Big Deals are good deals even based on high use titles
Sarah Glasser (2018) Judging big deals—take two, Journal of Electronic Resources Librarianship, 30:1, 27-33, DOI: 10.1080/1941126X.2018.1443904 (https://www.tandfonline.com/doi/full/10.1080/1941126X.2018.1443904)
Abstract: A follow-up to an earlier article published in this same journal in 2013, this article repeats an analysis of five Big Deal journal packages to which Hofstra University's Axinn Library subscribes. Improvements over the last study include more years of data, the removal of open access and archival content from the calculations, and a subject analysis. Results confirm the original findings that four of the Big Deals are good deals, but expand on these outcomes with a discussion of the subject breakdown of each package, the cost of individual title subscriptions, and the dangers of overreliance on usage statistics.
Big deal advantages [expanded access, increased efficiency (fewer licenses and access platforms to manage), and greater consistency for users (fewer platforms to become familiar with)] and disadvantages [runaway costs and lack of flexibility (cancellation restrictions)] still exist (as per lit review).
My thought: Pros and cons don’t exist separately. the disadvantage of “runaway costs” for example will eventually impact “expanded access” through not being able to afford either the deal in question or others.
Big deal analysis should focus on most used titles because determining whether something is a good deal depends on the content being used.
My thought: On the contrary, it should be based on what would have been subscribed to without the big deal. This may ideally be the same thing but you can’t assume that (1) you would have chosen the exact titles that ended up having the highest use in a package and (2) that having a big deal or not doesn’t affect use of specific titles (not having a specific journal title doesn’t always mean that the researcher simply goes without: it’s much more complicated than that).
Good idea to remove OA, archived, and separately acquired titles/volumes from the usage data.
No consistent cost-per-use, or percentage of high use titles.
Subscribing to titles with 15+ downloads/year would cost significantly more than the Big Deal cost for all but one of the five packages checked. Subscribing to titles with 50+ would still cost more than two of the five Big Deals. Subscribing to titles with 100+ would generate savings for all but one and would remove one of the deals (having no 100+ titles). The conclusion of this is that the package that would see savings after subscribing to only the 15+ titles may not be a good deal but the other’s, especially the one that remained cheaper after limiting to 100+ titles were good deals.
The two best big deals also happened to be the largest deals, with over 2k titles each. They were also primarily STEM with little to no Humanities and only a small number of Social Sciences ones.
My thought: I think this shows that either STEM strong vendors can afford to be more generous, giving discounts for big deals to get their titles out, what with individual titles being more expensive OR that use for STEM is by its nature more suited to seeming successful in this kind of analysis (e.g. electronic and journal content used more heavily or use tends to center on fewer popular titles instead of spread more evenly).
Before cancelling, the library would have to be certain that particular departments and areas of study are not unduly taxed and left with insufficient resources. (Don’t just use usage data to make cancellation decisions.)
Many things can affect use data, rightly or wrongly.
This analysis was used for info only: no decisions were made based on it.
My thought: There was no consideration of FTE or faculty numbers in this analysis, nor was there any mention of relative sizes of departments or user groups. I’m not sure comparing titles with say 15+ uses in an average year across the board is fair if, say there are simply more users of one kind of title than another. This would certainly be true when comparing two deals that differ in content at the high subject level (as they’ve done) but also within subjects, say between medicine and physics. Again, use is more complicated and therefore assessing “enough” use won’t be the same from deal to deal. What’s needed is a complex model of use that can use numbers and types of user, expected behaviours, historical trends, and maybe even use as a reaction to availability or not. Not easy. Perhaps impossible.
