Hybrid discrete choice models: Gained insights versus increasing effort

Area of Science:

• Discrete choice modeling
• Econometrics
• Psychological modeling

Background:

• Standard discrete choice models lack psychological depth.
• Hybrid choice models integrate latent psychological variables.
• Estimating hybrid models can be computationally intensive.

Purpose of the Study:

• Compare hybrid choice models with random parameter logit models.
• Evaluate the trade-offs between model complexity and performance.
• Provide guidance on selecting appropriate discrete choice models.

Main Methods:

• Comparative analysis of hybrid choice models and random parameter logit.
• Assessment of model efficiency and insights gained.
• Application of principles of scientific inference for recommendations.

Main Results:

• Hybrid models demonstrate increased efficiency due to additional information.
• The choice between models depends on analytical objectives.
• Hybrid models excel at disentangling preference heterogeneity.

Conclusions:

• Hybrid choice models are preferable for analyzing preference heterogeneity.
• Random parameter logit models may be superior for predictive accuracy.
• Recommendations are provided for the appropriate application of hybrid choice models.