Every year, right around this time, the same disagreement reemerges:

–Some people claim that on Super Bowl Sunday, domestic violence rates increase by 10%.

–Others claim this is a myth: Domestic violence doesn't spike on any particular day.

Neither claim seems accurate. A 2013 study showed that when an NFL team suffers an upset loss (meaning that they'd been expected to win by more than four points), there's a 10% increase in male-on-female domestic violence in the local community on the same day.

That's a troubling finding. And it illustrates two of the ways our society has been "statisfied", or transformed by modern statistics.

Most noticeable is the way statistical factoids can crop up in almost any discussion. Descriptions of domestic violence rates are commonplace. Same goes for the chances of rain tomorrow, the effectiveness of some vaccine, the performance of students on a national exam, and global mean surface temperature. People didn't talk about such things prior to the 20th century.

Statistics has also transformed society behind the scenes, providing researchers with tools for developing domestic violence interventions, as well as medical treatments, nutritional guidelines, psychotherapeutic techniques, classroom teaching practices, and much more. We're "statisfied" without necessarily being aware of it.

In this newsletter I'll be discussing a new study that provides a great example of behind-the-scenes influence. Statistics allowed the researcher to reveal something about cheating that might not otherwise be observed.

My clickbait-ish title has two meanings: The new study identifies a cause of cheating that's "hidden" in the sense that most people aren't aware of succumbing to it. And, because this is a scientific study, most people aren't aware of the findings. Not yet anyway.

Cheating and motivation

Cheating refers to the breaking of rules for personal gain.

Wherever there are laws, social norms, or other kinds of rules, you'll find cheaters. The contexts include tax returns, academic work, athletic competitions, and romantic relationships. Elon Musk admitted recently that he cheated many times on the video game Diablo IV.

What all forms of cheating have in common is the anticipation of some sort of benefit. People wouldn't break the rules if there were no incentives.

Or would they?

The study I'll be sharing with you illustrates that people will cheat when doing so provides no clear benefits, though the self-deception that's involved may be harmful.

Diagnostic self-deception

Imagine a student who skips class for two weeks and then cheats on the next quiz. If he gets an A on the quiz, should he conclude that he's mastered the material?

Of course not. If he did, he'd be engaging in diagnostic self-deception, meaning that he failed to remember that cheating boosted his performance.

The same sort of self-deception would occur if a person weighs themselves every morning, enters lower numbers on their dieting app, and then overestimates the extent of progress toward their weight-loss goals.

We know these sorts of things happen. In my opinion, they're especially likely in a "statisfied" society, where numerical self-assessments are pervasive. Here's a typical example from earlier this week:

This is from Fox News, but it's not a partisan phenomenon. CNN is loaded with self-assessments too and, as I suggested, we know that people will peek at the answers if they can.

What the new study reveals is that folks will not only peek but succumb to diagnostic self-deception, even though doing so isn't in their best interests.

The new study

This study was published last month in Journal of the Association for Consumer Research. Dr. Sara Dommer, the sole author, is an assistant professor of marketing at Penn State.

Dommer reported four cleverly-designed experiments. The second one, which consists of two parts, offers a particularly clear example of the connection between diagnostic self-deception and cheating.

Part 1

195 adults completed a 10-item IQ test administered online. These folks were recruited from Amazon MTurk, a crowdsourcing website where money is earned for completing discrete tasks online.

Here are a couple of questions from the test:

The answer to the first question is 24. The answer to the second question is that people who create IQ tests are sadistic.

Seriously, the answer to the second question is the third option. You have to rotate the square piece 180 degrees before fitting it into the upside down L piece. (No wonder people cheat on these things.)

Roughly half the participants completed the IQ test under ordinary conditions. In other words, they received no guidance about the correct answers. They simply took the test and estimated their score upon completion. This is the control condition; there's no opportunity to cheat.

The other participants were shown the correct answer to each of the 10 items, so that they could keep track of their score and report it at the end. This is the "cheating condition", because people could easily cheat if they wished.

Part 2

After completing the test, the screen displayed the person's actual score. Now they were asked to predict how well they would perform on a new 10-item IQ test.

This time though, participants in the cheating condition were informed that the correct answers wouldn't be provided.

Participants were also told that they'd receive a small sum of money for correctly guessing their score.

Main findings

No surprise that participants in the cheating condition got higher scores on the first IQ test. The means for the two groups were 8.82 and 5.36 out of 10, respectively.

(The fact that the mean for the cheating condition was high but not 10 out of 10 suggests that these folks cheated but tried not to be conspicuous.)

The results for the second IQ test are where things get interesting.

On this test, where cheating wasn't possible, people from the cheating condition predicted significantly higher scores than the control group did. But the two groups didn't differ significantly in actual scores.

Further analyses confirmed that folks who had cheated on the first IQ test tended to overestimate how well they would do on the second one, even though they knew cheating wouldn't be possible the second time around.

Diagnostic self-deception played a clear role here. People who cheated on the first IQ test got high scores that inflated their perceptions of their own intelligence. The effect was small but significant.

