Thursday, May 08, 2008

User types and cluster analysis - Part 1

I am now approaching the last three months of my research contract on OpenLearn although I am sure that the papers relating to this will spill over into whatever I may be doing in the future.

I have a lot of data still to analyse - particularly on the qualitative side - which always takes longer than the quantitative.

Recently I have performed some Elementary Linkage Analysis on a couple of the questions in the survey. This is a form of cluster analysis that might be useful in identifying types or categories within the data. Some of this work has been reported by me and Patrick in a paper which I will be presenting at ED-MEDIA 08. In it I identify two clusters based on a question asking users about what types of facilities they would like to see in OpenLearn.

One of these clusters we have labelled as 'volunteer students'. These are users that represent more traditional types of study and highly value the content. They would like to see interactive content and a large choice of content. They would like to see more assessment, and Q and A sessions with experts. This type is likely to represent the largest proportion of users.

We have identified the other type of user as 'social learners' but it is also likely to include a small proportion of educators interested in using the resources of OpenLearn in their own teaching. This cluster might be thought of in terms of two sub-clusters; media, and social aspects in relation to learning.

Many may flag social aspects of learning but these may not be the priority for many users, who desire a traditional content based resource but perhaps to include interaction with more knowledgeable others achieved by way of Q and A with experts. It is possible that OERs such as OpenLearn could occasionally invite an expert to give a talk, say via FlashMeeting, and invite responses from users, perhaps via FlashMeeting or a forum.

One of the problems of any form of cluster analysis is the challenge of finding a label to name and encapsulate the nature of any clusters emerging.

An earlier draft of the paper can be found at: