Post-Purchase Rationalization: Why the Thing You Bought Improves

A decision creates a problem that the evidence did not. Before the purchase the two options were close, which is what made the decision hard. After the purchase one of them is yours and the other is gone, and the reasons that supported the one you gave up are now reasons to regret. The cheapest repair is to adjust the evaluation: the chosen option gains weight, the rejected one loses it, and the decision stops looking like a coin flip. This page is about that movement in evaluation. The related drift in memory for which option had which feature is covered under choice-supportive bias.

Leon Festinger's theory of cognitive dissonance, published in 1957, is the usual explanation. Holding two things that do not fit, in this case "the option I rejected had real advantages" and "I chose well," is uncomfortable, and the discomfort is reduced by changing whichever of the two is easier to change. The facts of the purchase are fixed. The ranking is not.

In practice the adjustment arrives as reasoning rather than as a feeling. A buyer reads reviews of the model they already own, notices the drawbacks of the one they passed over, repeats the salesman's line about resale value, and discounts the specification where their machine is weaker. None of it feels like rationalization from the inside. It feels like finally understanding the purchase.

How much of the measured effect is really a change of mind is contested. Keith Chen and Jane Risen pointed out in 2010 that the standard experiment has a selection problem: a person's choice reveals a preference that the first rating measured only noisily, so ratings taken afterward will differ even if nothing moved. A design that answers the question has to subtract that artifact. One way is to keep the choice from revealing a preference, by making it blind or settling it with a coin. Another is to measure the same spread without a real choice in the middle: Coppin and colleagues ran the usual rate, choose, rate again sequence against a control group who rated, rated again and chose last, and in 2014 reported an effect of choice on the rejected odors, but not on the chosen odors and not on faces. The pattern is real enough to plan around; its size, and how much of it is a change of mind, are still open.

Jack Brehm published "Postdecision changes in the desirability of alternatives" in the Journal of Abnormal and Social Psychology in 1956, while he was a graduate student with Festinger, and it is the first experimental test of dissonance after a decision. Participants rated a set of consumer products for desirability, were then offered a choice between two of them as a gift, and afterward rated all the products again. In one condition the two items offered had been rated close together, which makes the decision hard, and in another they were far apart.

The second set of ratings had moved. The product each participant had chosen was rated higher than before and the one they had turned down lower, and the movement was larger when the two had been close. The design has been run many times since under the name free-choice paradigm, and the spread between chosen and rejected options is what it measures.

Robert Knox and James Inkster took the question to a racetrack in 1968. They asked bettors how confident they were that their horse would win, some of them in the line before placing the bet and others immediately after. The two groups were betting on the same races with the same information, and the only difference was which side of the ticket window they stood on. Confidence was reported at 3.48 on a seven-point scale before the bet and 4.81 after it.

The caution came much later. Chen and Risen's 2010 analysis showed that rating, choosing and rating again produces a spread even under the assumption that the choice changed nothing, because the choice itself is an extra measurement of a preference the first rating captured imperfectly. Their argument did not show that rationalization does not happen; it showed that the classic design cannot prove it does. Later work has turned to blind choices and coin-flip assignment, which stop the choice from revealing a prior preference, and to control sequences that put the choice after both ratings, which leave that link intact and instead measure how much spread the extra measurement alone produces. Coppin and colleagues used the last of those in 2014 and found a residual effect of choice on the rejected option in one of their two studies.

The adjustment follows commitment rather than purchase, so it appears wherever something has been chosen and cannot be undone cheaply. The rows below are illustrations, not findings from the studies cited on this page.

The adjustment happens after the commitment, so the useful work is done before it and with a record.

• Write the comparison down before you commit, including the advantages of the option you expect to reject, and keep the sheet where you will see it again.
• Write the conditions for regret at the same time: what would have to happen for this to turn out to be the wrong call. That turns a later review into a check rather than a defense.
• Use a return window as a test rather than as insurance. Try the thing against the criteria you wrote, not against your wish to be right.
• Separate the decision from the purchase where you can. A night between choosing and paying gives the comparison one last run before the commitment locks it.
• When you advise someone else, give them your pre-purchase notes rather than your current opinion. Your current opinion has had the adjustment applied.
• Expect it in reverse too. The option you rejected is probably no worse than it was on the day, whatever it looks like now.

Several effects follow a commitment. They differ in what gets adjusted.

• It is not choice-supportive bias, which is the drift in memory for which option had which feature. That changes the facts a person recalls; this changes the value they place on them.
• It is not effort justification, which raises the value of something because reaching it was costly or unpleasant. The cost there is effort rather than the act of choosing, and it applies to things nobody chose between.
• It is not the sunk cost fallacy, which is continuing to spend because of what has already been spent. That is a decision about the next payment; post-purchase rationalization is an opinion about a payment already made.