Measuring Implementation of a Structured Educational Model in an Urban School District: An Observational Approach

Most analyses of the National Follow Through Study (Stebbins et al., 1977; House et al., 1978) reported unexpectedly large degrees of variability in the results for each model. Several authors (House et al., 1978; Kennedy, 1978) have hypothesized that this variability was due in part to fluctuations in the extent to which a model was implemented at a given site. Yet only a small sample of classrooms was observed to measure fidelity of implementation in the original evaluation (Stallings, 1975) and those were not of students included in the Abt report. In many ways the Direct Instruction Follow Through Model is an ideal candidate for an implementation study. An almost fully articulated system exists (see, e.g., Becker, 1977; Becker et al., 1980), specifying in detail what teachers should be doing and saying, how much time should be spent on reading, arithmetic, and oral language each day; how fast lessons should be paced, and at what accuracy rate students should be responding. Thus, it was much more direct and simple to operationalize a behavioral coding system than to perform the same procedure on one of the open classroom models analyzed by the Stallings (1975) group in their Follow Through classroom research. Another unique component of the model is that the role of classroom aides is clearly specified-they are responsible for teaching academic curriculum material, usually oral language and arithmetic, to groups of 6 to 10 children. Thus, the same observation system can be used to analyze performance of both teachers and paraprofessionals.