The present study was conducted to examine the effects of early orientation of counseling related students to the two most prevalent paradigms of psychodiagnostic decision-making on first, the integration of the model, and second, on the ability to make proficient diagnostic decisions while in training. Using an experimental, pretest posttest design, 60 participants from two higher educational sites were randomly assigned to two treatment groups. Participants in each group were oriented to one of two treatment conditions -- a binary decision tree model or a problem-solving model (multiple competing hypotheses). Participants were then introduced to DSM Axis II diagnostic categories utilizing a computer assisted learning laboratory. Results suggested that participants learned diagnosis during the experiment. However, no significant difference in diagnostic proficiency occurred as a result of the two treatment conditions. Additional analyses raised questions about use of case studies as a means of assessing diagnostic proficiency. Item difficulty appeared to be linked to diagnostic clusters and individual diagnoses. Item difficulty factors influenced the internal consistency and validity of test instruments. The assumption of the unidimensial weight of syndromes in the construction of assessment instruments is suspect. Considering the preponderance of case study use for counselor training assessment, caution during instrument construction and use is advised. Evidence also existed that treatment groups responded differently to particular DSM diagnostic clusters and items. This suggested that cluster and item difficulty may be important to consider for instruction of diagnosis in the classroom. Results also suggested that as diagnoses become more complex, problem-solving diagnostic decision-making may become more important. Secondary analysis of computer assisted learning resulted in significant evidence that nonsequential, user-friendly computer assisted instruction may overcome teaching-study style mismatch, resulting in more even distribution of learning over the sample population. / Graduation date: 1998