Rational Choice and the Relevance of Irrelevant Alternatives

This experimental study investigates the inuence of irrelevant or phantom al- ternatives on subjects' choices in sequential decision making. Using experimental data from 45 subjects, we found that irrelevant alternatives bear significant rele- vance for decision making. We observe that only 38% of our subjects make the same choice after two phantom alternatives, as compared with the same decision problem when analyzed from scratch. Even allowing for a natural error rate as high as 25%, we find that between 40% and 60% of our subjects are led astray by the presence of phantom alternatives. Testing then basic postulates of rational choice, we find moderate violations of contraction monotonicity and static preference consistency, and substantial viola- tions of dynamic preference consistency. Finally we find that subjects exhibiting rational choice behaviour are far less susceptible to dependence on irrelevant alternatives than subjects which violate rational choice behaviour. Rational choice behaviour is thus a good proxy for the independence of a subject's choices of irrelevant alternatives.