Quality of Semantic Differential Scales - an Application of Multiple Correspondence Analysis

Semantic Differential Scales (SDS) are frequently used to assess self- and other-descriptions. The application of SDS is based on a number of assumptions: (1) the linearity of rating scales, (2) the symmetry of scales, (3) the equal distance of scale points; however, these assumptions are usually not matter of analysis. Against this background, the present study employs Multiple Correspondence Analysis (MCA) to analyze the underlying assumptions of SDS. A randomized between-subject experiment (split-ballot) was realized, varying three different forms of SDS. The first form was a classical form with polar adjectives (good – bad), the second form used unipolar adjectives (good-not good) and the third form was a bipolar form, but with nonsense pairs (good – passive). The respondents had to evaluate themselves and a well-known German politician by one of the SDS forms. We conducted a web survey on a probability sample of German residents (N = 552, 53% males, age M = 42.63, SD = 14.77). The results of MCA are reported and discussed with regard to their methodological implications.