If Thurstone Was Right, What Happens When We Factor Analyze Likert Scales?

Document Type


Publication Date



Likert scaling, rating scales, Thurstone scaling method, choice process, personality assessment

Digital Object Identifier (DOI)



Comments on 75 years after Likert: Thurstone was right! (see record 2010-25616-017) by Drasgow, et al. Drasgow, Chernyshenko, and Stark (2010) rightly explain how the unfolding model can be a better representation of people's responses to what they term introspective judgments (e.g., ratings of attitude or personality) than the dominance model. Drasgo et al. noted Davison's (1977) argument that a set of unidimensional items would form two factors when factor analyzed if those items fit an unfolding model. Drasgow et al. further noted that the veracity of arguments about the underlying structure of a set of items depends on getting the response model right. We elaborated on these points in Spector, Van Katwyk, Brannick, and Chen (1997), and produced what we believe is an empirical example of the artifactual two-factor phenomenon using job satisfaction data. In this commentary, we further discuss the issue of artifactual factors produced by respondents disagreeing with oppositely worded items and provide data that this phenomenon can occur in a sizeable proportion of a sample.

Was this content written or created while at USF?


Citation / Publisher Attribution

Industrial and Organizational Psychology, v. 3, issue 4, p. 502-503