The (Honest) Truth About Dishonesty

The possibility of using religious symbols as a way to increase honesty has not escaped religious scholars. There is a story in the Talmud about a religious man who becomes desperate for sex and goes to a prostitute. His religion wouldn’t condone this, of course, but at the time he feels that he has more pressing needs. Once alone with the prostitute, he begins to undress. As he takes off his shirt, he sees his tzitzit, an undergarment with four pieces of knotted fringe. Seeing the tzitzit reminds him of the mitzvoth (religious obligations), and he quickly turns around and leaves the room without violating his religious standards.

Adventures with the IRS

Using honor codes to curb cheating at a university is one thing, but would moral reminders of this type also work for other types of cheating and in nonacademic environments? Could they help prevent cheating on, say, tax-reporting and insurance claims? That is what Lisa Shu (a PhD student at Harvard University), Nina Mazar, Francesca Gino (a professor at Harvard University), Max Bazerman (a professor at Harvard University), and I set out to test.

We started by restructuring our standard matrix experiment to look a bit like tax reporting. After they finished solving and shredding the matrix task, we asked participants to write down the number of questions that they had solved correctly on a form we modeled after the basic IRS 1040EZ tax form. To make it feel even more as if they were working with a real tax form, it was stated clearly on the form that their income would be taxed at a rate of 20 percent. In the first section of the form, the participants were asked to report their “income” (the number of matrices they had solved correctly). Next, the form included a section for travel expenses, where participants could be reimbursed at a rate of 10 cents per minute of travel time (up to two hours, or $12) and for the direct cost of their transportation (up to another $12). This part of the payment was tax exempt (like a business expense). The participants were then asked to add up all the numbers and come up with their final net payment.

There were two conditions in this experiment: Some of the participants filled out the entire form and then signed it at the bottom, as is typically done with official forms. In this condition, the signature acted as verification of the information on the form. In the second condition, participants signed the form first and only then filled it out. That was our “moral reminder” condition.

What did we find? The participants in the sign-at-the-end condition cheated by adding about four extra matrices to their score. And what about those who signed at the top? When the signature acted as a moral reminder, participants claimed only one extra matrix. I am not sure how you feel about “only” one added matrix—after all, it is still cheating—but given that the one difference between these two conditions was the location of the signature line, I see this outcome as a promising way to reduce dishonesty.

Our version of the tax form also allowed us to look at the requests for travel reimbursements. Now, we did not know how much time the participants really spent traveling, but if we assumed that due to randomization, the average amount of travel time was basically the same in both conditions, we could see in which condition participants claimed higher travel expenses. What we saw was that the amount of requests for travel reimbursement followed the same pattern: Those in the signature-at-the-bottom condition claimed travel expenses averaging $9.62, while those in the moral reminder (signature-at-the-top) condition claimed that they had travel expenses averaging $5.27.

ARMED WITH OUR evidence that when people sign their names to some kind of pledge, it puts them into a more honest disposition (at least temporarily), we approached the IRS, thinking that Uncle Sam would be glad to hear of ways to boost tax revenues. The interaction with the IRS went something like this: