Abstract
Applications of the semantic differential technique, involving small numbers of concepts and scales and eliciting descriptions of the concepts rather than metaphoric meanings, present special problems of analysis, especially in accommodating "concept-scale interaction." In considering possible analysis strategies, Maguire's analysis of variance model of SD responses is examined. In this model, an individual's score on a scale for a particular concept can be attributed to various sources, including a concept by scale interaction effect, different from concept-scale interaction as generally understood in SD research. Maguire's model is valuable in identifying response sets in SD data, but fails to solve the problem of concept-scale interaction encountered in the reduction of scales to a set of dimensions common to all concepts by factor analytic methods. Comparison of concepts in terms of these dimensions is often the main aim of SD research. For this purpose, the model requires extension to the formation of the covariances on which the factor analysis is based. A number of methods of forming covariances are considered and exemplified using data from a study of teenagers and drinking. The paper concludes with the discussion of an optimum strategy for analyzing semantic differential data.
Get full access to this article
View all access options for this article.
