Regression effect and individual power functions over sessions.

Abstract

Stevens attributed individual differences in power function exponents mainly to differences in "the regression effect" and suggested that balancing for regression may largely do away with individual differences. Cross-modality matches of loudness to apparent time duration, and vice versa, were repeated over six sessions for each of 20 5s with 24 hr. between sessions. Individual differences, indicated by analysis of variance, Kendall's W, and significant positive correlations across sessions, occurred before and after regression balances. Group exponents for duration matches increased over sessions, but those for loudness matches and those obtained by regression balance did not change significantly.

Inverse cross-modality matching: A test of ratio judgment consistency for group and individual data

“…measured in radians, are also listed in Table 1. The values of the regression angles and the various exponents in this table are approximately the same as found by Wanschura and Dawson (1974) using the same two sensory continua. Further.…”

“…The straight lines accounted for 66% to 80% of the variability, on the average, for an individual subject depending on the task. The quality of the individual functions resembled that found in Wanschura and Dawson (1974). It is interesting to note that subjects who tended to have a high exponent in the direct matches also had high exponents (but negative) for the inverse matches.…”

“…In addition, a regression balance (S. S. Stevens & Greenbaum, 1966) was planned in an attempt to reduce the amount of intersubject variability. S. S. Stevens (1971) suggested that a reduction is apt to occur with such a balance, but Wanschura and Dawson (1974) found little, if any, such a decrease for cross-modality matches.…”

“…Individual functions have been found to follow Stevens' power law for the methods of magnitude estimation and magnitude production (J. C. Stevens & Mack. 1959) and for cross-modality matching (Wanschura & Dawson. 1974).…”

On bias in magnitude scaling and some conjectures of Stevens

“…Next are three studies by Dawson and colleagues (Dawson & Brinker, 1971;Lilienthal & Dawson, 1976;Wanschura & Dawson, 1974) in which loudness and duration were used in a counterbalanced design. The table shows that the exponent ratios, β/β′, are again consistent with typical findings: the values are all around 0.6, which is consistent with a loudness exponent of about 0.6 and a duration exponent of about 1.0.…”