The study also involved locating important predictor variables for future consideration when constructing a change model designed to manipulate these police behaviors. The study sought to determine the applicability of a multivariate data analysis to research in the area of interorganizational relations. Data were primarily obtained from a self-administered questionnaire to which 69 Muskegon police officers responded, out of a total of 76 officers. Questionnaire information also was obtained from 10 persons in juvenile probation, 57 in DSS, and 12 in CAAP, in order to make specific comparisons between police and members of these agencies. The questionnaire sought to measure police behavior with other agency personnel, social-psychological variables, and several structural variables. A cluster analysis technique was used to develop composite scores for the activities of interaction and referral. Study results showed that police officrs had the highest level of interaction and referral with juvenile probation, then DSS, and CAAP, respectively. Other results are also discussed. Among the implications of the study is the observation that social-psychological variables are poor predictors of police behavior with personnel of supportive-type agencies. The study also showed that multivariate data analysis techniques are extremely useful in describing or explaining relations among personnel of different organizations. Tables, figures, footnotes, and a bibliography containing approximately 108 citations are provided. Questionnaire items and additional items used in the analysis are appended. (Author abstract modified)