Combining and comparing consumers' stated preference ratings and choice responses

In this study we develop and test an econometric model for combining choice and preference ratings data collected from the same set of individuals. Choice data are modeled using a multinomial logit framework, while preference data are modeled using an ordered response equation. Individual heterogeneity is allowed for via random coefficients providing a link between the choice and ratings data. Parameters are estimated by Simulated Maximum Likelihood. An application of the model to consumer yoghurt choice in The Netherlands found that ratings based preference estimates differ significantly from choice based estimates, but the correlation between random coefficients driving the two is very strong