June 10, 2013 Whether it's an effort to increase recycling rates, reduce energy usage or cut carbon emissions, the conventional wisdom says that the best way to get people to do the right thing is to make it worth their while with cold, hard cash.
But Harvard researchers say there may be an easier, cheaper way -- by appealing to people's reputation, not their wallets.
Using enrollment of thousands of people in a California blackout prevention program as an experimental test bed, a team of researchers that included Erez Yoeli, a researcher at the Federal Trade Commission, Moshe Hoffman, a visiting researcher at Harvard's Program for Evolutionary Dynamics (PED), David Rand, formerly a post-doctoral fellow at PED (now a professor at Yale University), and PED Director Martin Nowak, showed that while financial incentives boosted participation only slightly, making participation in the program observable -- through the use of sign-up sheets posted in apartment buildings -- produced a three-fold increase in sign-ups. Their work is described in a study published this week in the Proceedings of the National Academy of Sciences.
"We wanted to see how observability compares to a cash incentive for getting people to act to benefit the common good. The answer is that observability is really dramatically better," Rand said.
Using a cash incentive of $25, the utility company sponsoring the program had seen participation increase from about three percent to four percent. When researchers made people's participation observable, however, participation jumped from three to nine percent. To get the same result using a cash incentive, Yoeli estimated, the company may have had to offer every person as much as $175.
Observability proved to be the key factor in the results, the researchers said, because it puts people's reputation at stake, encouraging those who might not otherwise sign up to do so.
"When people know it's a cooperative effort, they feel peer pressure to take part," Rand explained. "They think, 'If I don't do this, I'm going to look like a jerk.' But if it's not observable, then there's no problem with not participating."
"In fact, we think this is one reason why the Prius, for instance, is such a different-looking car," said Hoffman. "The designers at Toyota seem to have intuitively had this idea, designing a car that didn't look like any other car so your neighbors can tell you're driving a hybrid.
"You can also see this phenomenon when you go to vote, and you get an 'I voted' sticker," he added. "Or when you go to give blood and you get a pin you can put on your backpack."
To directly demonstrate the effect of observability in a highly relevant real world setting, Yoeli and colleagues turned to a large-scale California blackout prevention program. As part of the program, residents were asked for permission to install a monitoring device on their air-conditioning system. If power demand spiked, the device would automatically adjust the air-conditioning temperature to reduce electricity demand and not overload the power grid.
As part of their study, researchers randomly offered people one of two ways to sign up for the program.
In the first, people received a mailer that described the program, and were encouraged to sign up in their apartment building using a unique identification number. In the second, people received a similar mailer and identification number, but also had to write their name and apartment number when they signed up.
"The idea was that in one case, it's anonymous, and in the other, it's observable -- everyone can see who has and hasn't signed up," Rand said. "When participation is observable, people worry about their reputation, and wanting to seem cooperative drives them to sign up."
To support that explanation for the increased sign ups when participation was observable, Yoeli and colleagues point to three pieces of evidence: First, tests showed that the effect was greater in large apartment buildings -- where more people were likely to see the sign-up sheets -- than in row houses, which have less common space.
Second, tests showed that the effect was more pronounced among people who own their apartments than among renters. "People who own their apartments are real, permanent members of their community," Rand said. "They are more likely to care about what other people in the community think of them. If you are a renter, by comparison, you may not even know any of your neighbors."
Finally, Yoeli and colleagues said, tests showed observability only increased participation when the program was portrayed as a community-wide benefit.
"We showed that signing up only matters when others can see it, and particularly when people you care about can see it. But we also wanted to show that it matters whether people think you are being cooperative or not," Rand said. "If you think about the way reputation works, you can get a bad reputation for being selfish, but things that don't involve a cooperative element -- like signing up for a promotion offered by your utility that has no effect on others- aren't going to affect your reputation."
Ultimately, Yoeli and colleagues said, the study suggests that similar strategies could be relatively cheaply and easily employed to boost participation in a host of efforts to improve community-wide efforts.
"This finding is very policy-relevant, because we're talking about changing real-world behaviors that are economically significant," Rand said. "The moral here is that these type of reputation concerns deserve a prominent place in the toolkit used by policy makers to encourage people to do things that benefit the public good. We think that observability and reputation concerns are powerful tools that are being under-utilized."
Other social bookmarking and sharing tools:
Note: Materials may be edited for content and length. For further information, please contact the source cited above.
- Erez Yoeli, Moshe Hoffman, David G. Rand, and Martin A. Nowak. Powering up with indirect reciprocity in a large-scale field experiment. PNAS, June 10, 2013 DOI: 10.1073/pnas.1301210110
Note: If no author is given, the source is cited instead.