Memorandum, Department of Economics, University of Oslo 2002,21
The Duhem-Quine thesis asserts that any empirical evaluation of a theory is in fact a composite test of several interconnected hypotheses. Recalcitrant evidence signals falsity within the conjunction of hypotheses, but logic alone cannot pinpoint the individual element(s) inside the theoretical cluster responsible for a false prediction. This paper considers the relevance of the Duhem-Quine thesis for experimental economics. A starting point is to detail how laboratory evaluations of economic hypotheses constitute composite tests. Another aim is to scrutinize the strategy of conducting a series of experiments in order to hem in the source(s) of disconfirmative evidence. A Bayesian approach is employed to argue that reproducing experiments is not necessarily useful in terms of identifying correct causes of recalcitrant data.