Searching the Literature

by Dr. Brewer, ScienceDocs Editor and Writer

“Bernard of Chartres used to compare us to dwarfs perched on the shoulders of giants. He pointed out that we see more and farther than our predecessors, not because we have keener vision or greater height, but because we are lifted up and borne aloft on their gigantic stature.” – John of Salisbury, 1159

Science ultimately is a collective enterprise.  We learn from our predecessors and peers, and we seek to inform others of our discoveries.  The main way we learn from other researchers is to search the literature. 

By searching the literature, we can get or shape ideas, and find answers to our questions.  From the results of searches, we can put our own research in context and give credit to those whose work we have relied on, extended, or challenged.  Literature searches allow us to avoid reinventing the wheel and repeating others’ failures.

Searching the literature also teaches us important history lessons.  Most avenues of research ultimately become dead ends, or at least paths no longer taken in the pursuit of truth.  Few ideas or empirical results are original.  If we take these lessons to heart, we gain humility about our own and others’ research today. 

The best time to search the literature is before we carry out, or even plan, a research project.  Each new study is like a sentence in the worldwide conversation that we scientists have with each other over time.  If we want to be understood when we report our research, we must know what the conversation is about, what’s already been said, and what might be a useful addition.  Of course, we can and should search the literature at any point when we have a need, but it is essential to search the literature formally before we start new research.  Even though we learned this sequence as students, many professional researchers cease to follow it when they no longer must satisfy the requirements of their thesis or dissertation committees. 

Until about 20 years ago, the best ways to search the literature were to browse the shelves in a library, (as related material is shelved together), inspect journals’ tables of contents, find sources from references cited in articles and books (backward citation search), and use print copies of citation indexes to identify sources citing particular works (forward citation search).  These tools, although useful, were very time-consuming. 

The landscape changed in the late 1990s with the advent of online databases of journal articles, such as PubMed.  These databases focus on subsets of journals in broad academic fields.  They store information on bibliographic details, abstracts, and subject classifications, and can be searched easily. 

Then, in 2004, Google released Google Scholar, a search engine that indexes the full text of the scholarly literature – journals, books, reports, and other documents.  I believe that Google Scholar has become the single best tool for searching the literature.  It has many advantages: it is free, returns documents published in any language, enables easy forward citation searches, identifies articles related to documents returned in a search, exports citations to bibliographic software, and often provides links to free, full text versions of the documents identified.  To my knowledge, no other current scientific literature database or search engine offers full-text indexing.  That means if a search term is not in the abstract, subject classification, or bibliographic fields, but is in the text of the document itself, Google Scholar can find the document, but the other databases and search engines will not. 

In the past, Google Scholar sometimes had more gaps for relatively old material, but those gaps have been and are steadily being filled as older books and articles are digitized.  In fact, other databases and search engines are almost useless for documents published more than 50 years ago, because they do not index abstracts of the small amounts of old material they include.  In my experience, no other search engine or database covers old material nearly as well as Google Scholar. 

Compared to other databases and search engines, however, Google Scholar produces more irrelevant records in a search, although these tend to be concentrated toward the middle and end of the records returned in a search.  A few fields, such as chemistry, are not represented well due to uncooperative publishers.  I have found excellent coverage in the health and social sciences.  Indeed, as of January, 2013, Google Scholar captured 87% of the scholarly literature, while other services performed much worse. 

For almost every reason scientists search the literature, it is essential that the search be as complete as possible.  In our own studies, we wouldn’t throw away half or more of our samples that are ready for analysis.  Similarly, it would be foolish to neglect half or more of the relevant research when searching the literature.  With Google Scholar, and a little extra effort, we can obtain a fairly complete picture of the literature.  As the situation demands, we can also supplement Google Scholar with other databases and search engines, especially those that focus on particular kinds of documents that don’t always appear on the World Wide Web, such as dissertations and theses. 

Despite the remarkable advances in tools for searching the literature, over the last 25 years, I have noticed a remarkable decline in the quality and completeness of literature searches in research reports, including meta-analyses and systematic reviews.  Just as it has become very easy to search the literature well, it seems that many researchers do not attempt to do so.  I do not know the reasons for the decline, but it represents a corrosion of scientific practice and threat to the efficient cumulation of knowledge.  As George Santayana observed in 1905, “those who cannot remember the past are condemned to repeat it.”

Learn more about ScienceDocs Editor Dr. Bewer