75 Years After Likert: Thurstone Was Right!

Abstract

For over three-quarters of a century researchers and practitioners have analyzed rating scale data using methods that assume a dominance response process wherein an individual high on the trait assessed is assumed to answer positively with high probability. This approach derives from Likert's famous 1932 approach to the development and analysis of rating scales. In this paper, we argue that Likert scaling and related methods are misguided. Instead, we propose that methods that have evolved from Thurstone (1927, 1928, 1929) scaling provide a better representation of the choice process underlying rating scale judgments. These methods hypothesize an ideal point response process where the probability of endorsement is assumed to be directly related to the proximity of the statement to the individual's standing on the assessed trait. We review some research showing the superiority of ideal point methods for personality assessment and then describe several settings in which ideal point methods should provide tangible improvements over traditional approaches to assessment.