The sharply drawn distinction between research and evaluation, and the equally sharp distinction between the carefully controlled laboratory-type experiment and the study occurring in the field, are seen as arbitrary and not very helpful. In illustrating this thesis, the author develops a design generally regarded as a "research" design and shows the evaluative purposes for which the design may be used. A Teacher Corps/Trenton (New Jersey) State College/Trenton Public School inservice program for the improvement of reading instruction skills was chosen as the subject of the study. Observational data on teacher change due to training, effectiveness of varying training programs, and the relationship between change/level of teacher skills and pupil gains in reading are presented. The data, essentially of an evaluative nature, is extensively interpreted to illustrate the contention that researchers are also learning about the nature of teaching and the structure of teaching performances. The evaluative data also yield information on the relationship between kinds of teacher performances and pupil learning. Through this interpretation, the author illustrates that by carefully designing a study, one can achieve two goals: first, one can develop an understanding about teaching performances related to pupil learning; and two, the effectiveness of ideas about what constitutes the appropriate substance of training and whether or not the chosen methods are effective can be evaluated. (MJB)