The purpose of this dissertation was to, from the point of view of interaction, study how the factors group size, group composition, group norm and problem type influenced group problem solving. Three classroom studies were performed in grades four and five of the Swedish nine-year compulsory school. In addition to the participants in various pilot studies, the experiment included 1146 pupils. The performance was measured both regarding group results and the individual short-term and long-term retention of knowledge. A fifth factor, group productivity, was formed on the basis of these two measurement values.Group size 1, 2, 3 and 6 were studied. The group composition was defined on the basis of, on one hand, the ability levels high, middle and low and, on the other, on the basis of sex. The group norms were included through instructions about cooperation and competition. A third group norm, so-called free norms, was also introduced. Two problem types were used. One of them was designed so that the other group members would easily be able to judge if a response from one of the group members was correct or not (high accessibility). The other problem type was defined in terms of low accessibility. The first study included the factors group size and group composition (ability), the second study included group composition (sex), group norm, problem type and group productivity, and the third study included all five factors. The factor group size turned out to be of great importance and interacted with each of the other factors. Only high ability pupils could make use of group size in interaction with e. g. the factor group composition. Group composition was also interesting regarding sex. Co-operation favoured girls and competition favoured boys. The results also indicate that girls remember what they learn better than boys. Interactions between the factor group norm and the factors group size, problem type and group productivity showed that, in no case, did co-operation produce poorer results than competition and free norms. Various interactions with the factor problem type and other factors showed that larger group sizes are more favourable to the solving of problems with a high accessibility than to the solving of problems with a low accessibility. However, the individual group members had a better recollection of the solutions to problems with a low accessibility than to problems with a high accessibility. The factor group productivity was included in interactions with each of the other factors and in the three-way and four-way interactions. These interactions indicate that the factor group productivity, in terms of both group results and individual results, is an important factor in group research and in other classroom research.By way of conclusion, the results are discussed regarding educational implications and continued research. / digitalisering@umu