Vague Linguistic Expressions and the Problem of Equidistance in Verbal Response Scales


This contribution examines the problem of equidistance of vague linguistic terms (LT) of verbal response scales, which are used in psychological questionnaires or interviews. The study design (N = 92) employs an empirical translation procedure for the numerical translation of LTs in an example questionnaire (i.e., measuring chronic stress) that utilizes a verbal response scale with frequency expressions (e.g., sometimes). The data are modeled using fuzzy membership functions (MFs) that reflect the LTs’ meanings. Results show that the LTs of the original questionnaire scale are not distributed equidistantly and, therefore, that the presuppositions for data analysis are statistically violated. Further, this violation biases study results in such a way that measured stress levels are overestimated. A proposed alternative scale with different LTs shows nearly equidistant response categories. To solve the problem of equidistance, a fuzzy analysis of response data is presented and compared to conventional analyses.

Back to Table of Contents