An interactive preference ordering model and its monte carlo evaluation

A computer-interactive preference ordering procedure that is based on assumed transitivity of relations is described. The procedure includes features that are designed to make it robust in the presence of inconsistency. With errorless data, the method will find the underlying order in close to log₂ n! comparisons. Two simulation studies are described where the responses are assumed to follow the Thurstone Case V model. It was found that the method produces orders that are more valid than does a considerably larger preselected subset of pairs.This research was supported in part by Public Health Service Grant MH 16474. It was part of a larger project, and major contributions were made by Jerald F. Kehoe and Rex S. Green.