Findings of educational research offer little indication of a relationship between teacher performance and pupil achievement. A review of reported studies suggest that the difficulty may be methodological. Most investigations are either correlational studies or comparative studies using control group designs. Most correlational research is characterized by a lack of theoretical guidance or explicit rationale, and results are often not replicable. Group comparison studies are impeded by difficulty in establishing preexperimental sampling equivalence of groups, inadequacy of traditional data analysis techniques, and within-group variance. Two alternative methodologies which may be useful for educational research are the time-series and multiple baseline designs. The former is an example of a design where a discontinuity in a series of measurements that occurs coincidentally with the introduction of the treatment variable suggests a possible relationship. The multiple time-series design introduces a nonequivalent control situation in order to rule out the possibility that other potential influences on the dependent variable were responsible for the observed changes. The multiple baseline design is an extension of the time-series model in which the researcher attempts to replicate the effect of a treatment variable across a number of behaviors. Baseline behaviors are established, treatments applied successively, and behavior changes recorded. (HMD)