Some statistical models and computational methods that may be useful for cognitively-relevant assessment
Brian W. Junker · 1999
On leave at Learning Research and Development Center (LRDC), 3939 O’Hara Street, University of Pittsburgh, Pittsburgh PA 15260. I would like to thank a number of people, both for general conversations on cognitively relevant assessment models, and in some cases also for specific and copious comments on an earlier draft of this report. These include Gail Baxter, Pam Beck, Lou DiBello, Al Corbett, Karen Draney, Bob Glaser, Ken Koedinger, Bob Mislevy, Peter Pirolli, Lauren Resnick, Steve Roudenbush, Bill Stout, and Mark Wilson. In addition, I was very fortunate to be an occasional visitor to a graduate seminar on intelligent tutoring systems at LRDC, led by Kurt VanLehn, during the Fall 1999 semester. Two unpublished manuscripts, Louis Roussos’ Summary and review of cognitive diagnosis models and Mislevy, Steinberg and Almond’s On the design of complex assessments, were very influential as I began the task of writing this report; I also appreciated and benefitted from the comments and reactions of members of the Committee on the Foundations of Assessment when I presented a draft of this report to them on October 1, 1999. Many of the people people listed here will be surprised to see their names at all, and others will wonder why they expended so much effort on communicating their views to me, given the small effect it seems to have had on this report. Nevertheless, any special insights contained in this report are due to the influences of these people, and the many errors that remain are mine. Finally I would like to thank the NRC staff, including Naomi Chudowski, Jane Phillips, and John Shepard, for their patience with my on-deadline (and usually past-deadline) writing habits.