Yet another mechanism exists that, in many cases, circumvents the major problem of keyword searching by enabling you to recognize highly relevant sources whose terminology you cannot think of in advance. This mechanism is related record searching. Web of Science and Scopus provide the best starting points for this kind of inquiry, although the search technique is starting to show up in other databases as well.
In order to do a related record search you must first do a keyword search. In the Web of Science, for instance, if you find any good record that way, click on it to bring up the “Full” display of the citation (showing its abstract). The same display will present you with a column to right of the screen, usually providing you with a clickable option saying “View Related Records.” Pursuing this link will provide you with a list of all other articles in the database that have footnotes in common with your starting-point article.
These are not articles that cite each other—rather, they are articles written independently that nonetheless cite several of the same sources in their footnotes. Articles having shared footnotes are usually covering the same subject area; the important point, however, is that they may be referring to their common topic with entirely different keywords. The list of related records, generated from the starting-point article, will be displayed in ranked order—those articles having the most footnotes in common with it at the top, down through those having only a single shared reference.
In the previous chapter I mentioned the researcher looking for articles on “the economics of antiquities looting.” He found one good article having those exact words in the title, through a keyword search; then, through a citation search, he found many subsequent articles because they referenced that first article. In the same instance, however, he could find still other relevant articles through a related record search because they had multiple footnotes in common with that same initial source. Among these articles having shared footnotes were such titles as the following:
Related record searching is thus another mechanism that brings to your attention in a systematic manner relevant sources whose keywords you could never think up in advance. Further, since your search is tied to an already-discovered relevant article, that linkage provides a kind of contextual anchor that precludes widespread retrieval of thousands of articles having keywords wandering off into contexts that are irrelevant.
Another example is provided by a criminologist who was looking for articles on “statement analysis.” I had never heard of this topic before she came to the reference desk, but she explained that it has to do with detecting signs of deception in either written or oral communications. A keyword search in Web of Science will turn up a few good hits within a welter of retrievals on “financial statement analysis” (which is something very different), with additional articles that cite the few relevant ones. But a related record search, starting from those few good keyword articles, produces scores of articles having such titles as these:
What is especially useful about this search technique is that it can produce relevant hits no matter what their temporal relationship to the starting-point article—that is, the related record articles may be not just subsequent articles (as with citation search results), but also previous articles (as with footnote chasing), or even articles written in the same year. This search technique thus offers advantages over both of these other techniques for exploiting footnote references.
Further, you can “cycle” related record articles: if you find any good hits through this approach, you can then see if they lead to still other related records, because the same “View-Related Records” option is likely to show up on their own “Full” displays.
One limitation of Web of Science and Scopus coverage is that both databases “see” only scholarly journals—those having footnotes—to begin with. They do not cover the many other newsstand or commentary journals, which may well be of high quality, whose articles do not have these scholarly appendages.
I mentioned above that these two databases are no longer unique in offering related-record search capabilities. As of this writing several others (e.g., ProQuest’s Dissertations and Theses and ABI/INFORM Complete) offer this search option. Undoubtedly others will follow. Some—not all—of the initial bibliographic citations you find in these databases will provide a link that says “Documents with shared references.” Clicking there will produce a display of related citations in ranked order—i.e., those with the most shared footnotes at the top, down through those with only a single shared reference—with explicit listings of how many “references in common” there are for each. The qualification is that, at least in ProQuest databases, not every record has such a link. In Dissertations, for instance, none of the pre-1997 records have it.
Remember that whenever you find a good article on your topic, you can do several things with it to find additional conceptually related sources that may be using keywords you could not think of on your own:
All four of these search mechanisms are capable of producing results that cannot be matched by searching keywords. The advantage they offer, each in a different way, is that of laying out menus of works relevant to your topic in ways that enable you to simply recognize good sources whose terminology you would not otherwise think of. These are some of the best solutions to the problem of “what words do I type in to get the best results?”—the most serious difficulty routinely encountered in Internet searching. They show you directly relevant records whose search terms you don’t know how to ask for, and they do it in ways that assure conceptually anchored retrievals that do not have tens of thousands of hits having the right words in the wrong contexts. (Yet another search technique having comparable capabilities is that of using published subject bibliographies, to be discussed in Chapter 9.)
A repeated theme of this book is that no one way of searching does everything. And here I would re-emphasize that Internet keyword searching alone—no matter how the terms are relevance-ranked—is particularly problematic if your goal is to get an overview of the full range of literature on your topic. The same applies to “discovery” and federated searches of multiple subscription databases merged into a single pool. You simply have to be aware of the unavoidable trade-offs among search techniques so that your overall strategy can balance their respective strengths and weaknesses against each other. Each will show you different parts of “the elephant.” What you cannot see with one way of searching, you can see with the others.