Thurstone Model: Analysing Survey Data and Ranking Stimuli – Model or Procedure?

Abstract

This study explores the suitability of the Thurstone model in preference studies through empirical validation. While the Thurstone model has been widely utilised for establishing linear orders among stimuli, its appropriateness has come under scrutiny. The research proposes a novel empirical validation test to assess the model’s applicability, particularly concerning large sample sizes. Findings suggest limitations in using the Thurstone model with extensive samples. Consequently, the paper offers insights for researchers relying on this model for stimulus ranking and provides guidelines for verifying its suitability. Moreover, it discusses alternative pathways for addressing poor model fit, including incorporating different standard deviations and treating the Thurstone method not as a definitive model but as a procedure for ordering objects. Additionally, the study compares the Thurstone method with the average rank method, highlighting the advantages and challenges of each approach. Overall, this research contributes to understanding the complexities of preference studies and provides valuable insights into refining methodological approaches in this domain.