Significant clustering: Implications for strategic groups research

Research on strategic groups has been hindered for decades by inability to test for significant clustering. Do firms actually clump together in distinct strategic groups? In lieu of a significance test, heavy emphasis has been placed on tests of construct validity, and the group membership-performance link has emerged as the de facto litmus test for the existence of strategic groups. Paradoxically, these tests for construct validity have led to distortions of the concept itself to fit the available tests. This is particularly disturbing given that the group membership-performance link is itself invalid. Several programs are now available with tests for significant clustering. A multimethod approach exploits the complementarity of permutation and Monte Carlo techniques. Strategic groups are identified by the interdependence that binds firms together, rather than the mobility barriers that keep them apart. This approach is illustrated using the European airline industry. Both theoretical arguments and industry experts predicted that strategic groups would not differ in performance. Findings support the existence of two strategic groups: Low-cost airlines and full-service airlines. Results support the face validity and predictive validity of these iconic strategic groups and contradict the logically flawed group membership-performance link. This approach heals a schism in the field between those who view strategic groups as subsets of similar but independent firms (which is consistent with nonsignificant clustering) and those who view them as distinct groups of interdependent, interacting firms (a special case that emerges due to significant clustering). Nonsignificant clustering can also yield additional insights. In this example, hybrid groupings shed light on the convergence on the mainstream middle in the European airline industry.