The heart of performance- and competency-based teacher education, evaluation, and accountability programs is the establishment of empirical relationships between teacher behavior as an independent variable and student achievement as a dependent variable. Before researchers can adequately establish those relationships they need to deal with the problems of instrumentation, methodology, and statistics. Workers in this area must come to grips with the inadequacy of standardized tests, the unknown predictive validity of tests from special teaching units, the problem of building multivariate outcome measures, the problems of measurement of appropriateness of teacher behavior, the lack of experience in choosing an appropriate unit of analysis for describing teaching behavior, and the lack of stability of many teacher behaviors. Also discussed are the problems of how student background affects measures of teacher effectiveness, what subject matters should be examined, how normative standards and volunteer teachers affect what can be said about teachers and teaching, how individual students react to teaching skills, and how students monitor and interpret a teacher's behavior in ways that may or may not coincide with how educational theorists interpret the phenomena. (JMF)