Estimating Independent Locally Shifted Random Utility Models for Ranking Data

Area of Science:

• Behavioral Economics
• Psychometrics
• Marketing Science

Background:

• Conjoint experiments are widely used to understand consumer preferences.
• Estimating probabilistic ranking models often involves complex high-dimensional integrals.
• Partial rankings are valuable for efficient data collection in conjoint analysis.

Purpose of the Study:

• To develop an efficient method for estimating probabilistic ranking models in conjoint experiments.
• To extend existing approximation techniques to a broader class of random utility models.
• To facilitate the analysis of partial rankings.

Main Methods:

• Utilized approximate rather than exact ranking probabilities to avoid high-dimensional integration.
• Extended Henery's (1981) approximation technique to independent locally shifted random utility models.
• Applied the method to independent random utility models with common distributions (e.g., normal, logistic) and scale.

Main Results:

• Successfully estimated independent random utility models with common shapes and scales.
• Demonstrated the applicability of the method to partial rankings.
• Validated the approach through reanalysis of existing datasets (career and holiday preferences).

Conclusions:

• The proposed approximation method offers computational efficiency for probabilistic ranking models.
• The approach supports the estimation of a wide range of independent random utility models.
• This facilitates practical conjoint analysis by enabling efficient partial ranking data collection and analysis.