Stacy Bruss, Reference Librarian at the National Institute of Science and Technology (NIST), described how she used bibliometric analysis in the NIST library to develop the collection. (The conference website contains supplementary material with details of Bruss’s methods and sample outputs.)
The annual collection impact study examines cited references and NIST publication authors to measure the impact of the collection. Doing this manually used to take 5 librarians a month. Now, using bibliometric analyses, it takes Bruss only 16 hours. Here are the key fields measured (for journals–books are not a focus of the NIST collection):
Here are the study findings and outcomes:
Outlying data were checked to see why journals are not in the collection; for example, a highly cited article on astrophysics by a NIST reseracher was published in a journal that was not in the library’s collection. But astrophysics is not a significant NIST area of research, so the journal was not added.
It was a surprise to realize how much historical journals (over 25 years old) continue to be used in current research.
The biannual research study is done every other year. It looks at all NIST indexed journals over the past two years and correlates citations with journal subscriptions.
Here are the study output and findings:
Many journals are obtained under contract, so changes sometimes cannot be made. It is important to note that this study prioritized journals based on downloads, publications, and citations, but not costs.
Some journals are designated as “core journals”, and those not subscribed to are often purchased if they have many downloads.
Collection insights show where most of the downloads come from; the above example shows which publisher’s journals are cited the most. Journals from the top publisher are cited more than 3 times more than those of the one in second place. This study gives evidence-based data for subscription decisions. Formerly, printed surveys were sent to the researchers, but they never suggested dropping anything and just added titles! These data showed that the survey was not representative of actual use, so it was dropped, which was supported by the researchers. The advantage of a survey like this is that budget changes can be made quickly.
The methodology for an archive journal holdings study. has been developed, but the study has not yet been conducted. Archive journals take up 20,500 feet of shelving in the NIST library. It is difficult for researchers to shift their work practices if they must come to the library to get an article in an archive journal or through ILL.
Here are the anticipated outcomes of the study:
The NIST collection contains materials that may not be available anywhere else in the world; they must continue to be held somewhere. The archive journal holdings study will identify “nice to have” items as well as “must have” ones.
Below are the types of tools and resources used in these bibliometric studies. Note that Google Scholar does not work very well in working with bibliometric data, and Excel 2013 has a relational database built into it.
Bruss concluded that bibliometrics has come a long way and is finding new uses in libraries; they are “not your mother’s bibliometrics any more”!