The smallest meaningful sample of classroom transactions

Abstract

Abstract In order to explore the question of the number of observations necessary to obtain a trustworthy sample of classroom transactions, the data from an existing study were reanalyzed. Sixteen teachers, all of whom were trained to use the materials and procedures of the Science Curriculum Improvement Study, were observed four times with a five‐category system which classified teacher questions. In this sample, teachers were extremely variable across occasions. The reliability of a single observation was zero for four of the five variables. For the fifth variable, asking students to hypothesize, it was estimated that 11 to 20 observations would be necessary to obtain a stable mean score for each teacher.