Experiment Now

We should conduct experiments to inform social policy

Posted Jul 20, 2012

Should we do experiments to inform social policy? On the one side of this question are pundits who argue that psychology isn’t really a science and has little to contribute to society. Needless to say I think this view is wrong, as I argued in a recent op-ed piece in the Los Angeles Times. Others, from different quarters, have joined me in arguing for the importance of behavioral science.

The economist Richard Thaler, in a New York Times article, points to advances in Great Britain. David Cameron, the British prime minister, created a Behavioral Insights Team that is headed by a social psychologist (Thaler is an unpaid adviser to the team). The group has had some notable successes, such as finding ways to get more small business owners to pay their taxes on time and getting home owners to take advantage of a government energy savings program. How did they do it? In part by applying lessons from basic research in psychology to policy questions. Most importantly, though, they did their own experiments to find out what works the best. To increase tax collections, for example, they wrote different kinds of appeals, sent them to 140,000 business owners, and sat back and observed which appeals elicited the most payments (the particular letter a business owner got was randomly assigned, of course). As it happened, the appeal that worked the best was one that communicated social norms, telling business owners that a high percentage of businesses in their area paid their taxes on time. Government officials estimate that if this type of appeal were used throughout the country, annual tax revenues would increase by £30 million.

Jim Manzi, in his book Uncontrolled, is also a strong advocate of doing experiments. Manzi is a business CEO who stumbled on the idea that when it comes to understanding and predicting consumer behavior, all the analysts in the world, and all the fancy software, are not as good as an old-fashioned experiment. Many businesses have caught on—Capital One, Google, and Harrah’s Entertainment, for example, conduct thousands of experiments in which consumers are randomly assigned to receive one kind of appeal or another to see what works. Manzi quotes Gary Loveman, the CEO of Harrah’s, as saying that there are three things that will get you fired at his company: stealing, harassing someone, or failing to include a control group.

Manzi argues that we should learn from the business world and conduct experiments to see what works in all segments of society, including economics, education, politics, and criminology. He is well-versed in the history and philosophy of science and makes a persuasive case.

I hope that policy makers across the political spectrum take notice and come to the same conclusion as Thaler and Manzi. It is in everyone’s interest to find out what works the best to address social problems.