When does paying people change what they do, and when does it backfire?

Rewards often work, even tiny ones

Three field studies, each with thousands of people, show rewards changing real behaviour outside a lab.

Setting	Reward	What happened
American Red Cross blood drives (Lacetera, Macis & Slonim, 2012)	Small economic rewards	Across nearly 14,000 drives, rewards raised donations, and bigger rewards raised them more. Donors were no more likely to be ineligible.
40 US primary schools, 8,000 children (Loewenstein, Price & Volpp, 2016)	A small prize for eating fruit or vegetables at lunch, for three or five weeks	From a baseline of 39%, the share of children eating a serving doubled. Two months after the prizes stopped, it was still 21% above baseline after three weeks, and 44% above after five.
German Wikipedia (Gallus, 2017)	A purely symbolic award to new editors, chosen at random	Award winners kept editing more than others for the whole following year. The award had no cash value and no career benefit.

The Wikipedia result matters most. A badge on a pseudonymous profile carried no money at all. It worked because it told newcomers they belonged and their work was noticed.

But rewards can backfire

Two famous studies show the opposite.

The day-care fine

Economists Uri Gneezy and Aldo Rustichini studied day-care centres in Israel, where some parents arrived late to collect their children. The centres introduced a fine for lateness. Late pickups went up, not down.

Their explanation, in the title of their paper: "A fine is a price." Before the fine, parents felt they were imposing on the teachers. The fine turned lateness into a service they could buy, with a known and modest cost.

Paying for blood

In 1970 the British social researcher Richard Titmuss argued that paying for blood would drive away donors who gave out of goodwill. In 2008 two Swedish economists tested it. People could sign up as blood donors for nothing, for a payment of about $7, or for $7 they could give to charity.

• For men, the payment made no difference.
• For women, the number signing up fell by almost half when payment was offered.
• Letting people pass the money to charity cancelled the drop completely.

The money did not make blood less needed. It changed what donating said about the donor.

The pattern behind both

A 1999 review of 128 experiments by Edward Deci, Richard Koestner and Richard Ryan found that tangible rewards, given for doing or finishing a task people already found interesting, lowered how much they chose to do it afterwards, on their own time. Children were more affected than university students. Positive feedback, by contrast, raised interest.

This finding is contested. A comment published alongside it questioned how far its conclusions reach. But the field studies above fit its broad shape:

• Rewards help when people are not yet doing the thing, when the reward signals recognition or belonging, and when it tips a choice they were half-inclined to make.
• Rewards hurt when people already act out of duty, care or pride, and the reward recasts that as a trade.

A catch: some gains are borrowed

The Red Cross study also found that many extra donors at rewarded drives had simply left nearby drives with no reward. The more valuable the reward, the more donors moved. Count only the rewarded drives and you overstate what the reward achieved. Any scheme needs to measure the whole system, not just the place where the reward sits.

Using this

1. Before adding a reward or a fine, ask what it will make the act mean. "Thanks for helping" and "here is your fee" are different messages.
2. Try recognition first. Public thanks, awards and badges cost little and can work for a year or more.
3. Use a short reward to start a habit, then remove it. The school fruit study shows effects can last.
4. Don't fine people for something they already feel bad about. You may sell them permission.
5. Let people give the money away. The Swedish donors kept donating when the payment could go to charity.
6. Measure everywhere. A reward that moves behaviour from one place to another has not changed the total.