The implications of imperfect measurement for free-choice carry-over effects: Reply to M. Keith Chen’s (2008) “Rationalization and cognitive dissonance: Do choices affect or reflect preferences?”

Examinations of post-choice decision-making behavior often involve two successive choices. At time 1, participants choose between two equally attractive items. At time 2, participants choose between the unchosen item from time 1 and a new item that is roughly equal in attractiveness to the other two. The option rejected at time 1 will tend to again be rejected at time 2, a tendency often attributed to a psychological carry-over effect. Chen (2008) (Available from: http://cowles.econ.yale.edu/P/cd/d16b/d1669.pdf) discounts these psychological explanations. He argues that, given the time 1 choice, there is, in fact, a baserate probability of 66.7% that the initially unchosen item will be rejected again at time 2. However, Chen’s argument rests on the unwarranted assumption that the time 1 choice provides a perfectly reliable measure of subjects’ preference for the chosen item over the unchosen item. With more realistic estimates of the association between preference and choice, Chen’s statistical explanation cannot fully account for the carry-over effect. Alternative experimental methodologies that eliminate Chen’s statistical explanation are discussed.