number of classes in this study alllowed the investigators to provide the variability in group structure and patterns of behavior required by such a large number of independent variables.

Only a small selection of the findings can be presented here. Readers may refer to the original publication to understand the intricacies of this study and its manifold findings. Collaborative groups, whether structured or unstructured in terms of students' interaction patterns, appear to enhance the quality of students' interactions regarding complex tasks. Explicit interdependence structures may be unnecessary to generate productive student interactions. Lower achieving students benefited more from working in pairs than in groups, and pairs produced higher collaborative ratings for these students, while other group members collaborated comparably in dyads and in small groups. Greater disagreement and negotiation that improve learning were generated by small groups rather than by pairs.

As always, further research is required to explore many topics that could not possibly have been included in the Fuchs et al study, such as problem solving with broad social/historical/intellectual or scientific issues, not only with mathematical problems. Furthermore, the teachers in this study seemed to have received minimal preparation for their role, and the duration of the study was very short. This experiment is so highly sophisticated that these investigators seem to be best qualified to continue this line of research that, in time, may provide some definitive answers to many more of the perplexing problems of cooperative learning.

Learning Disabilities
Fred Brandt and Nancy Ellsworth (1996) reported a study using Learning Together with 72 adolescents and young adults ages 14 to 20 in 12 classrooms for the learning disabled in a New York city high school. The teachers learned the Learning Together model in a 45 hour preparatory course. The experiment lasted 15 weeks, during which 6 classes (of 5 to 9 students per class) were taught with the cooperative method at least 3 days per week for at least 20 minutes per day, and six classes - with the traditional whole-class approach. Each class was taught Spanish, English, math and science (bilingual). The goal of the study was to assess the effects of Learning Together on these students' academic achievement and self esteem. It was hypothesized that the cooperative environment would provide welcome relief for these students from the highly competitive nature of traditional instruction, and would offer them the support and peer encouragement needed for more sustained concentration on learning tasks.

The effectiveness of cooperative learning with exceptional students is far from unequivocal. The group-interactive nature of this approach could possibly constitute a hindrance for learning disabled students rather than a source of support (Tateyama-Sniezek, 1990), a view that was widely subscribed to in the literature on learning disabilities for several decades. On both standardized and criterion-referenced tests students in the Learning Together classrooms registered higher scores than their peers in the traditional classes. No differences were found between the groups on measures of self esteem. Teachers reported fewer discipline problems in the cooperative classrooms, as well as a general atmosphere of friendship. Many students said that they enjoyed their class assignments and valued what they were learning. In the control classes teachers noted a distinct climate of apathy on the part of the students, ethnic rivalry, and hostility toward the classroom. They felt their studies were irrelevant to their lives.