Morality judgments: Tests of an averaging model.

Pairs of items describing objectionable behaviors were rated for their overall morality and the ratings were demonstrated to depend upon the range as well as the average scale value of the component behaviors.

Abstract

Pairs of items describing objectionable behaviors were rated for their overall morality. Contrary to additive or constant-weig ht averaging models, the ratings were demonstrated to depend upon the range as well as the average scale value of the component behaviors. A range model accounted for more than half of the variance left unexplained by the additive models. One interpretation of the range effect postulates that each component stimulus produces a distribution of values. The value of the stimulus combination is assumed to be the mean value in the overlap of the component distributions, which is closer to the item with the narrower dispersion. How immoral is it to both pocket the tip the previous customer left for the waitress and poison your neighbor's dog whose barking bothers you? Anderson (1968b) has suggested that 5s combine information by averaging psychological values associated with each component stimulus. Thus, 5 independently assesses the morality of pocketing the tip and the morality of poisoning the dog and then averages his two assessments to arrive at an overall rating. According to this view, the psychological value of the whole is simply an average of the psychological values of the parts. The theory is a special case of a general additive model (Rosenberg, 1968), which can be written: