1,965 words 1 May 2004 Online 57 ISSN: 0146-5422; Volume 28; Issue 3 English Copyright 2004 Gale Group Inc. All rights reserved.
All three reviews in this issue focus on economic literature databases. Economics has been the turf of the EconLit database for decades. In the past few years, however, there have been many projects to make at least the abstracts of substantial economic research papers freely available through the Web. One of the most successful of the international collaborative efforts is the RePEc (Research Papers in Economics) archive, which is being implemented with different features by talented economists and programmers in many countries as varied as Sweden, Russia, the U.K., and the U.S.
My first pick is IDEAS, one of the many excellent implementations of the RePEc collection of free abstracts for more than 200,000 journal articles, working papers, books, book chapters, and software. Over half of the abstracts have links to the full-text digital version. The other pick is a spectacularly well-implemented bibliometric service that delivers very informative statistics about the papers, journals, series, and authors in RePEc. The pan is the American Economic Association's EconLit database, which is widely licensed by many college libraries and research centers, but is becoming increasingly less and less state of the art.
It was not easy to choose my pick of the several excellent RePEc implementations. Considering all the features, IDEAS, by the Department of Economics at Connecticut University [ http://ideas.repec.org/], is my favorite. It is a labor of love by associate professor Christian Zimmermann, who is also the brain behind the EDIRC (Economics Departments, Institutes and Research Centers in the World) database, which contains information on about 7,770 institutions, making it a potential pick in and by itself.
Instead of trying to describe the richness of this version of RePEc, let me reproduce one of the overview statistics of IDEAS. This is the kind of information I begged for in a guest editorial on database nutrition labeling over a decade ago ("A Proposal for Database 'Nutrition and Ingredient' Labeling," DATABASE, February 1993, pp. 7-9).
The size of the database is impressive, and so is the fact that more than half of its records have abstracts. Very importantly, 56 percent of the records link to the full-text documents (although only a fraction of them are open access). Still, most users at research libraries are likely to have a subscription to the digital archives of economics journals. Since IDEAS' links are to the article level, access is very efficient.
The records are displayed in a well-structured, easy-to-scan manner, with jumps to the different sections of the record, such as the cited references, which are available for 14 percent of the records. Even more records take you to the papers that cite the item being displayed. In cases where the author is registered in EDIRC, there is a link to the author record that in turn shows which of his or her articles are open access. And all this is free, courtesy of the publishers, the participating research institutes, and the individual researchers.
The search software is ht://Dig, which I don't dig, because it does not offer exact phrase searching, field-specific searching, limiting to year range, and other useful options. This database deserves a better open access software, such as Swish-e. Luckily, Zimmermann brings the most out of ht://Dig. IDEAS has some very useful statistics, such as the top 5 percent of authors, institutions, countries, and states, ranked by the number of works in RePEc, the number of times their abstracts were viewed, papers were downloaded, and cited in other papers that are covered by RePEc and could be analyzed. It complements very nicely another free RePEc service that deserves kudos (Citations in Economics) as well as my other pick, LogEc.
The popularity of the RePEc archive and its services is superbly illustrated by the LogEc service [ http://logec. hhs.se/], run by Sune Karlsson at the Economic Research Institute of the Stockholm School of Economics, which analyzes the server logs of the participating RePEc servers. Smart graphs--I cannot do justice to them with these black and white illustrations--depict the traffic on the participating servers, the file download and abstract view activities, broken down by document type. From 1998, there have been nearly 6.8 million file downloads and nearly 46 million abstract views, excluding visits by robots. The charts clearly illustrate the importance of working papers (most of them electronic pre-prints to be published a year or two later).
Beyond these global statistics, there are dynamic tables for the past year, quarter, and month, with highly informative traffic indicators ranging from the analytic level (working paper, article, software, individual authors) to the aggregate level (journals, working paper series).
With a click of a button on the appropriate table header, you can resort the list by any of the indicators on the fly. Automatically the top 25 ranked papers, authors, journals, (series) are listed, but you can change it to the top 50 or top 500, or get the entire content ranked by one of the applicable criteria instantly.
The top-ranked journal article for January 2004 was a paper about the effect of mad cow disease on futures prices. It ranked sixth even in the quarterly ranking. Particularly interesting is that this is the first paper by an award-winning graduate student, Newton N. Paiva, which was published in the Journal of Agricultural and Applied Economics in August 2003 before the outbreak of the mad cow disease in the U.S. Talk about prescience and good timing. Another click presents the rise in popularity of the article in the selected period, which in turn leads you to other useful statistics, such as the ranking of other papers from the same journal or working paper series. It is simply awesome.
