A study was undertaken to explore cost-effective ways of making career ladder teacher evaluation system decisions based on fewer measures, assessing the relationship of observational variables to other data and final decisions, and comparison of compensatory and conjunctive decision models. Data included multiple scores from eight data sources in the 1985-86 Tennessee Career Ladder Teacher Evaluation System. The data sources include: (1) classroom observation; (2) teacher/evaluator dialogues in planning, teaching strategies, and evaluation; (3) a peer questionnaire; (4) a principal questionnaire; (5) student questionnaires (primary, elementary, and secondary levels); (6) the Tennessee Career Ladder Professional Skills Test; (7) Professional Development and Leadership Activities Summary; and (8) evaluator consensus judgments. Findings indicate that less data can be gathered without having any major impact on the decisions reached if one uses optimal weighting. Several fairly accurate models using various reduced sets of data were proposed. Although classroom observation data were not highly related to the other variables, there was evidence indicating that the decisions reached without such expensive-to-gather data would be highly similar to the decisions actually reached. It was not possible to compare a purely compensatory model with a purely conjunctive model using these data. However, the comparison of the actual decisions reached in Tennessee with those made using a conjunctive data combination model gave no support for preferring a compensatory model. (TJH)