Deescalation strategies: A comparison of techniques for reducing commitment to losing courses of action.
Previous research suggests that decision makers have a tendency to become locked into courses of action—to throw good money after bad in dealing with losing projects. The present study directly compared the effectiveness of several deescalation strategies designed to make decision makers more responsive to the available evidence. Three deescalation procedures were found to be most effective: (a) making negative outcomes less threatening; (b) setting minimum target levels that, if not achieved, would lead to a change in policy; and (c) evaluating decision makers on the basis of their decision process rather than outcome. The theoretical and practical implications of each of these strategies are discussed. A long stream of empirical research has investigated the commitment of decision makers to losing courses of action. Under the rubrics of sunk cost effects (Arkes & Blumer, 1985; Garland & Newport, in press), entrapment (Brockner & Rubin, 1985), too much invested to quit (Tegar, 1980), and the escalation of commitment (Staw, 1976,1981), numerous studies have shown how individuals can become locked into failing courses of action. To date, this literature has isolated a wide variety of determinants of behavior in escalation situations, ranging from psychological to social, organizational, and project concerns not typically dealt with in the economics of investment (Staw & Ross, 1987). Much of the research in this area has focused on the determinants of escalation, and little attention has yet been given to procedures that might help people avoid the escalation trap. This is unfortunate because, just as debiasing research has helped behavioral decision theory provide policy recommendations (e.g., Fischhoff, 1982), deescalation research could potentially aid managers in avoiding the overcommitment of resources. So far, however, only the rudiments of a deescalation literature are in place. In such a stream of research, one might place Tegar's (1980) and Brockner, Shaw, and Rubin's (1979) studies showing that limit-setting can reduce escalation. Also in this group would be Nathanson et al.'s (1982) experiment showing that information about the problem of entrapment can deter individuals from initially engaging in an escalation situation. McCain's (1986) experiment showing the reduction of investment when losses are repeated and clear-cut, as well as Staw and Ross's (1987) more general discussion of possible techniques for reducing escalation are also relevant. Nonetheless,