Invalid results from the method of constant stimuli

Abstract

The Method of Constant Stimuli was shown to produce constant errors in the direction expected by E. In an experiment, 89 Ss compared variable lines with a standard, each S producing a psychometric function and Point of Subjective Equality (PSE). Four groups differed in the particular range of variable stimuli used; whether the stimuli averaged longer or shorter than the standard, and whether they had a wide or narrow range. All sets of variable stimuli encompassed the standard. PSE was consistently between the standard and the mean of the variables, as predicted by Adaptation Level Theory. In many experiments, the set of variables is chosen on the basis of theory or pilot studies; the present study shows that PSE will be spuriously drawn toward the center of the variables, hence toward E’s pre-experimental expectations.