Handling Excess Zeros in Count Models for Recreation Demand Analysis Without Apology

Abstract

Hurdle and zero-inflated models are the two foremost methods used to deal with excess zeros. However, their reliance on the nonparticipation assumption is a drawback when applied to recreation demand analysis because there is not a theoretical framework convincingly explaining presence of non-participants. This paper discusses how latent class count models represent a theoretically consistent alternative to handle excess zeros. The theoretical model behind a latent class model does not require the non-participation assumption. Instead, excess zeros is explained as the accumulation of Corner solutions from individuals belonging to different classes.