Abstract
225 judges (Js) predicted the responses of 4 targets (Ts) to interest test items. Two characteristics of the information given the Js about the Ts were varied: (1) number of relevant information dimensions (dimensions) and (2) the amount of irrelevant information (noise). It was hypothesized that increases in both variables would lead to an increase in error. Analysis of the total data indicated that an increase in noise led to a significant increase in error but that an increase in dimensions led to a decrease in error. It is suggested that this latter result is due to response set. An analysis of partial data not susceptible to such a set showed that amount of noise had a significant retrograde effect on accuracy but that the effect of number dimensions was not significant. These results are, in part, consistent with those found in concept identification tasks.
Get full access to this article
View all access options for this article.
