I wanted to mention that a fellow colleague here at Woodbury has now joined the blogging community.

http://djcremer.typepad.com/transdiscipline/ 

He'll be writing about transdisciplinarity as he starts in a new position as the directory of the Institute of Transdisciplinary Studies.

As I have been writing about in this blog, my research group at Claremont Graduate University has been using Elgg to support a set of transdisciplinary courses.  We have now had the site up and being used for over a month and a half.

 

Overall, it has been a very positive experience.  We have four supported and one unsupported class using the system.  In plugging the system to the professors, we pushed each of the major elgg features (blogs, wikis, file sharing) as ways to solve class organizational and collaborative problems.  We have not emphasized the social and portfolio aspects, instead mainly focusing on classroom uses.

 

The professors have been very positive.  Two have said that they are doing things in elgg that they could not do in WebCT, something that we like to hear.  A large number of additional professors have expressed interest in using the system in future semesters.  We have not yet gotten much student feedback, but that will come at the end of the semester.

 

The biggest problem we have had so far is in speed of the site.  We will never again use godaddy as a host.  At this point, the professors are actually calling it slowdaddy.  While the support people are responsive, they have zero interest in actually solving our performance problems.  We are probably going to have to cancel our contract with them, and are working with our campus technical staff to host it locally.

 

The second major issue is usability.  Students are having problems keeping track of activities going on in their various communities.  Especially for professors, they really want a centralized place to see all of the activity in the system.  One professor has something like eight groups for one of her classes.  I put together a global RSS system, but before I was able to add-on security I came down with tendonitis and been unable to finish it.

 

The third issue has been administrative.  It took a lot of time to get all of the students’ accounts created and joined to the proper community.

 

One thing I have been very surprised about is the sheer amount of bandwidth used by the system.  This month, we have had 2 GB of bandwidth with only five classes supported.  None of these classes have really been using many files, so this is almost exclusively page views.

 

We also have a lot of unique visitors on the site.  Even though we only have around a hundred students on the site, we average around 120 unique visitors each day.  The times that these visitors log onto the site varies, with our heaviest traffic around 11 o'clock at night.  The only time the system is not being heavily used is from 2-4 in the morning.  There is definitely a weekly pattern of more activity right before each class meets, but even our slowest day has at the least half the traffic as the most active day.  We don't have many Apple users (5%), but 20% of our users are using Firefox. 

 

At the beginning of the semester, we surveyed the different classes to find some basic demographic information.  Attempting to get an early feel for it, I ran a regression analysis attempting to predict each user’s overall activity (files, blogs, comments, and wiki's) by their demographic information.  The data is mainly speculative, but does it give a few early ideas.

 

Overall, our collected demographic information and course membership data accounts for 74% of the variation in actual activity.  Obviously, the class in which people are enrolled has a major impact upon their activity.  The only demographic information charted so far that is statistically significant is a technology familiarity index.  This value was created by asking students to rate on a scale of one to three their familiarity with blogs, RSS, wikis, social networks, file sharing, and e-mailing their professors.

 

Demographic information that was not significant included a long survey from another study used to predict online achievement, an introvert/extrovert personality survey, and information about their primary language.  I found it rather ironic that our simple 6 question familiarity index was a much better predictor than the validated “predicting online achievement” survey.  With more data, introvert/extrovert and primary language variables will probably become significant.

 

Anyway, this is still very speculative, but I thought that I would share our early experiences with the community here.  We will be writing papers and conference presentations, and I will be sure to post copies online for interested parties.

Nathan