Go to the U of M home page

Friday, May 25, 2012

Pre-Demographic Results

When we began work on the Library Data and Student Success project, we knew that we'd need to combine the data we collected with demographic and performance data tracked by the Office of Institutional Research (OIR). But even before we matched students with their library use, we were having fun with the numbers.

To recap, we collected U of M Internet IDs from 13 different service and resource areas during the Fall 2011 semester. Each of these 13 datasets has its own set of caveats regarding exactly what we can capture. For example, if a librarian has an instruction session with a particular class, anyone registered for that class is counted as receiving Course-Integrated Librarian Instruction. We have no way of knowing which students were in the room that day, so we're most likely over-counting in that area. On the flip side, reference staff don't typically log the patron's Internet ID in an in-person reference transaction, and the ID is sometimes not included even in an online transaction, so reference interactions are most likely under-counted.

With those caveats in mind, we set about doing what analysis we could while we waited for the OIR analysts to perform their magic. We added each of the 13 datasets to a Microsoft Access database as a table. Each table consisted of a list of Internet IDs. Using just these tables, we were able to determine the answers to questions like these:

  • How many individuals interacted with the Libraries in any measurable way?

  • How many interactions of each type did each individual have?

  • What was the total number of interactions of each type?

  • How many different types of interactions did an individual have with the Libraries?

  • How many individuals who did one thing (were registered for a course with course-integrated library instruction, for example) also did another (checked out a book)?

These questions were addressed and answered in aggregate, but having the Internet ID in each log allowed us to separate individuals from interactions. Here are some of the tidbits we were able to calculate:

  • 61,195 individuals used the Libraries in some measurable way in Fall 2011.

  • 10,455 people accessed an e-book, and 21,993 checked out or renewed at least one item.

  • 38,328 people used a database, and 30,105 accessed an e-journal.

  • 23,807 people used only one of the 13 service and resource areas. 2,774 used six or more.

  • 47,197 people used some type of digital resource (database, website, e-journal, or e-book) for a total of 1,110,727 digital interactions.

Tuesday, May 1, 2012

A word about privacy

Last Friday, April 27, our Library Data and Student Success team presented our findings at the ARLD Day at the Minnesota Landscape Arboretum. Our session was preceded by a presentation by two representatives of the ACRL Value of Academic Libraries project so the day actually flowed quite nicely. First, attendees were given a broad overview of how they can start measuring the value of their libraries, and then our presentation highlighted a tangible example of a library doing just that.

Our presentation went well, but through questions and other attendee comments it became clear that privacy implications are a big hold-up for other libraries doing this kind of work. Obviously, in order to do this kind of work libraries must track our usage in ways we maybe haven't done before, including retaining some user information regarding material check-outs and renewals. This should come as no surprise, but most libraries have strict policies in place that prohibit the retention of user data around material circulation and resource/services usage in general.

When we started this project at the University of Minnesota we quickly realized that we needed to alter our privacy practices while at the same time maintaining at least a baseline of privacy for our users that they would be comfortable with. To be crystal clear, we realized we needed to retain user information, namely the U of M Internet ID of our users for each resource or service usage. By retaining the U of M Internet ID we could then get at some of the demographic and success measures we were seeking to find out about our users.

What we also realized, however, is that we didn't need to retain exactly what our users checked out or accessed. We only needed to tie user ids to broad activities such as "checking out a book" or "using an ejournal." In other words, we are not retaining specifics about user activity. Maybe this table will help describe our efforts further:

We kept this:But not this:
Checked out X booksActual book titles
Attended X workshopsActual workshops
Reference interactionSubstance of interaction
Logged into library workstationDate, location, duration
Used an ejournalActual ejournal title

Hopefully this makes our activities clearer. To be blunt, we had to stretch our privacy policies to make this project a reality. For the first time, we are retaining some user information in order to find out 1) who our users are, 2) what types of resources they use, and 3) how this use impacts their success in the classroom. There is no way this project could have happened if we didn't tie actual users to their library activities in some way. However, we are confident that user privacy is still being maintained. Data is only being reported in the aggregate, and the data we are retaining is being kept in a secure location with the Office of Institutional Research.

So far, so good. Any questions? Let us know!