Data were gathered, using pre and posttests and interviews, on the learning of individual students during a seven-week unit on a science topic (conservation) in a fourth grade classroom. Case studies of the learning experiences of each child, based on recordings and detailed observations of all class activities, were used to predict student learning and memory for each content item in the posttests. Analysis of the predictions made by researcher and class teacher revealed the significance of type of class activity, student learning and friendship hierarchies, prior knowledge and experience, and the teacher's ability to diagnose student misunderstanding. A model of effective teaching and the Haberlee Evaluation technique are described. (Author)