sla2006: notes from the Math Roundtable (other speakers)
My notes on my session are below. These aren't meant to be complete, but perhaps somewhat contemplative.
Nisa reviewed her D-Lib article
comparing citation counts for JASIST from 1985 and 2000 using Scopus, WoS, and Google Scholar. She also introduced her forthcoming paper looking at citation overlaps between the three services for oncology and condensed matter physics: 2006) Three options for citation tracking: Google Scholar, Scopus and Web of Science.
Their figure 1, copied from page 11, is very powerful. Look at it for a minute:
Yes, this is not generalizable in a scientific or methodological sense, but what does it say about how we should be searching if we want to be more comprehensive in our searching? Obviously some of this is expected and perhaps desirable: do we want to include citations from patents and student papers? Conference papers and technical reports? If you're really looking at the influence of some work, then perhaps you should search across all three? Obviously it depends on the task, but this is very interesting
It was mentioned that condensed matter physics isn't the most popular category in arxiv, but it was a practical choice for other reasons.
Next, we heard from J. Parker Ladwig and Andrew J. Sommese reviewing their article on Using Cited Half-life to Adjust Download Statistics from College & Research Libraries. There are several noted weaknesses (and strengths) for any one way of measuring journal usage. The real answer is that we need to make up a weighted measuring system for the particular collection and needs of the users (note: not needs of the library, but needs of the user). Parker and Andrew are suggesting a correction to cost per download metrics based on the JCR cited half life. JCR reports for many math journals that the cited half life is >10. IOW, mathematicians are still using older journals. To me this speaks of the need to maintain older collections and to buy backfiles in this subject -- perhaps get back on jstor to collect more heavily in math. I see how this works for print subscriptions and (if they exist) online subscriptions where you buy the year and have that year in perpetuity. So you buy 2006 now, and you know that over the next 20-50 years it will be used, if the immediacy and 2 year impact factors aren't high, the overall number of times cited might add up. However, I don't believe that model is the most common for ejournals. I think it's more likely that you'll buy the backfile in a lump sum, in which case it might be easier to do the calculation OR you'll have to keep paying for the same year over and over again. Anyway, requires more thought. It was nice to have another "real" mathematician in the room although he was amazingly unwilling to speak for his whole profession :)
Finally Kris Fowler talked about whether a European group interested in km for math should have SLA representation. We all agree on the general subject, but perhaps not participating in this group. Several librarians are on their mailing list, though, so will probably continue to monitor.