The authors develop a model for multiple paired comparison product tests. Tested on several data sets, the model is shown to describe the data accurately. The model provides an estimate for the frequency of "errors" in the data.
BEMMAOR, A.C. and WAGNER, U. (1996). Capturing Chance Effects in Pairwise Product Tests: A Stochastic Model for Preference Ratings. In: 20th Annual Conference on Classification, Data Analysis and Knowledge Organization, March 6-8, University of Freiburg, Germany.