What support for gathering usage data do you have in your current CMS? What is critical to your institution as you consider moving to Sakai?
Blackboard has some built-in functionality for administrators to look at usage data. Victor Maijer at University of Amsterdam is doing some sophisticated cluster analysis using this data to characterize different types of sites that instructors have there. Other schools using Blackboard were Arizona State University and Virginia Tech.
OnCourse at Indiana University provided usage data for instructors regarding their own course sites, in addition to standard web server access reports.
More regarding usage data at University of Amsterdam:
Currently using Blackboard 6.0.
Victor looks up role per course based on enrollment in the course, not by going back to the user directory.
So you're an instructor if you're an instructor in the site.
He is able to differentiate between student and instructor activity based on role above and where the activity is logged in Blackboard. (The Blackboard control panel is instructor activity; the main left nav is usually student activity). So you can distinguish between administrative activity in the course and just reading activity.
He logs the "static" data - a value at a particular point in time. So he counts the number of content objects. The clusters of course types are based on the mean values for number of content objects. In his text reports, columns define object types and each row is a course site's data with numeric values. For example -
resource, assignment-test, survey, discussion
21, 2, 0, 3
means there are 21 resources, 2 assignment-tests, 0 surveys, and 3 discussions.
He did log linear analysis to tweak the cluster definitions. So "this type of site" is defined by "about this number of resources, about this number of announcements, etc." and he can use this data to track changes in the ways course sites are used over time.
He defines "dynamic" data as the time series data.
Historical data is stored in same oracle instance as producton, but in a different table area. Data from the last three months is stored in production area that's actually written to. The historical data is later exported to another machine to work with offline (where he does the clustering analysis).
Eventually we could look at creating sakai tools to do the clustering, not just going offline to another tool.
Amsterdam have a client tool they use with Blackboard to pull out this information. Maybe this could be applied to Sakai.
Powered by a free Atlassian Confluence Open Source Project License granted to Apereo Foundation. Evaluate Confluence today.