Understanding Unbiased Dimensions: The Use of Repertory-Grid Methodology

The unbiased nature of the dimensions derived from multidimensional scaling poses a problem of interpretation. Subjective labelling by the researcher assumes that his judgments correspond to those of the respondents and is unsatisfactory. Identification of the dimensions needs to be based upon information gathered from the respondents themselves and in terms of the manner by which they were originally construed. Such information can be derived from the use of the repertory-grid test, which, like multidimensional scaling, requires subjects to make judgments of similarity between objects. The repertory-grid test also provides verbal labels for the distinctions that are made. A principal-components analysis of the repertory-grid data provides a number of components which are shown to be equivalent to the dimensions derived from multidimensional scaling. The use of component scores that relate to the verbal labels allows the dimensions to be identified in terms of the evaluations and perceptions of the respondents.