THE TREATMENT OF NONPARTICIPANTS IN TRAVEL COST-ANALYSIS AND OTHER DEMAND MODELS

When estimating demand models, such as those used in travel cost analysis, the procedures used to collect data must be considered. When the statistical assumptions underlying the chosen model are inconsistent with the sample design, biased results will be produced. In this paper, three types of bias related to the improper treatment of nonparticipants are reviewed: censoring, truncation, and endogenous stratification. Methods that control for these sources of bias are considered in the context of both continuous and count models.