This research utilized a structural analysis of final judgment and a process-tracing method to examine the effects of situational involvement and task complexity on information acquisition and the decision-making process. In addition, the predictive accuracy of the linear model in predicting drug choice across decision situations was assessed. A contingency model for the selection of decision strategies based on a cost/benefit principle was used as a framework in the study. A randomized mixed model factorial design was conducted in which a sample of forty-eight physicians, recruited from the University Medical Center at the University of Arizona, indicated their preferences and choices for hypothetical anti-infective drugs. Subjects were randomly assigned to experimental and control groups. Subjects in the experimental group were told via the written scenario to assume that his/her decision would be reviewed and evaluated by peers and (s)he would be asked to justify drug choice. No mention of peer review was made in the scenario used by the control group. Subjects in both groups were required to perform the two choice tasks and conjoint ranking task. The number of drug alternatives in a choice set was varied between three and six alternatives. The Mouselab program was used to monitor physicians' information acquisition behavior. The measures concerning the amount of information searched, order, duration, direction of search, and the chosen alternative were observed. The conjoint LINMAP program was used to estimate drug attribute importance weights. Results of the study provide support that the information acquisition and decision-making process is contingent on the characteristics of decision tasks. A significant number of subjects shifted from using compensatory to noncompensatory decision-making processes when task complexity increased. The study did not find support for the effects of situational involvement on the decision-making process. However, subjects in the two groups were found to differ in choice outcomes and the attention given to specific drug attribute information. Finally, the study found support for the robustness of the linear models in predicting drug choice across contexts.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.