Experiments as exhibits and experiments as tests

Two roles of experiments in behavioural economics are distinguished - experiments as exhibits and experiments as tests of theories. An exhibit is an experimental design which reliably induces a surprising regularity, typically combined with an informal hypothesis about the underlying causal mechanism. 'Deviation theories' - generalisations of received theories which incorporate additional causal mechanisms - are constructed so as to be consistent with known exhibits, and tested in new situations. The paper argues that the principles of good practice are different for the two types of experiment, in part because of a tendency for successful exhibits to be produced by several causal mechanisms.