An on-site versus a household survey approach to modelling the demand for recreational angling: Do welfare estimates differ?

This paper compares recreational fishing travel cost demand modelling results from an on-site angler intercept survey to results from a household survey where the respondents represent the same underlying population of interest. We employed a Poisson and negative binomial count data model with and without the econometric corrections for the on-site sampling issues of endogenous stratification and truncation as the onsite modelling approach and use Poisson and negative binomial count data hurdle specifications to control for excess zeros in the household modelling approach. We find that welfare estimates differ substantially across the two samples and argue that the underlying samples may represent two different types of anglers.