Averaging processes in ratings and choices based on numerical information

Ss were presented sets of numbers said to represent sample price information from grocery stores. Ss rated the relative desirability of shopping at each of a series of stores and were also asked to make paired-comparison preference choices. In Experiment I, some sets contained favorable information, some contained unfavorable information, and some contained neutral information added to the favorable or unfavorable information. The addition of neutral information led to a decrease in the polarity of responses. This is consistent with an averaging, but not an adding, formulation of how the information is integrated. In Experiment II, sets of favorable or unfavorable information varied in size and a set-size effect was obtained. The greater the amount of favorable or unfavorable information, the more extreme the response. This was accounted for by assuming that an initial neutral expectancy is averaged with the information presented.