This paper develops and empirically evaluates alternative econometric strategies for accounting for non-participation - repeated choice of the same alternative or same type of alternative - in data sets that are typically analyzed within the repeated discrete choice framework. Random coefficient single and double hurdle variants of the repeated discrete choice model are developed and applied to stated preference data. The empirical results suggest that significant statistical improvements in fit can arise with the single and double models relative to more traditional models. However, similar to Haab and McConnell's (1996) findings in the context of count data models, these gains are diminished when the analyst accounts for unobserved heterogeneity through random coefficients.