Modeling preference heterogeneity in stated choice data: an analysis for public goods generated by agriculture

Abstract

Stated choice models based on the random utility framework are becoming increasingly popular in the applied economics literature. The need to account for respondents' preference heterogeneity in such models has motivated researchers in agricultural, environmental, health, and transport economics to apply random parameter logit and latent class models. In most of the published literature these models incorporate heterogeneity in preferences through the systematic component of utility. An alternative approach is to investigate heterogeneity through the random component of utility, and covariance heterogeneity models are one means of doing this. In this article we compare these alternative ways of incorporating preference heterogeneity in stated choice models and evaluate the sensitivity of estimated welfare measures to which approach is selected. We find that a latent class approach fits our data best, but all the models perform well in terms of out-of-sample predictions. Finally, we discuss what criteria a researcher can use to decide which approach is most appropriate for a given data set

Optimal management of wetlands: Quantifying trade‐offs between flood risks, recreation, and biodiversity conservation

“…The Ben-Akiva and Swait test results rejects the null hypothesis that the RPL and Cov Het models are the preferred specification. Similar to the work by Birol et al [2006b] and Colombo et al [2009], preference heterogeneity in this specific data set is better explained at the segment level that at the individual level.…”

“…In this paper LCM is used to understand heterogeneity in the population, rather than the RPL model, because the aim was to recommend targeted policies for specific population groups that may derive varying levels of utility from the selected wetland management program attributes. For a comparison of methods of representing heterogeneity in choice modeling, see Colombo et al [2009].…”

Motivations matter: Behavioural determinants of preferences for remote and unfamiliar environmental goods

“…An alternative model for exploring heterogeneity in preferences and WTP across respondents is the latent class (LC) model (Pacifico and Yoo 2013, Colombo et al 2009, Scarpa and Thiene 2005, Boxall and Adamowicz 2002. This model estimates discrete sets of coefficients , which are indexed over classes .…”

The design of agri-environmental schemes: Farmers’ preferences in southern Spain

“…Within the class, choice probabilities are assumed to be generated by the multinomial logit model. The probability (P) of an individual n, who makes a sequence of choices (y1, y2,… yT) among a particular set of alternatives J, to belong to class s is given by the following common formulation (Colombo et al, 2009):…”