How problems are solved: Reconsidering the phase theorem

The literature on problem-solving is dominated by the phase theorem. The descriptive facet of the theorem suggests that problem solvers follow a certain sequence of phases. Its prescriptive facet suggests that problem solvers are more likely to succeed if they follow a certain sequence of phases. Despite the ubiquity of the phase theorem, there is no empirical support for either its descriptive or its prescriptive validity. This study tested the descriptive and prescriptive validity of a six-phase probabilistic problem-solving model in seventy-five pairs of retrospective reports of successful and unsuccessful problem-solving in organizational settings. In both types of cases, the location of phases in the problem-solving process was consistent with the sequence prescribed by the model, and the order of consecutive phases, irrespective of location in the sequence, was also consistent with the model. Differences between success and failure processes were found in frequency of diagnosis, the compatibility of diagnosis and action, and the extent to which diagnosis preceded action. The results suggest that, contrary to received opinion, the descriptive validity of the phase theorem is still unresolved. Further research is required to establish whether the descriptive and prescriptive models suggested by these results reflect actual problem-solving processes, or implicit schemata of how problems are, and should be, solved. (C) 1996 Academic Press, Inc.