The careful interpretation and analysis of the results (which require the careful reading of the explanation at the "About" page of the site), comparing LogEc findings with Thomson-ISI's Journal Citation Reports and the Social Science Citation Index, opens up new opportunities for bibliometric research into prestige and potential impact of primary sources and authors, as well as of research trends.
EconLit, produced by the American Economic Association and available on many commercial online services, including CSA, Dialog, EBSCO, OCLC, Ovid, Science Direct, and Silver Platter, has ruled the waves in print, CD-ROM, and online formats for decades. In my first "Picks & Pans" column, published in the June/July 1995 DATABASE, I panned the database for the extent of carelessness in handling journal names while professing utmost care and commitment to quality through idle PR promises. The publicity convinced the publisher to correct the most brutal errors, although not quite all of them, as I found out when writing this column. According to EconLit, The Indian Journal of Quantitative Economis (rather than Economics) and the Populaiton Research and Policy Review (rather than Population) are viable journal titles.
I pan EconLit this time, however, for different reasons. The handling of links is disappointing. First of all, the links are often embedded in the long text of the "availability" field and are not uniformly hyperlinked by the commercial services hosting the database. CSA optimizes them; DialogClassic does not, for example. Second, the links are directed to the journal or publisher level, and never to the item. This makes it necessary for users to cut and paste the URL, often also to modify it just to get to the journal or publisher Web site. Once there, they need to find the site engine, type in a query, or drill down to the item through the browsable title index or table of contents--all the while not forgetting the title of the item and/or the volume and issue number.
All this could be spared by making the links a little smarter for instant gratification. For example, a URL like this: http://qed.econ.queensu.ca/pub/ papers/abstracts/index.html takes you to the home page for Queen's Institute for Economic Research Working Papers. From there you need a series of unintuitive steps and clicks to get to the PDF article.
This smart URL http://qed.econ.queensu.ca/pub/p apers/abstracts/download.html# 1001r would take users directly to the free document, no muss, no fuss. It may take a little longer to create such a link, but often the elements on which the links are built already appear in the record. For example, by replacing the subdirectory name and adding the report number for an actionable click with instant gratification, this URL www.econ.upenn.edu/Centers/pier/Archive/03-001.pdf would have saved the users the trials and tribulations caused by being taken to the too generic address of www.econ.upenn.edu/Centers/pier/WorkingPapers/index.htm. Try out both examples and feel the difference.
To me, EconLit's attitude is like that of a taxi driver who drops you at a busy intersection without giving directions, versus the cabbie who provides door-to-door service and even opens the door for you. EconLit could have been easily endowed with more than 300,000 smart links to full-text documents with a little extra work, whose logic and syntax would be easy to master by the committed and paid indexing and abstracting staff, as they follow a pattern within archives of working papers and publishers, and thus can be applied for thousands of records by routine.
Another reason for panning EconLit is the apparent abandonment of working papers, which are as important in economics as in physics, math, and computer science. For decades EconLit has added 1,900 and 2,500 per year, resulting in more than 41,000 records for this document type in the database. However, after 2000, the number of these records plummeted, and as of early February 2004, the trend is sadly clear: I found a mere 19 such records dated 2003. [Editor's note: An EconLit spokesperson told us the decline in working papers results from its supplier not delivering the data as promised. There is no policy decision to exclude working papers and EconLit agrees on the importance of this type of data to economic literature. EconLit is working to develop procedures to obtain working papers directly rather than through a third-party distributor. The plan going forward is to collect rather than license the data.]
College students and researchers who heed the advise to rely only on sources from well-known publishers--and not to accept candy from strangers--will find themselves more and more disadvantaged using EconLit versus those who venture to use the innovative and brainy services built around the ever-growing RePEc archive, which covers many of the same journals EconLit has been covering, also increasingly including cited references. RePEc covers far more working paper series, including actionable hotlinks to full text and abstracts even when EconLit delivers not even an abstract, as is the case with the article whose abstract was the most viewed in RePEc in January 2004 and was ranked 6th for the last quarter of all the journal articles in RePEc.
It does not increase EconLit's credibility that its database blurb claims, "Virtually all records include abstracts." Our definition of the word "virtually" may differ, but this is an outrageous assertion. Out of the 637,970 records only 237,238 had abstracts. That is 37.18 percent.
Peter Jacso [jacso@hawaii.edu] is professor of Library & Information Science at the University of Hawaii's Department of Information and Computer Sciences.Comments? E-mail letters to the editor to marydee@xmission.com.
COPYRIGHT 2004 Information Today, Inc.