This document reports on a study to develop an empirical method for analyzing and reporting classroom interaction data. Data were collected according to a classroom observation scheme involving the frequency of various interactions (question, answer, or management), the cognitive level of the interaction, the role of the actor, the level of personalization, and the length of the interaction in words. Tape recordings made of 24 upstate New York high school biology and chemistry classes on 6 different occasions during the 1985-1986 school year were analyzed by a panel of expert judges. Each interaction in the randomly selected 10-minute period was classified and the interactions were accumulated. Results of the analysis of these data indicated that observation variables could be grouped into scales that discriminated between teachers, and that the scales were meaningful. (TW)