Testing Probabilistic Models of Choice using Column Generation.

Area of Science:

• Behavioral Economics
• Computational Social Science
• Decision Theory

Background:

• Random preference models analyze probabilistic choice, often using linear or weak orders.
• Evaluating choice consistency with these models presents significant computational challenges.
• Current limitations restrict the scale of empirical research in behavioral economics.

Purpose of the Study:

• To address computational difficulties in testing random preference models.
• To develop efficient algorithms for evaluating choice consistency.
• To enable the analysis of larger datasets in probabilistic choice research.

Main Methods:

• Proposed a family of column generation based algorithms.
• Implemented and evaluated algorithms on diverse datasets.
• Focused on computational efficiency and scalability.

Main Results:

• Observed substantial improvements in computation time.
• Demonstrated the ability to efficiently test larger datasets than previously feasible.
• Validated the effectiveness of the proposed algorithms.

Conclusions:

• The developed algorithms overcome previous computational limitations.
• Efficient testing of random preference models is now possible for larger empirical studies.
• This facilitates broader and more scalable research in behavioral economics.