Problems with So-Called “Direct” Scaling

It was once thought that magnitude estimation was a better method for the measurement of subjective value than the method of category rating. However, recent research shows that magnitude estimations do not have advantages over category ratings and are in fact no more than ordinal scales of subjective value. When subjects are asked to judge "differences" and "ratios" of subjective value, it appears that subjects compute subjective intervals for both tasks. Magnitude estimations of "ratios" appear to be an approximate exponential function of subjective differences, and category ratings appear to be an approximate linear function of subjective differences, though the exact form of these functions depends on the stimulus and response distributions. Scales of subjective value that are derived from the subtractive model of within-mode stimulus comparison appear to be largely independent of the stimulus spacing and the response procedure. Measurements of subjective value should be derived from a theory of empirical relationships, and should show generality across contexts and across empirical domains. So-called "direct" scales have not been successful in predicting empirical relationships, whereas subtractive model scale values have shown promise.