The Estimation of Salience Weights From Similarities Choice Data in Multidimensional Scaling With Application to The Assessment of Industrial Supervisors

82 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1982.A complete methodology with its conceptual foundation, intended to establish multidimensional scaling as an assessment and behavior prediction tool for an individual subject, is presented.Procedures to estimate the salience weights of the weighted Euclidean distance model, from an individual's similarity choices with random error, assuming the existence of a known group stimulus space, are presented. One method assumes that the individual weights are independent normally distributed random variables and employs a maximum likelihood approach for the estimation of the mean weights. A second method makes no assumption about the nature or the source of the random error and calculates weights estimates that minimize the number of discrepancies between the data and the model. Simulation experiments establish the accuracy and robustness of these estimation methods.A procedure is presented for the creation of stimuli with built-in attributes relevant to the industrial supervisory domain, together with a method for presenting the stimuli to the individual subjects. Choice similarity data collected from industrial supervisors about the created stimuli provide evidence about the potential validity of this methodology for individual assessment and prediction.U of I OnlyRestricted to the U of I community idenfinitely during batch ingest of legacy ETD