Overjustification Effect

​

What is the Overjustification Effect?

The Overjustification Effect, as counted here, is a smaller share of later classroom free-choice time with magic markers after a child agreed to draw in order to receive a Good Player Award.

If the external justification provided to induce a person to engage in an activity is unnecessarily high and psychologically “oversufficient,” the person might come to infer that his actions were basically motivated by the external contingencies of the situation, rather than by any intrinsic interest in the activity itself.

That sentence is Lepper, Greene, and Nisbett’s statement of the hypothesis in 1973. The everyday picture is two children and one set of markers. One child has seen the card and agreed to draw for it. The other draws and only then receives the same card. A week or two later the card is not in the room. The share that moved, in Table 1, is the first child’s. Psychological Reactance is a push back when a freedom feels threatened. This page is a later play share after an expected prize, not that push.

​

Overjustification Effect in 3 Depths

• Beginner: In 1973 the later free-choice means were 8.59 percent, 16.73 percent, and 18.09 percent. The low cell is the award the child expected. These are shares of time while the markers were out. They are not baseline shares.
• Practitioner: Write all three cells before you quote 8.59. The surprise award sits next to no award. A spoken “You really did a good job” was said in every cell, including the two that did not drop.
• Advanced: Deci, Koestner, and Ryan’s 1999 abstract puts engagement-contingent, completion-contingent, and performance-contingent rewards at d = −0.40, −0.36, and −0.28 on free choice. Unexpected tangible rewards in their Figure 1 are d = 0.01 across 9 studies. Keep 8.59 on the nursery row.

​

Origin

The name Overjustification Effect is the hypothesis Lepper, Greene, and Nisbett stated in 1973, as an implication of self-perception theory. The paper says self-perception theory was offered as another reading of the dissonance studies of insufficient justification, and it cites Daryl J. Bem, Psychological Review 74, 1967, pages 183–200, and Bem’s 1972 chapter. Those two were not re-read. It also cites Elliot Aronson, 1966, for insufficient justification. That chapter was not re-read. Insufficient justification, in the paper’s own contrast, is a small external reason for an unpleasant or inconsistent act, after which people infer that they wanted to act. Cognitive Dissonance is the discomfort those studies start from. The drawing result is the other direction: an external reason stronger than the drawing required. The 1973 paper does not give this result Festinger’s name. Edward L. Deci, Journal of Personality and Social Psychology 18(1), April 1971, pages 105–115, is the study the 1973 paper calls the first intentional test of a closely related idea. The 1971 PDF was not read as text. The PsycInfo abstract says two laboratory experiments and one field experiment used 24, 24, and 8 undergraduates. Money as an external reward tended to decrease intrinsic motivation. Verbal reinforcement and positive feedback tended to increase it. Rewards were given to the experimental students in the second period only. The abstract does not assign those three head counts to one table. Deci, Koestner, and Ryan, 1999, describe that first experiment as completion-contingent: students worked on four forms of a three-dimensional puzzle called SOMA and received $1 for each puzzle completed in the time allowed. No second count from 1971 is on this page. The counted means are the 1973 nursery study, below. The publisher record for that paper is 10.1037/h0035519. The PDF read here does not print a DOI. Lepper and Greene were at Stanford University. Nisbett was at the University of Michigan. Edward L. Deci, Richard Koestner, and Richard M. Ryan, Psychological Bulletin 125(6), November 1999, pages 627–668, meta-analyzed 128 studies. The author note places Deci and Ryan at the University of Rochester and Koestner at McGill University. The publisher record is 10.1037/0033-2909.125.6.627. The abstract says engagement-contingent, completion-contingent, and performance-contingent rewards undermined free-choice intrinsic motivation, d = −0.40, −0.36, and −0.28. Positive feedback enhanced free-choice behavior, d = 0.33, and self-reported interest, d = 0.31. Judy Cameron, Katherine M. Banko, and W. David Pierce, The Behavior Analyst 24(1), Spring 2001, pages 1–44, read the same literature as limited. Their abstract says negative effects appear on high-interest tasks when rewards are tangible, expected, and loosely tied to performance, and that rewards linked to the level of performance increase intrinsic-motivation measures or do not differ from a nonrewarded control. The publisher record is 10.1007/BF03392017. Their tables were not re-read. No d from that paper is on this page.

​

Key Points

Four readings keep each number on the row that produced it. A later play share is not a quality score, and a pooled d is not the nursery cell.

​

Applications

Write the comparison cell, the timing, and the outcome before an 8.59 leaves the nursery.

