Hal Arkes and Catherine Blumer stopped people buying season tickets to a theater series and sorted them into three groups: full price, small discount, large discount. The discount came out of a draw and reflected nothing about the plays. Who came to more of them?
Over the next six months, the people who paid full price attended more plays. The discount had been handed out at random and said nothing about the quality of any performance, so there was no reason to behave differently. What drove the gap was reluctance to let money already spent feel wasted.
What makes this experiment good is how little it interferes. People buying season tickets to a theater series were sorted at random into three groups, charged full price, a small discount or a large one. Nobody knew they were in a study.
The randomness is the whole design. Had the discount gone by request or by status, any later difference could be explained by saying the full-price group liked theater more. Because it was drawn by lot, the groups are alike in everything except the sum they handed over.
Over the following six months the full-price group attended more plays. Nothing about the performances differed, and everyone already held a ticket. The gap came only from an amount that had been paid and was not coming back.
From that comes the principle: we treat a as though it could still be rescued, and so it keeps steering decisions it has no business steering. The organizational version is , which is why failing projects keep getting budget. In fairness, some argue the behavior buys a reputation for honoring commitments, and is not purely a mistake.
wat. Theater subscribers got their discount by lottery, and six months later the ones who had paid full price had attended more plays.
wat wat
The same principle explains why organizations keep pouring resources into failing projects, a pattern called escalation of commitment.
How do we know?
Random assignment of the discount is what makes this an experiment rather than an observation, because it rules out the possibility that people who paid more simply liked theater more.
What is still unknown
The money is equally gone either way, so this is a genuine error in reasoning. Some argue it still serves a social purpose, by maintaining a reputation as someone who honors a commitment.
Learn something?
Where to next
What happened when the Dunning-Kruger effect was run on random data?bias
Why do we remember a long experience by only two moments of it?decision making
Is a coin toss really fifty-fifty, or does the hand tilt it?bias