A comparison of mixed logit and latent class models to estimate market segments for seafood faced with ocean acidification

Abstract

Abstract This study uses a choice experiment to characterize market segments (consumer preferences heterogeneity) based on three attributes of seafood (mussels) that are affected by ocean acidification: shell appearance, meat color, and nutritional composition. Using a sample of 1,257 individuals from two main cities in Chile, we estimate both the Mixed Logit model and the Latent Class model. We use the individual-specific posterior (ISP) parameters’ distribution to categorize consumers’ heterogeneity based on the signs and intensity (i.e., like or dislike dimension) of these ISPs. We compare the pattern of preferences and whether people are classified within the same preference pattern in both models. In general, we observed that the models identify a different number of segments with various patterns of preferences. Moreover, the models classify the same people into different groups. Since the segmentation is sensitive to the chosen model, we discuss strengths, inconsistencies, biases, and best practices regarding methodological approaches to establishing market segments in choice experiments and future ocean acidification conditions.