The problem in this study concerned Judge's (J) hostility as a factor in his perception of other persons. It was hypothesized that in comparison to a J of average hostility, a J of high hostility would rate higher, a J of low hostility would rate lower: (1) intensity of moderately hostile wants, and (2) the suitability of low and moderately hostile performances for achieving moderately hostile wants.

Js gave self-reports of reactions to hostility-provoking situations, and rated several parameters of intentional action: (1) intensity of wants; (2) intensity of constraint against wants; (3) ability to carry out performances; and (4) suitability of performances as means to wants as ends.

Groups of 5 Js in each group predicted the behavior in the hostility-provoking situations of the other members of the group and rated the hostility of the other members of the group.

There were two experiments, in which the Js predictions were simulated by a computer model. In Experiment I the parameters for the simulation of each J were the ratings which that J had made. A baseline simulation employed these ratings without change. For experimental simulations these ratings were modified before simulation as a function of the hostility of the J being simulated so as to compensate for the hypothesized hostility-related differences among Js.

In Experiment II the parameters for the simulation of each J were the means over all Js of parameter ratings. A baseline simulation employed these means unchanged. Before the experimental simulations these means were modified as a function of the hostility of the J being simulated so as to reproduce the hypothesized hostility-related differences among Js.

Degree of inter-judge disagreement was assessed by rated dissimilarity between predicted performances. It was predicted that in Experiment I the experimental simulations would produce less inter-judge disagreement than would the baseline simulation. This prediction was confirmed. Comparisons of simulations relative to the original predictions, however, did not reveal any relative differences between baseline and experimental simulations.

It was predicted that in Experiment II the experimental simulations would differ less in inter-judge disagreement from the original predictions than would the baseline simulation. This prediction was confirmed, although the experimental simulations were not better simulations of individual Js taken singly than was the baseline simulation.

The hypotheses of the study were supported by the results. The apparent inconsistencies revealed by the lack of relative differences between simulations in Experiment I and by the findings concerning relative goodness of simulation in Experiment II may be explained by the multi-dimensionality of the dissimilarity ratings by which inter-judge disagreement was assessed, and by the relatively limited scope of the computer simulation model which was employed.

The implications of this study for clinical practice, for theoretical explanations of hostility, for further applications of computer simulation technology, and for observational methodology were pointed out and discussed. [183 pp.]