ABSTRACT
In most of the magnitude estimation experiments performed by S. S. Stevens (see Stevens, 1975), subjects were given a modulus and requested to assign numbers to subjective magnitudes of stimuli in such a way that the ratios among the numbers were directly proportional to the ratios among the subjective magnitudes . Fundamentally, the choice of the modulus was arbitrary, in agreement with the multiplicative transformations allowed by ratio scales. Nevertheless, Stevens noted that the scale, as evidenced by the functional relation between the assigned numbers and stimulus intensities, depended nonlinearly on the modulus . This suggested to Stevens (1956) "that 0 weights his judgment by a factor related to the absolute level" (p. 10). In some experiments he allowed the subjects to choose their own moduli and found that the results agreed best with what he considered to be minimally biased scales. He stated with respect to these experiments that they seem "to go about as far as an experiment can go toward getting the Os to make absolute judgments of loudness" (p. 21) . Stevens , however, normalized these results to a common modulus obtaining a ratio rather than an absolute scale, thus making only the slope and not the absolute values of the function meaningful.
