Incorporating a multiple discrete-continuous outcome in the generalized heterogeneous data model: Application to residential self-selection effects analysis in an activity time-use behavior model

This paper makes both a methodological contribution as well as an empirical contribution. From a methodological perspective, we propose a new econometric approach for the estimation of joint mixed models that include a multiple discrete choice outcome and a nominal discrete outcome, in addition to the count, binary/ordinal outcomes, and continuous outcomes considered in traditional structural equation models. These outcomes are modeled together by specifying latent underlying unobserved individual lifestyle, personality, and attitudinal factors that impact the many outcomes, and generate the jointness among the outcomes. From an empirical perspective, we analyze residential location choice, household vehicle ownership choice, as well as time-use choices, and investigate the extent of association versus causality in the effects of residential density on activity participation and mobility choices. The sample for the empirical application is drawn from a travel survey conducted in the Puget Sound Region in 2014. The results show that residential density effects on activity participation and motorized auto ownership are both associative as well as causal, emphasizing that accounting for residential self-selection effects are not simply esoteric econometric pursuits, but can have important implications for land-use policy measures that focus on neo-urbanist design.