An examination of methodological issues relevant to the use and interpretation of the semantic differential

A number of methodological issues have been raised regarding the semantic differential technique. This study re-examined several key problems, particularly the assumed bipolarity of scales, instructions regarding use of the midpoint, and concept-scale interaction, all of which may contribute to a lack of precision in the technique. In addition, this study utilized an analysis of variance model to partition variance in semantic differential ratings. Forty subjects responded to one of four instruments on two occasions. Instruments differed in terms of polarity type (bipolar or unipolar) and presence or absence of an irrelevance option. Twenty-four concepts uniformly distributed throughout semantic space were judged on either 15 or 30 scales. Results indicated that the Evaluation- Potency-Activity (EPA) structure was both robust and reliable. However, several features of the data argued for caution in the use of the semantic differential technique. Both the Concept x Scale interaction and the Scale x Concept x Person interaction accounted for substantial proportions of variance in semantic differential ratings. Suggestions were offered to minimize such effects