Quantal Choice Models and Conjoint Analysis: Diagnostic and Predictive Comparisons

Quantal choice models are used to link a dichotomous or multichotomous criterion variable, measuring actual choice, with predictor variables which define attributes of alternatives. Usually, individual differences in parameter estimates are allowed for only through measures of demographic, socioeconomic or other variables. In conjoint analysis, stated preferences for hypothetical objects are used to estimate the influence of object characteristics. The parameters for such preference models can be estimated at the individual level if a sufficiently large number of objects is used. In this paper we discuss the suitability of (quantal) choice and preference (conjoint analysis) models in marketing contexts. We also present empirical results to compare diagnostics and predictions for these approaches, using data on consumers’ perceptions, preferences and choices of transportation alternatives.