This paper compares three approaches to using attitudinal data to describe heterogeneous preferences for non-market goods. Two latent class models and one random parameter logit model are included. Each model makes different assumptions about the role of attitudes in the decision process. Specifically, each model assumes a different relationship between attitudes and preferences and these differences are discussed in terms of economic and social psychology theory. The three models are then used to examine individual preferences for water clarity improvements in Green Bay, Wisconsin. The results suggest that the choice of models has important implications on the quantitative results and on the nature of the preference heterogeneity, but does not affect the qualitative implications of the results. The estimates of expected WTP were nearly identical across the three models.