Improving judgment by reconciling incoherence

Abstract

This paper discusses how a subject's judgments about his actions, uncertainties and values may be improved by seeking out and reconciling inconsistences between related judgments. Decision theory tells us that there are relationships between coherent judgments, such as between a prior, likelihood and posterior, but does not tell us how a subject is to reconcile his own, possibly incoherent, views. The role of coherence in improving judgments is not clear. This paper discusses whether there is a unique, best reconciliation of incoherent judgments and how such reconciliation could improve decision-making. The basic principles of the coherent view are explained and a language developed in which the problem of reconciliation can be discussed. The practical importance of the problem is emphasized, illustrated by several examples and possible solutions are explored.Theory and Decision 14 (1982) 113-132.

A philosophical basis for decision aiding

“…An example of such work is the development of techniques to help the DM to reconcile his inconsistent judgements, rather than simply pointing them out (Brown and Lindley, 1982;Lindley et al, 1979;Freeling, 1980Freeling, , 1981b. Such a theory would be "better" than DA to the extent that it was more useful.…”

New Methods for Estimating Business Markets

“…underlying the methodology of plural evaluation (Brown and Lindley, 1982; is the principle that combining several approaches to the same problem generally leads to more confidence in the results. This principle has long been part of the ''folk wisdom" of market researchers, and has great intuitive appeal.…”

Alternate Theories of Belief and the Implications for Incoherence, Reconciliation, and Sensitivity Analysis

“…However, he argues that since one piece of evidence might support only a set of events, rather than a single event, the belief induced by that evidence should be apportioned to that set of events, and not to any particular event. For example, consider the annual Oxford v. Cambridge boat race (as have Smith, 1961;Brown and Lindley, 1981;Freeling, 1981a). An event of relevance to the outcome is a coin toss, for the winner of that can take an important inside bend on the River Thames.…”

“…(For a discussion of that, see Shafer, 1976.) To effect a "reconciliation" of the form discussed above is very difficult in the Bayesian framework--witness the literature produced on the subject, e.g., Morris (1974Morris ( , 1977; Lindley, Tversky, and Brown (1979); Brown and Lindley (1981);…”

“…The DM is then being inconsistent , or exhibiting incoherence. In our previous work on RIJ (Brown and Lindley, 1981;Lindley, Tversky, and Brown, 1979;Freeling, 1980bFreeling, , 1981a we have examined the possibility of producing mathematical techniques to provide a single set of consistent, reconciled values from the inconsistent set provided by the DM. While we have had some success in producing such techniques, it has become apparent that there is no unique reconciliation of a set of inconsistent values.…”