I was impressed by a project at Argosy University to standardize an assessment of all their courses, matching course objectives to program outcomes, designing assessments that are a natural part of the class, but also useful for determining whether or not students are meeting the objectives. This is pretty much what I think the U of M is trying to do with its Student Learning Outcomes. The difference is that Argosy is a for-profit institution with no tenure and standardized courses. Whatever one may think of the quality of their courses, I'm really impressed by what they've done in terms of institutional assessment. That's probably one way in which standard colleges and universities will never be able to compete with for-profits--they can move fast.
There was a presentation on e-Coaching from some Ohio State folks. Looking backwards (20/20 hindsight), their problem and solution was obvious. The challenge was teaching students to interact productively in chats in online courses, since students do not naturally interact productively, nor do they get better at it over the course of a class. From the research literature on cooperative grouping, this makes sense. Most students do not know how to work productively in groups unless they get some kind of training. What type of training helps with this is well-known. After implementing coaching on group work with their students, there was an improvement. Another illustration of the non-triviality of transfer.
Random note: Several of the sessions that I went to during this conference seemed more like sales pitches than conference presentations. As I noted before, it really depends on the role of the person giving the presentation.