Skip to main content
Open Access Publications from the University of California

Heterogeneous Preferences for Water Quality: A Finite Mixture Model of Beach Recreation in Southern California


This paper uses a finite mixture logit (FML) model to investigate the heterogeneity of preferences of beach users for water quality at beaches in Southern California. The results are compared with conventional approaches based conditional logit (CL) and random parameters logit (RPL). The FML approach captures variation in preferences by modeling individual recreator preferences as a mixture of several distinct preference groups, where group membership is a function of individual characteristic and seasonal variables. The FML parameter estimates are used to calculate welfare measures for improvements in beach quality through a reduction of water pollution. These bound the traditional CL and RPL mean welfare estimates, and have the advantage of highlighting the distribution of the population sample's preferences. The data indicate the existence of four representative preference groups. As a result, willingness to pay measures for improvements in water quality can be weighted across individuals to calculate the distribution of individual welfare measures.

One of the groups is people who go to the beach with small children. An interesting finding is that these people have a lower mean WTP for improving water quality than people who go without a small child. This may well be an example of cognitive dissonance: parents find they go to the beach more often than others who don't have small children, since that keeps the children occupied and happy, and they adapt their perception of the water quality to be consistent with their behavior.

Previous environmental and resource economic applications of the FML have been limited to applications with small choice sets (6) and group membership variables (4). This paper extends the FML model through the estimation of a large (51) choice set with 9 membership variables. This application is the first to incorporate seasonal variables into the group membership function to capture seasonal heterogeneity.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View