A reaction to the ELI Brief, "Learning Analytics: Moving from Concept to Practice"

EDUCAUSE Learning Initiative has just issued a new brief, Learning Analytics: Moving from Concept to Practice. It is a synthesis of discussions at the Learning Analytics and Knowledge Conference (LAK12) and the ELI 2012 Spring Focus Session.  Here are some reactions from an academic librarian:

Learning analytics systems are built around assumptions about the variables that predict and indicate academic success of students.

At academic institutions using learning analytics, one of the most important decisions is what pieces of information about a student will predict his/her success or indicate that he/she is succeeding? If a student's high school GPA will predict their performance in their first year of college, then we need to feed that information into the system and use it in our predictive models.  If the number of times a student eats in the cafeteria in week two of the semester is unrelated to academic success, then we don't need to get data from the cafeteria.  But what if we don't yet know--because we had no way of mining that data until now-- what variables are truly indicative or predictive of academic success?  It would seem that getting as much data as possible into your system, and then mining it, would be the way to go.  If library usage is correlated to academic success, then we need to put it into the system, but what if we don't really know yet that it is correlated? Then, it would seem that mining library data as part of learning analytics is the way to prove this. 

Visualization tools in learning analytics make the data understandable to users of the system, including students and faculty.

However, Santos and Duval of Katholieke Universiteit Leuven report that some students said they didn't like other students being able to see their activity on the analytics dashboard -- in the cases where each individual student's effort is compared with others in the course to benchmark individual effort. A potential intellectual chilling effect? A violation of privacy? This point is related to library values regarding user privacy and confidentiality.

Learning analytics, in the end, are only as good as the followup. I concur. 

If institutions do not act on the information they gather from learning analytics, then it is simply surveillance and not truly related to teaching and learning. Perhaps this is where libraries can best be involved in learning analytics: helping at-risk students with learning interventions. More on this later.