Including Opt-Out Options in Discrete Choice Experiments: Issues to Consider.

Area of Science:

• Health Economics
• Behavioral Economics
• Econometrics

Background:

• Discrete choice experiments (DCEs) are vital for understanding real-life choices, especially in health.
• The modeling of opt-out alternatives in DCEs, crucial for realistic scenarios, has received insufficient attention.
• Accurate modeling of opt-out behavior is essential for reliable health-related choice analysis.

Purpose of the Study:

• To demonstrate methods for accounting for various opt-out effects in choice models.
• To enhance understanding of how to model opt-out choices in health studies.
• To illustrate the consequences of incorrectly modeling opt-out effects.

Main Methods:

• Generated synthetic data on medication choice using Monte Carlo simulation.
• Considered three definitions of the opt-out alternative and four candidate models for each.
• Applied a frequentist-based multimodel inference approach with performance indicators.

Main Results:

• Misspecifying opt-out effects impacts marginal willingness to pay estimates.
• Incorrect opt-out modeling leads to inaccurate market share forecasts.
• No single model is universally best; exploring multiple models is recommended.

Conclusions:

• Properly accounting for opt-out effects is critical in discrete choice experiment modeling.
• Researchers should evaluate multiple models to assess different hypotheses of opt-out behavior.
• The study provides practical recommendations and R code for DCE practitioners.