The Influence of Rating Scales and Question Attributes on the Validity and Reliability of Generalized Trust Scales

Trust is important for a range of societal outcomes. Despite its significance, there is considerable debate about how best to measure trust. In the context of a newly developed measure of generalized trust—the Stranger Face Trust scale (SFT)—this study evaluates whether different features of survey scales affect the reliability and validity of SFT, which relies on the standard 4-point unipolar scale used by many survey institutes. In a survey experiment conducted with a non-probability sample of U.S. adults (N = 4252), we randomly assigned intensity scale midpoints, polarity, and “don’t know” options to SFT. Results indicate that 7- and 9-point bipolar scales without a “don’t know” option slightly outperform all other scales on some psychometric tests, particularly those related to formal properties of the scales and factorial validity, but not on psychometric tests assessing survey environment or convergent, discriminant, and concurrent validity.