Comparing theories: What are we looking for?

Abstract

Two recent papers, Harless and Camerer (1994) and Hey and Orme (1994), were both addressed to the same question: which is the 'best' theory of decision making under risk? A second question that both addressed was: are any of the new generalizations of Expected Utility theory (EU) significantly better than EU (in some appropriate sense)? These are important questions: much theoretical effort has been expended in trying to produce a 'better' story of decision making under risk than that apparently provided by EU. What has been the purpose of this effort? Surely to improve the predictive power and descriptive validity of economics. These, of course, are competing objectives in general: other things being equal, the greater the predictive power of a theory, the lower the descriptive validity of that theory. However, the purpose of 'better' theory is to make other things not equal. Nevertheless, there generally (and as it happens in the context of recent theories of decision making under risk, specifically) is the need to make some judgement of the appropriate trade-off between predictive power and descriptive validity: simply because if one theory was better in both predictive power and descriptive ability than a second, the second would simply be discarded - it would be dominated by the first.