Comparing individual-specific benefit estimates for public goods : finite versus continuous mixing in logit models

Multi-attribute stated preference data, derived through choice experiments, is used to investigate the consequence of a finite number of preference groups in a sample of Yorkshire Water residential customers on the conditional distributions of willingness to pay in the sample. The research focuses on public good' values, and retrieves the implicit customer specific welfare measures conditional on a sequence of four observed choices. We assess and contrast the sample evidence for the presence of a finite number of 2, 3, 4 and 5 latent preference groups (classes), and contrast these with the presence of a continuous distribution of parameter estimates using mixed logit models. The main focus is the conditional valuations in the form of marginal values for the consequence of waste water handling and treatment, namely: river water quality, area flooding by sewage, presence of odour and flies, and other water related amenities.