9. Dealing with Heterogeneity I: The Latent Class Logit Model

The chapter delves into the complex issue of taste variation in discrete choice analysis, focusing on the Latent Class Logit Model as a means to account for systematic variations in utility across different population segments. It begins by discussing the limitations of traditional models like the logit model, which can lead to inappropriate substitution patterns due to residual correlation. The Latent Class Logit Model introduces a hierarchical decision-making structure, allowing for more flexible substitution patterns. The chapter uses a real-world dataset on transportation choices in Sierra Leone to illustrate the application of this model. It walks readers through the estimation process, highlighting the importance of understanding unobserved population segments in decision-making. The chapter concludes by emphasizing the model's value in providing behavioral insights and its role as a precursor to more advanced mixed logit models.