Regardless of when you went to college, or whether you went at all, there’s likely a strong association in your mind between the American college experience and drinking. A freewheeling culture of keg parties and mass alcohol consumption is central to nearly every college movie, joke and meme. And as a result, many students feel like fitting in means drinking hard. But the truth is that college students have a warped understanding of how much their peers drink. They vastly overestimate the “normal” rate of alcohol consumption on campus, and, unfortunately, the more students think their friends drink, the more they will drink themselves.

University administrations are acutely aware of this trend, and virtually all of them have attempted to address it with an intervention system called Personalized Normative Feedback, or PNF. The theory behind this practice is that when students are exposed to the true rate of alcohol consumption at their school, they will change their own drinking habits to bring their behavior in line with with their new understanding of the norm. Typically, a college PNF program involves asking students to estimate the percentage of their classmates they believe drink on a regular basis. Once the students answer, they are given the true percentage, as determined by nationally representative surveys, along with a graph to help them appreciate the difference between their guess and the reality. More often than not, these graphs reveal that the respondents’ initial figure was much too high.

Studies show that a college’s use of PNF predicts small to moderate reductions in students’ future alcohol use—not bad for something that takes five minutes and can be done on a smartphone. Not bad—but not great either.

The main problem with this approach is that the heaviest student drinkers are usually also the ones most resistant to change, and in fact, they often increase their alcohol consumption after receiving a dose of PNF. The reason appears to be something called psychological reactance, a term coined by psychologist Jack Brehm in 1966 to describe a fundamental problem in the way human beings experience efforts by others to influence their behavior.

Chances are you’ve experienced psychological reactance yourself throughout your life—for example, when a grown-up told you not to jump on the bed, and you promptly threw your shoes to the side and leaped onto the mattress. Essentially, it is an emotional response to a belief that one’s autonomy is being impinged on or threatened, and it typically involves an impulse to go out of one’s way to break the offending rule and thereby regain freedom and independence. The response explains both the teenager who sneaks out when grounded and the irritation you feel when a co-worker gives you “tips” on how to be “more efficient.” When people experience psychological reactance, they generally become more set in their ways—even if they’re being told those ways are “wrong.”

Hence the limits of college PNF programs, which students, of course, recognize as efforts to shape their behavior. The heavier drinkers among them, who tend to think of alcohol as a large part of their identity, are particularly likely to see PNF as a threat to their freedom and to resist change.

Given the pervasiveness of this kind of resistance in human life, and the toll it takes on our ability to interact with one another and to give and take advice and criticism, my colleagues and I wanted to see if we could find a way for people to bypass psychological reactance altogether.

This was the line of thought that inspired the study we conducted at Loyola Marymount University’s HeadsUP Labs in 2018. Using simple gaming mechanics and a virtual slot machine, we hoped to introduce PNF to heavy-drinking college students without triggering a negative response.

Our hypothesis was that by making PNF delivery seem more “random,” we could decrease feelings of psychological reactance and help students hear advice more openly. We speculated that if they believed drinking statistics were being presented to them purely by chance, they might see the message simply as helpful information rather than manipulation by know-it-all adults. Our findings bore this supposition out and implied that perceived randomness might actually play an important role in how, and how much, people heed unsolicited advice geared at improving their well-being.

To test our hunch, we conducted an experiment involving 138 college students at a midsize private university. All of these students were asked to make guesses about their classmates’ alcohol consumption. They were then given feedback about the accuracy of their estimates, just as in typical PNF studies.

But there was a twist. For one group of students, the study was disguised as an online game that featured a blue slot machine—our own covert agent of randomness.

When these students logged on to the study’s Web site, they learned they were going to be playing a game that involved guessing their classmates’ habits and behaviors across a variety of topics.

Next, our slot machine appeared on the screen with the accompanying text “Selecting your topics ....” The machine would whir with options for “social media,” “sex/relationships,” “drugs,” “grades,” “television” and, of course, “alcohol.” The device implied a degree of randomness, but in fact, every single student was given the topic “alcohol.” Next the students proceeded to answer their questions and view their feedback.

For the other group of students, there was no slot machine and no mention of other possible topics. When these students logged on to the study, they were informed they would be playing a game that involved guessing and receiving feedback on a topic chosen by the researchers, which was revealed to be “alcohol.” These students then answered the exact same questions as their peers in the slot machine group and received identical feedback.

The results were precisely what we’d speculated they would be. A survey taken after the game revealed students using the slot machine experienced significantly lower levels of psychological reactance. And when we followed up with them a few weeks later, they had reduced their alcohol consumption significantly more than the control group did. This finding implies that the slot machine—and the randomness it suggested—broke down the students’ mental barriers and allowed them to view PNF with an open mind.

In other words: when advice seems random, or “organic,” rather than intentional, it is more likely to be accepted and to have a positive impact on its audience.

Our study is simply a jumping-off point for more research on the matter, but its implications are worth exploring. If we can find ways to make advice about people’s negative habits feel more “random,” we may be able to convince others to accept help they would otherwise reject.

Parents facing rebellious, problematic behaviors from their children could use this approach, as could people whose loved ones struggle with alcoholism, drug abuse, eating disorders, and other serious maladies and mental illnesses. It could also have uses in the workplace, where managers would surely find it helpful in addressing performance issues. We might all benefit from the art of making meaningful conversations seem less like premeditated confrontations and more like the sharing of natural “random” thoughts.

Perhaps the simplest application of this research is to try prefacing your advice with a reference to whatever it was that made you think of it. Instead of “I want to talk about your drinking,” a parent could say something like, “I heard an interesting story on the radio this morning about underage drinking. Can we go for a walk and talk about it?” Likewise, rather than “I have some advice on how to format your reports,” a manager might instead use a line like “Reading your recent report made me remember a couple of things I’ve forgotten to teach you. Can you stop by my office this afternoon to discuss them?”

By highlighting what spurred you to think of the advice in the first place, you may be able to add a small but authentic feeling of randomness to the encouter that will cause that advice to be accepted with less resistance. The research is ongoing, and this particular application isn’t something that has been empirically tested, so I can’t guarantee it will work. It’s just something I thought of randomly last night, so I wanted to pass it along.