Measuring the Importance of Positive Constructs: A Test of Alternative Rating Procedures

When constructs are positive in nature, individuals rating a set ofthem in terms of importance tend to end-pile their ratings toward thepositive end of the scale, resulting in little differentiation amongthe items rated. Results of two experiments indicate that a procedurewhere respondents first pick their most and least important items in aset and then rate the entire set of items (least-most) and aprocedure where respondents rank the items prior to rating(rank-then-rate) provide more differentiation and lessend-piling than a simple rating procedure (rate-only). Theresults also show that the increased differentiation for the least-mostmethod results in improved fit of latent structure, compared to arate-only procedure. These results generalize across two types ofpersonal values scales (Rokeach Value Survey and Kahle's List of Valuesscale), number of items rated (9 and 18) and number of rating points (1to 10 and 0 to 100).