A LATENT CLASS BINOMIAL LOGIT METHODOLOGY FOR THE ANALYSIS OF PAIRED-COMPARISON CHOICE DATA - AN APPLICATION REINVESTIGATING THE DETERMINANTS OF PERCEIVED RISK

A latent class model for identifying classes of subjects in paired com parison choice experiments is developed. The model simultaneously esti mates a probabilistic classification of subjects and the logit models' coefficients relating characteristics of objects to choices for each respective group among two alternatives in paired comparison experimen ts. A modest Monte Carlo analysis of algorithm performance is presente d. The proposed model is illustrated with empirical data from a consum er psychology experiment that examines the determinants of perceived c onsumer risk. The predictive validity of the method is assessed and co mpared to that of several other procedures. The sensitivity of the met hod to (randomly) eliminate comparisons, which is important in view of reducing respondent fatigue in the task, is investigated.