The Failing Measurements of Attitudes: How Semantic Determinants of Individual Survey Responses Replace Measures of Attitude Strength

Abstract

The traditional understanding of data from Likert scales is that the quantifications involved are resulting from measures of attitude strength. Building on our recently proposed a semantic theory of survey response (STSR), we claim that survey responses tap two different sources; a mixture of attitudes plus the semantic structure of the survey. Exploring the degree to which individual responses are influenced by semantics we hypothesize that information about attitude strength is actually filtered out as noise in the commonly used correlation matrix. Applying a linguistic algorithm termed MI, we separated semantics from attitude strength in four samples of altogether 7781 respondents covering 8187 pairs of items. The surveys spanned commonly used organizational behavior surveys on leadership and motivation, as well as a short 5-factor personality inventory, the NEO-FFI. As hypothesized, the findings indicate that levels of attitude strength did not contribute uniquely to the correlation matrices except for in the NEO. This is contradictive to the prevalent understanding of what survey data represent. This problem has been overlooked, possibly contributing to reduced predictive value from research relying on Likert scale data.

Keywords: Semantic analysis, Surveys, survey responses, semantic theory of survey response (STSR), attitude strength