Discriminating Among Probability Weighting Functions Using Adaptive Design Optimization.

Area of Science:

• Decision Theory
• Behavioral Economics
• Cognitive Psychology

Background:

• Probability weighting functions are key to modeling choices under risk in cumulative prospect theory.
• Empirically distinguishing between proposed parametric forms of these functions is challenging due to their qualitative similarities.

Purpose of the Study:

• To investigate the efficacy of adaptive design optimization for discriminating between different parametric forms of probability weighting functions.
• To identify which functional forms are most prevalent in empirical data.

Main Methods:

• Utilized simulation experiments to test the discriminative power of adaptive design optimization.
• Employed choice experiments with human participants to gather empirical data.
• Applied adaptive design optimization, a computational method for model discrimination.

Main Results:

• Simulation results confirmed that adaptive design optimization can conclusively discriminate the correct data-generating probability weighting function from alternatives.
• Empirical experiments revealed significant heterogeneity in participants' preferred functional forms.
• The Prelec-2 and Linear in Log Odds models were identified as the most common best-fitting models in the empirical data.

Conclusions:

• Adaptive design optimization is a powerful tool for model discrimination in the context of probability weighting functions.
• Individual differences in probability weighting are substantial, suggesting a need for flexible modeling approaches.
• The findings provide insights into the underlying assumptions and empirical validity of different probability weighting function models.