The measurement of personal values in survey research - A test of alternative rating procedures

When survey researchers are interested in measuring the personal values of respondents, they often use a rating rather than a ranking method because it is easier and faster to administer and yields data that are amenable to parametric statistical analyses. However, because personal values are inherently positive constructs, respondents often exhibit little differentiation among the values and end-pile their ratings toward the positive end of the scale. Such lack of differentiation may potentially affect the statistical properties of the values and the ability to detect relationships with other variables. Two experiments were conducted via mail surveys to general population samples to test alternative rating methods designed to increase differentiation and reduce end-piling in the rating of personal values. The results suggest that a procedure in which respondents first pick their most and least important values, then rate them (most-least), provides more differentiation and less end-piling than a simple rating procedure (rate-only). Increased differentiation for the most-least method influenced the fit of latent structure and resulted in more robust relations between the values ratings and other criterion variables. These results generalized across type of values scale, number of values rated, and number of rating points.