Comparing the Conditional Logit Estimates and True Parameters under Preference Heterogeneity: A Simulated Discrete Choice Experiment

Health preference research (HPR) is the subfield of health economics dedicated to understanding value and health-related objects using observational or experimental methods. In a discrete choice experiment (DCE), utility in set may differ systematically between persons due interpersonal heterogeneity (e.g., brand-name medication, generic no medication). To allow for heterogeneity, probabilities be described logit functions with fixed individual-specific parameters. However, practice, study team ignore preferences estimate conditional (CL) model. this simulation study, we examine effects omitted variance correlations (i.e., heterogeneity) parameters on estimation coefficients, willingness pay (WTP), predictions. The simulated DCE results show that CL estimates have been biased depending structure used data generation process. We also found these biases coefficients led substantial difference true estimated WTP up 20%). further were similar each other was less than 0.08) regardless underlying structure. imply that, under from their means, differences can substantive estimates. More specifically, underestimated failure recognize bias HPR indirectly underestimates treatment, substantially reducing quality care. These findings important implications because remains widely practice.