Last November the U.S. government released a report detailing the devastating impact of global warming for the U.S. economy: they predicted that GDP would shrink by more than 10% by the end of the century if nothing was done to reduce rising temperatures.

The report makes clear that a significant reduction in energy consumption is needed to help meet critical temperature thresholds. New research we conducted points to a way to help consumers work toward this goal – one that doesn’t rest on changing people’s personal beliefs about climate change. It turns out that something else matters even more. Our research suggests that whether you believe your neighbors care about energy conservation is an important motivator for how you consume energy.

We partnered with the utility provider Opower, acquired by Oracle in 2016, to find out what predicts whether someone will or will not reduce their energy consumption. Opower’s flagship product is their Home Energy Report (HER), which tells residential energy customers not just how much energy they use, but also how much energy their neighbors consume. Prior research finds that this single-page document helps customers reduce their energy use, on average about 1-2% per year.

But averages don’t tell the whole story. Every time Opower implements their intervention in a new region or with a new utility company, they conduct a randomized controlled trial, comparing a treatment group that receives the HER with a group that does not. We received access to the 211 randomized controlled trials Opower conducted over the last decade, which included over 16 million households in 27 states.

Puzzlingly, we found that the average reduction in energy usage by households that received the HER varies: in some cases, households achieved a 2.55% reduction in energy consumption, whereas others reached only 0.81%. Why is the impact so inconsistent? The answer has to do with the core principle and motivator baked into every HER: social norms.

The HER is essentially a simple 3-bar graph called the neighbor comparison. It shows recipients how much energy they’re consuming, how much energy “efficient neighbors” consume, and how much “all neighbors” consume. It also rates them on how they’re doing in terms of energy reduction. The graph depicts a “social norm” by communicating how others like them behave––and challenges them to do better if they’re falling short of the neighbors.

Opower expected that the HER would influence consumers equally. After all, decades of social science research show that people generally want to be like their peers (or better), and no one wants to be seen as caring less about the environment than their neighbors. Yet some people seemed to be less motivated by the report than others.

To find out what drove this difference, we surveyed individuals in the 27 states that Opower’s randomized controlled trials were in, asking them two questions: whether they themselves believe that energy conservation helps to save the environment, and whether they believe that the majority of their neighbors believe that saving energy helps to save the environment. We then explored whether the effectiveness of Opower’s randomized controlled trials was associated with participants’ responses within those same states.

Surprisingly, what matters more than one’s own attitudes and beliefs—how concerned we are with our own energy use and the environment—is whether we believe our neighbors view saving energy as important to saving the environment. Receiving the HER with information about your neighbors’ energy consumption has a stronger influence on your own energy use when you believe that your neighbors care about saving energy as an issue.

Imagine receiving this report stating that you consume more energy than your neighbors. You may wonder: “What are my efficient neighbors doing?” And, a beat after, you may think, “Haven’t they been on vacation?” “Aren’t their kids already in college?” “Are they even doing it on purpose?” “They were surely just lucky this month!” If your neighbors are using less energy than you by accident, rather than out of concern for the environment, then perhaps their efficient energy use is hardly a valid basis for you to change light bulbs and turn off the AC.

However, if you believe that your neighbors do deeply care about reducing their energy usage, then you may view your own consumption of more energy through a different lens, asking yourself: “What steps are they taking to conserve energy?” “Perhaps I could ask them for help?” or even, “They seem to be taking this seriously.” In other words, the more you think that your neighbors actually care about saving energy to help the environment, the more you will engage in energy reducing behaviors.

The suggestion here is that the stories we tell ourselves about why others behave the way they do drive our own behavior, as much as our personal attitudes and beliefs and what we observe others doing. The implications of this perspective go beyond energy consumption.

Imagine that you’re a manager, seeking to improve your employees’ work-life balance. As a student of successful management practices and behavioral science, you look at the data and decide to employ a social norm message, such as: “70% of your colleagues don’t check email on nights and weekends” or “85% of your colleagues use all of their vacation days by the end of each year.” Our results imply that you might want to go one step further: You need to help your employees understand that others in the firm not only engage in these behaviors, but also deeply care about them. So you might be better off writing, “We all value time with our families, and so does our firm. That’s why 85% of your colleagues use all of their vacation days by the end of each year.”

Above all, our results remind us that whenever we attempt to change human behavior, we must go one step beyond seeking to change what a person believes, and instead also pay attention to what they think others believe. We are social beings and care deeply about not just what our neighbors and co-workers do but also what they think.