Do Reward Stickers Backfire? The Studies, and How to Use a Stamp Calendar

In 1973, Mark Lepper, David Greene and Richard Nisbett published an experiment run at a nursery school on the Stanford University campus. The children were 40 to 64 months old and had already shown an interest in drawing with magic markers in class. The final sample was 51.

Each child went to a separate room and drew for about six minutes. One group was shown a Good Player Award first and asked if they wanted to win one: a card with a gold star, a red ribbon and a space for their name. A second group drew with no mention of any award, then received the same card as a surprise. A third group got nothing.

One to two weeks later, observers watched the classrooms without the children knowing. The share of free time spent with the markers was:

• Promised an award: 8.59%
• No award: 16.73%
• Surprise award: 18.09%

Only the promised group dropped significantly from where it had started. The groups drew about the same number of pictures during the session, but judges rated the promised group’s drawings lower, at 2.18 on a five-point scale against 2.85 and 2.69.

This experiment is a standard citation for the idea that rewards backfire. The same paper warns against reading it too broadly. The authors write that nothing in their data suggests a promised reward will always, or even usually, reduce interest. They deliberately chose an activity children already liked, so the reward was superfluous. Rewards may well help, they add, when interest is very low or when an activity only becomes appealing after some mastery, and banning token systems broadly would be a mistaken overgeneralization.

What 128 studies add up to

In 1999, Edward Deci, Richard Koestner and Richard Ryan published a meta-analysis of 128 experiments. These are their effect sizes (d) for whether people choose the activity in their free time. A negative number means a drop.

• Expected tangible rewards: -0.36 (92 studies)
• Unexpected tangible rewards: 0.01 (9 studies)
• Rewards for doing the task, however it goes: -0.43 for children, -0.21 for college students
• Rewards for finishing the task: -0.36 (20 studies)
• Verbal praise: 0.43 for college students; 0.11 for children, which was not significant (7 studies)

For children, the drop did not fade quickly: studies measuring more than a week later found -0.53. But studies comparing interesting and dull tasks in the same experiment found -0.68 for interesting tasks and a non-significant 0.18 for dull ones.

Not everyone accepts this picture. In a 1996 paper, Robert Eisenberger and Judy Cameron argued that detrimental effects appear only under highly restricted, easily avoidable conditions, and Deci’s team set out why their results differ from earlier analyses, including that one. Researchers have argued over this point.

The sticker trials

Two trials used actual stickers. Cooke and colleagues (2011) worked with 422 children aged 4 to 6. Each child ranked six vegetables, and their fourth choice became the target, offered for 12 tasting sessions. Some children could choose a sticker if they tasted it, some were praised, some simply tasted it, and a control group had none of this.

Liking rose more in all three tasting groups than in the control group, with no difference among the three. The authors say the results gave no support for the idea that rewards reduce liking. Intake stayed higher than control at three months in the sticker and praise groups, but not in the taste-only group. Their explanation is that liking started low and had little room to fall, and that earlier declines came from studies of foods children already liked.

Corsini and colleagues ran a home trial in Adelaide, Australia, with parents who found it hard to get their 4- to 6-year-olds to eat vegetables. Children tasted a vegetable daily for 14 days. In the reward group they put a sticker on a 14-day chart after each taste, choosing from yummy, just okay and yucky faces. Ten or more tastes were reached by 77.2% of the sticker group and 37.95% of the taste-only group. Liking rose in both groups compared with controls, with no significant gap between them, and the authors note the study was underpowered. Parents were told to give mild encouragement and avoid lines like “if you don’t taste it you can’t have a sticker.”

The helping experiment

Other work points the other way. Felix Warneken and Michael Tomasello (2008) tested 36 German children of about 20 months. An adult dropped a pen or similar object and could not reach it. When a child helped, the adult gave a cube for a noisy toy, or said “Thank you,” with the child’s name, and “that’s really nice,” or responded with nothing.

Later, with no reward on offer, children helped on 53% of trials after cubes, 81% after praise and 89% after no response. There were only 12 children per group. The authors also raise the possibility that material rewards help when a child’s inclination to help is very low.

Where the results divide

Read side by side, these studies differ in a few clear ways.

Did the child already like the activity? The drops appear for things children chose on their own, like drawing. They do not appear for disliked vegetables or dull tasks.

Was the reward promised in advance? In both Lepper’s experiment and Deci’s review, promised rewards lowered interest and surprise ones did not. Deci’s team add a caution: children can start to expect the “unexpected” if it comes often.

How was it given? Deci’s review discusses a difference between rewards used to control and rewards that give information, while noting that only one study has tested this for tangible rewards. The line Corsini’s parents were asked to avoid is an example of the controlling kind.

The rewards in these studies were certificates, stickers and toy cubes. None tested a stamp on an app screen.

Praising days or praising what they did

None of the studies read for this article directly compares praising days in a row with praising the work itself. The closest is Melissa Kamins and Carol Dweck (1999). Kindergarteners, 64 of them with a mean age of 5 years 6 months, role-played successes with dolls and were praised by a teacher doll. Children told “You’re a good girl” showed more helpless reactions, including self-blame, in a later setback than children told “You must have tried really hard” or “You found a good way to do it, can you think of other ways that may also work?” The comparison was between kinds of praise. There was no group without praise.

Jennifer Henderlong and Mark Lepper’s 2002 review concludes that praise can raise, lower or leave motivation unchanged. As long as it is seen as sincere, it helps most when it points to causes the child can control, supports autonomy, builds competence without leaning on comparison with others, and conveys standards the child can reach.

In ordinary words, that looks like describing what the child did: “You went slowly around that curve.” Looking at a calendar together and saying “there are four stamps this week” is another plain statement of fact.

The stamp calendar in BrainDojo Kids

BrainDojo Kids is an iPhone and iPad app I make for ages 3 to 6, where children write on the screen with a finger or a pen to practice letters and numbers.

After each finished training game or course stage, whatever the score, the calendar opens. The child taps today’s date and a red seal marked “OK” appears, in the style of the seals used on Japanese documents. Only today can be stamped. Because the screen opens after each finished game or stage, a day can collect several stamps, and from the second one on the seal shows a count. “Back to home” (“Back to map” in courses) closes the screen without stamping. Each player has their own calendar, and it can be opened any time from Calendar on the home screen.

The stamps stay on the calendar as a record. They cannot be traded for anything, they do not open new games, and there is no streak counter. A family arrangement that swaps stamps for snacks or toys would be close to the promised award in the 1973 experiment. For something to talk about, the learning report in the parent area shows the number of characters written and a line such as “Wrote on 5 of the last 7 days.”

The three daily games in Today’s Training and the four Recommended games are free each day, and every other game can be tried once. In the parent settings there is a daily time limit, on by default at 10 minutes. There are no ads and no account to create. The app is free to start, and a Premium subscription (monthly or yearly) opens all courses and games. House rules are covered in Tablet Rules for Preschoolers, and the first session in Getting Started with BrainDojo Kids.