​

Case Study

At Bing Nursery School, on the Stanford University campus, Lepper, Greene, and Nisbett watched children ages 40 to 64 months during the first hour of three class days. The children were predominantly white and middle class. The paper says they were of average or above-average intelligence and prints no test score for that phrase. Three Black children who would otherwise have been included were excluded so the population would be easier to define. Teachers set magic markers and drawing paper, materials not usually in the rooms, on a hexagonal table in front of a one-way mirror. Crayons, scissors, and other paper were put away while the markers were out. Adults were asked not to sit at the table. A child counted as interested while sitting at the table or holding a marker. Two observers’ records for 47 children correlated r = .99. The question was whether agreeing to draw for a Good Player Award would lower later free play with the same markers. Of 139 children present at some point in the three hours, 102 played at least a little. Children whose total play exceeded 4 minutes were blocked by class and sex. The potential pool was 24 boys and 45 girls. Twelve of the 45 girls and 2 of the 24 boys refused the “surprise room” three times. Fifty-five children drew: 19 expected award, 19 unexpected award, and 17 no award. Each child had 6 minutes, with up to 30 extra seconds. The second experimenter did not know the condition for the first 5 minutes. The folder opened at 5 minutes showed only whether an award would be given, so the expected-award and unexpected-award children could not be told apart in that last minute. Four children were missing from the later observation, three absent throughout and one present for 10 of 180 minutes. The analyzed sample is 51 children, 19 boys and 32 girls: 18, 18, and 15 in the three conditions. The paper reports no significant sex difference and no sex-by-condition interaction. The later means were 8.59, 16.73, and 18.09 percent. The expected-award group also fell from its own baseline, t = 2.61, p below .02. The other two groups showed small increases, and both of those t values were below 1. The baseline percents are not printed, so 8.59 is not a subtraction from a known starting share. During the session, expected-award drawings scored 2.18 against 2.85 and 2.69, while the picture counts stayed close. Among children below the median of initial interest, only the unexpected-award half showed a substantial rise, t = 2.35, p below .05. The paper says both halves of the expected-award group declined, and that the data do not separate a stronger effect in the high-interest half from a floor under the low-interest half. No half’s mean is printed. The lesson is narrow. An award the child had agreed to win went with less later free play, and with lower-rated drawings in the session, in this nursery sample. The same award, delivered after the drawing, did not. The authors write that the study does not say an expected reward always, or even usually, reduces interest. They chose an activity the children already liked, so the prize was superfluous. They say it would be a mistaken overgeneralization to ban token economies from that result. The token-economy papers they cite were not re-read. No count from those papers is on this page.

​

Boundaries and Failure Modes

The Overjustification Effect stops being the nursery result as soon as the prize was a surprise, the task started dull, or the outcome is something other than later free choice. Boundary 1: A card after the drawing did not lower the share. The unexpected-award mean was 18.09 percent, beside 16.73 with no card. Figure 1 of the 1999 review puts unexpected tangible rewards at d = 0.01 across 9 studies. Task-noncontingent rewards, given for something other than doing the task, are d = −0.14 across 7 studies, interval from −0.39 to 0.11, and are not marked significant. Verbal rewards on free choice are d = 0.33 across 21 studies. The spoken “good job” in 1973 sat in every room. Boundary 2: Performance-tied pay is a split row, and the reviews disagree. Deci, Koestner, and Ryan’s overall performance-contingent free-choice result is d = −0.28 across 32 studies. The same paper’s split is d = −0.15 when everyone got the maximum reward and the control heard no feedback, d = −0.88 across 6 studies after one outlier was removed from a 7-study pool of less-than-maximum rewards, d = −0.20 when the control heard the same positive feedback, and d = −0.03 across 3 studies when the reward signaled poor performance and the control heard the same negative feedback. That last interval, from −0.37 to 0.31, includes zero. Cameron, Banko, and Pierce’s 2001 abstract says rewards linked to performance increase the measures or do not differ from no reward. Deci, Ryan, and Koestner, Review of Educational Research 71(1), Spring 2001, pages 43–51, answer that Cameron’s comment did not show a problem in their meta-analysis. That reply was not re-read past the abstract. The two abstracts do not become one number. Common misuse: Reading 8.59 percent as proof that pay removes love of the work, in an office, a classroom sticker chart, or a household. The 8.59 is one later cell for children ages 40 to 64 months at one campus nursery, on markers that were new to the room. The expected-award children were still at 8.59 percent, not at zero. The authors’ own discussion refuses the jump to a general ban.

​

Common Misconceptions

The name travels farther than Table 1. Each claim below has a row that already answers it.

​

Related Concepts

The later play share sits beside a bounty that changes the act, a measure that becomes the target, a score used as a decision, a push back against a threat, and a higher value on something one made.

​

One-Line Takeaway