sustain

I met with University of Washington Assistant Professor Hendrik Wolff last week to discuss the economics of eco-feedback interfaces. Hendrick has done research on environmental economics and management but has focused largely at the macro scale rather than the micro scale, which is where most of the eco-feedback work fits. One of the focuses of our conversation was the amount of resources that are often necessary to run experiments out in the field rather than in the laboratory. Hendrik mentioned John List, who is a professor at the University of Chicago known for adapting methods that are well established in medical science to the social sciences, mainly, real-world experiments relying on randomized trials.

The New York Times has a really interesting article on John List, which includes a personal biography and some highlights from his more well-known research studies, one of which is on philanthropy–in particular, why do people give? From the article:

Philanthropy in America

For a long time, philanthropy was mostly ignored by social scientists. It’s not an especially large part of the economy, and most charities operate on a shoestring, without the resources to finance research projects. But this is starting to change. Americans gave $295 billion to charity in 2006, equal to 2.2 percent of the country’s gross domestic product, up from about 1.8 percent from the mid-’70s to the mid-’90s, according to the Center on Philanthropy at Indiana University. Most philanthropy still comes in the form of small gifts, but there is also a growing group of donors, like Bill and Melinda Gates, who are interested in bringing some of the quantitative rigor of big business to philanthropy

Charities as Laboratories to Study Human Behavior

Academics, for their part, have come to realize that charities provide an excellent laboratory for studying human behavior, in part because so many of them are desperate for the kind of free-of-charge consulting Karlan was offering. When charities are designing their donor appeals, they often go by nothing more than a few rules of thumb, some of which may be profoundly insightful and others a good deal less so. “I think some fund-raisers have developed terrific intuitions, passed on through the fraternity of fund-raisers,” says Paul Brest, president of the William and Flora Hewlett Foundation in Menlo Park, Calif., which often works with charities. “But a lot of the intuitions don’t work. Look at how much junk mail you get.” Matching gifts were another good example. People figured that they worked, because — well, how could they not? They seem so sensible.

So, John List and Dean Karlan, an economics professor at Yale, put together a field experiment to uncover how well “matching gifts” work in social programs. Matching gift programs work by asking for a donation and touting that some other organization (or person) will match that gift thereby making your original donation much more significant. Most matches are two-to-one (e.g., you donate $100, another organization donates $100–doubling the size of your contribution) but some go up to a four-to-one match.

Earlier Research on Match Gifts

In addition to common sense, some of the earliest economic research on philanthropy supported the idea that matching gifts should make a big difference. In the 1970s, economists began studying the tax deduction for charitable giving, and they found that it clearly affected how much people gave. When tax rates were higher — and deductions were thus more valuable — people gave more. It seemed to follow that they would be equally rational about a match.

The Experiment

Late in 2004, List and Karlan started working on different solicitation letters for a political organization. The letters were similar except for the part that mentioned (or didn’t mention) a match. In one letter, sent to the control group, there was no match. Another letter said that a donor had agreed to match any gift, dollar for dollar. In a third, the match was increased to two to one, and in a fourth it was three to one.

The Results

When Karlan and List got their results, however, they realized that the conventional wisdom about matches was only partly right. The existence of a matching gift did very much matter. In their experiment, 2.2 percent of people who received the match offer made a donation, compared with only 1.8 percent of the control group. That may not seem like a big difference, but it is — more than a 20 percent gap between the two response rates, which is certainly large enough to justify making the effort to solicit a hefty matching gift.

But the size of the match in the experiment didn’t have any effect on giving. Donors who received the offer of a one-to-one match gave just as often, and just as much, as those responding to the three-to-one offer. That was surprising, because a larger match is effectively a deeper discount on a person’s gift. Yet in this case, the deeper discount didn’t make an impact. It was as if Starbucks had cut the price of a latte to $2 and sales didn’t increase.

Why Do People Give?

In the late 1980s, an economist named James Andreoni argued that the internal motives for giving were indeed more important than many people had acknowledged. He came up with a name for his idea — the “warm glow” theory — and it stuck. In the warm-glow view of philanthropy, people aren’t giving money merely to save the whales; they’re also giving money to feel the glow that comes with being the kind of person who’s helping to save the whales.

Andreoni’s argument was a merely theoretical one, but the experiment by List and Karlan suggested that it was correct. Donors did not, in fact, seem to do a rational analysis of how they could best help promote liberalism. And there was one more layer to their results that made the findings even more striking. In blue states — defined as those that voted for John Kerry — even the existence of a matching gift had only a minor effect. It lifted the response rate by about 5 percent. In red states, though, a matching gift increased donations by about 60 percent. For isolated liberals living in states that had just voted for Bush’s re-election, the glow that came from joining up with another liberal seemed to be much stronger. “Giving is not about a calculation of what you are buying,” Karlan said. “It is about participating in a fight.” It is about you as much as it about the effect of your gift. As much as fund-raisers say that they understand these mixed motivations, charities often continue to behave as if donors were automatons. Thus the existence of big matching gifts.

I found this study incredibly compelling for a number of reasons. First, their method allowed them to test a number of conditions at scale in the field. This is the primary principle behind A/B testing and will, no doubt, play a huge role in future eco-feedback systems (e.g., like Google’s PowerMeter and Microsoft’s Hohm) that will allow the designers to quantify the benefit/effectiveness of specific feedback features and interfaces. Second, their results further underline how very irrational humans can be and that we are not, for whatever reason, always motivated to maximize rational economic gain. If you’re interested in the theory of decision making, I recommend Tversky’s The Framing Of Decisions And The Psychology Of Choice, Tversky’s Judgment Under Uncertainty Heuristics And Biases, and Thaler’s Mental Accounting Matters (to name a few). Note that I believe each of these articles rely solely on laboratory experiments to make their arguments. Finally, I’d be interested in knowing whether visualizations of how gift matching works on the letters themselves would have an effect–that is to say, are some people simply not getting the fact that gift matching can make a huge difference?