Dear David, I tried your formulation and it is OK with small amount of replies. When I went to mass discussion out of the university forums, the d value is very high and the interpretation can't tell anything (what is d=13.84? how much this value is bigger than 2.4?). Additionally, Rafaeli and Jones proposed a hierarchy of interactivity (e.g. http://gsb.haifa.ac.il/~sheizaf/publications/finalGROUP99Submission.htm) that takes into account more variables. It seems relevant to your paper. Avner Caspi SOHAM The Open University of Israel 16 Klauzner St. Tel Aviv Israel -----Original Message----- From: David Wiley [mailto:dw2@opencontent.org] Sent: Thursday, August 08, 2002 7:41 PM To: air-l@aoir.org Cc: david.wiley@usu.edu Subject: [Air-l] Abstract: Measure of Discussion Activity in Threaded Discussion Spaces I've been lurking on this list long enough, I suppose. David Wiley, Dept. of Instructional Technology, Utah State University. How do you do? This paper in progress describes a manner of calculating a measure of discussion activity (*not* quality) in threaded discussion spaces. Based on an operationalization of discussion activity in which deeper reply levels suggest more intense discussion activity, a primitive formula for calculating mean reply depths for discussion archives is presented, together with methods for adjusting the measure to account for misthreaded messages and non-discussive messages in the archive. Sample calculations are shown and discussed for six month archives of two mailing lists relating to open source software. http://wiley.ed.usu.edu/docs/discussion08.pdf I would love to receive any comments anyone has on this work, especially pointers to related work. I would also be happy to share the software we used if people are interested. It currently runs the calculation on Mailman mailing list archives (and, incidentally, converts these archives into either edge lists or adjacency matrices for other kinds of analysis), though we're expanding it to handle webboards like Slashdot and kuro5hin. Theoretically it could be extended to process any archive with a consistent presentation format, as long as you aren't afraid of regular expressions <g>. Anyway, pleased to meet you. Here's to more AoIR fun, David _______________________________________________ Air-l mailing list Air-l@aoir.org http://www.aoir.org/mailman/listinfo/air-l