited references of a specific article or book as done by Scopus and CSA Illumina.

Overall, the WoK version of PsycINFO is a very good  choice, especially for those users who also have access to the WoS database. The latter can very well complement PsycINFO by showing cited references for about three times as many articles from psychology journals covered but not enhanced by PsycINFO. Conversely, PsycINFO can provide records for books, book chapters, conference proceedings and papers not available in WoS, and show abstracts for about four or five  times as many psychology journal articles than WoS, which has abstract only if one appeared in the original article.  There are additional advantages of having 3rd party databases not only aggregated but nicely integrated by WoK, which brings  out the synergy of searching multiple databases in one fell swoop.