Discrete choice experiments: An overview on constructing D-optimal and near-optimal choice sets

Area of Science:

• Behavioral Economics
• Experimental Design
• Survey Methodology

Background:

• Discrete choice experiments (DCEs) are vital for estimating and forecasting individual choice behavior using stated preference data.
• Effective DCEs rely on well-constructed experimental designs, often involving a limited number of choice sets and alternatives per set to enhance response efficiency.
• While algorithmic (efficient) designs are common, optimal (orthogonal) designs are still used, especially when prior population preference information is unavailable.

Purpose of the Study:

• To provide an overview of discrete choice experiment construction methods in the literature for models focusing solely on main effects.
• To compare various optimal and near-optimal design construction techniques based on their efficiency in minimizing the number of choice sets.
• To address practitioner concerns regarding the performance of different design techniques in achieving high efficiency with minimal choice sets.

Main Methods:

• A literature review was conducted to identify and categorize different construction approaches for discrete choice experiment designs.
• Comparative analysis of various optimal and near-optimal design construction methods was performed.
• The primary metric for comparison was the ability of each method to minimize the number of choice sets required for a survey.

Main Results:

• The study reviewed and compared different techniques for constructing discrete choice experiment designs, focusing on models with main effects only.
• The comparison evaluated the effectiveness of various construction methods in reducing the number of choice sets while maintaining high design efficiency.
• Key findings highlight the trade-offs between different design strategies concerning the number of choice sets and overall experimental efficiency.

Conclusions:

• The research offers insights into the performance of various discrete choice experiment design construction methods.
• Findings illuminate the optimal sample sizes necessary for conducting efficient experiments in this domain.
• This work aims to assist researchers in designing more effective discrete choice experiments by understanding the strengths of different design approaches.