Problems associated with combining observations taken from intact classrooms were examined. Clarification of basic entities to be studied was a critical step in research planning. Classroom observers usually study instructional arrangements and other classroom phenomena. The author used activity structure and its component activity segments as the focus for study. An activity segment was defined as a unique time block in a lesson which occured in a fixed physical setting. Segments have instructional forms designed for the accomplishment of specific tasks. Daily activities in a classroom conform with the general type of curricular approach and goals. Activity structures are produced to accomplish certain goals. When observational data is characterized by distributions which are bimodal or have many zeroes or non-occurences, summary with means and standard deviations may result in erroneous conclusions. The author observed 20 fifth-grade math classes from diverse school districts. He determined that combining data within two clusters, highly individualized and whole class programs, provided more meaningful and interpretable results. Researchers and writers need to include background descriptive information about the programs and contexts which they have studied to facilitate sorting of data. (DWH)