Integration of Choice and Latent Variable Models

摘要: This paper presents a general methodology and framework for including latent variables—in particular, attitudes perceptions—in choice models. is something that has long been deemed necessary by behavioral researchers, but often either ignored in statistical models, introduced less than optimal ways (e.g., sequential estimation of variable model then model, which produces inconsistent estimates), or narrowly defined structure. The focused on the use psychometric data to explicitly perceptions their influences choices. requires an integrated multi-equation consisting discrete model’s structural measurement equations. estimated simultaneously using maximum likelihood estimator, function includes complex multi-dimensional integrals. applicable any situation one modeling behavior (with type combination data) where (1) there are important variables hypothesized influence (2) exist indicators responses survey questions) variables. Three applications provide examples demonstrate flexibility approach, resulting gain explanatory power, improved specification