Notice that the cheaters were actually hurting themselves. Since they could've earned money from the second test by correctly estimating their score, they should've been thinking: Well, I cheated the first time, so I can't expect to do well this time around. But their predictions were slightly inflated. It's as if they forgot that cheating contributed to their prior scores.

"To thine own self be true", Polonius advises Laertes. The cheaters should've heeded his advice.

Attributional data

Dommer also reported an experiment in which participants rated the extent to which their performance reflected their own intelligence as opposed to the difficulty of the test.

Cheaters showed a greater tendency to attribute their performance to their own intelligence. They also rated the test itself as a more accurate measure of intelligence.

There's a hint of narcissism here. If I did well on the test, it must be because I'm smart – and because it's a good test.

Why does this happen?

Why would anyone engage in diagnostic self-deception?

Clearly there's potential here for people to feel better about themselves. Whether the outcome is a higher test score, or more pounds lost, or something else, it feels good to perform well. Living in a statisfied society, we tend to engage in a lot of numerical self-assessment.

The evolutionary psychologist Robert Trivers hypothesized that self-deception yields two specific advantages. I'll use the example of cheating on an exam to illustrate:

1. Reduced cognitive load.

Viewing yourself as smart takes less effort than keeping in mind the discrepancy between your true abilities and your cheating-enhanced performance.

This reduction in cognitive load makes it easier to deceive others. If you've fooled yourself into thinking you're smart, it's easier to come to class and maintain the impression that you've broken no rules. In other words, you don't need to work at pretending to be innocent if you actually believe you are.

2. Greater confidence.

Viewing yourself as smart will make you more confident about your performance on the next assessment. Even though you're fooling yourself, you'll feel more confident than you would've been had you acknowledged that cheating overestimated your actual skill.

How does this happen?

How could a person fail to notice that they've cheated and are now overestimating themselves?

In a 2011 paper, Trivers and his colleague William von Hippel hypothesized that people repress the memory of having cheated, while remaining conscious of their performance. Then, repeating the lie to themselves and others ("I aced the test!") further reinforces their false portrayal of the situation.

I reached out to Dr. Dommer for her thoughts on how diagnostic self-deception plays out. She suggested a two-part process: First, we unconsciously persuade ourselves that we haven't cheated. Then, over time, we tend to remember positive details (i.e., our performance) while forgetting the negative ones (i.e., the cheating). Thus, if I've cheated on a test,

"I'm more likely to remember that I scored well, and if I've already convinced myself I'm not cheating, I've totally forgotten that this was even part of it. So even if there's still an inkling back there, if I've deceived myself into believing I wasn't cheating, that is completely erased from the memory... I might just remember that 'Oh, I scored really well'..."

When does this happen?

In Dommer's experiments, cheating is easy because the correct answers are highlighted on the screen. I asked Dr. Dommer what might happen when cheating is a little more involved (e.g., a student posing a question to ChatGPT, then modifying the response to better fit the original question while pruning AI botspeak).

Dr. Dommer had two thoughts on this. First, she indicated that when cheating becomes more effortful, it's harder to deceive oneself about having cheated. Describing an earlier experiment of hers, she noted that

"instead of having the answers right in front of them, students had to flip some papers to see the answers, and I found that this significantly reduced cheating. So I do think that to the extent that cheating becomes more involved, it's harder to engage in that deceptive process."

Dommer added an important qualification: AI support can make it easy to trick ourselves into thinking we're not actually cheating, but rather just harvesting ideas, just as we might do when using Google.

In sum, Dr. Dommer's research shows one reason why people cheat when there are no apparent rewards.

The old adage "when you cheat, you only cheat yourself" seems literally true in this case.

Students who cheat, for instance, not only deprive themselves of learning but may also become overconfident in their abilities. People who cheat on health assessments may end up with a dangerously positive impression of their current health.

How to prevent diagnostic self-deception?

Here are two strategies that strike me as especially promising:

1. Promote uncertainty.

In Dommer's final experiment, participants took a financial literacy test, but some people were informed in advance that most American adults are unable to pass such a test.

This statement reduced the prevalence of cheating among those who had the opportunity to do so, presumably because they were now more motivated to learn the true extent of their financial savviness.

This strategy might not reduce academic dishonesty, but it seems worth incorporating into the instructions for fitness trackers and other health-related assessments that people complete themselves. Perhaps it should be emphasized that most people are unable to accurately guess their own blood pressure, or weekly exercise times, or whatever.

2. Promote mindfulness.

Diagnostic self-deception clearly involves something short of full, reflective consciousness. The person is unaware of what they're doing. Or they're aware, but they soon repress the memory of having cheated.

The most direct solution would be to promote mindfulness, which begins by letting people know that diagnostic self-deception is a natural process but not necessarily a conscious one.

I would be delighted if every high school student were presented with a three-minute synopsis of Dommer's study. Students have heard a lot already about the moral aspects of cheating and how it hurts others as well as oneself. Not all students appreciate abstract discussions about ethics and society. Diagnostic self-deception is a threat to one's own self-interests and thus of immediate interest to everyone.

Thanks for reading!