Key points are not available for this paper at this time.
Three methods of gathering and evaluating value profiles for use in market segmentation are compared. Different reliability estimates are found to produce different conclusions as to the relative test-retest reliability of the three methods. However, the most appropriate statistical measure, Kendall's tau B , reveals that Likert rating scales are significantly less reliable than either rank ordering or paired comparison procedures. No differences in subject interest are found among the methods, but Likert scales and rank ordering require less response time than do paired comparisons. Implications for method choice and future research are discussed.
Reynolds et al. (Sat,) studied this question.