by Alfie Kohn
Given that success and competition are conceptually distinct, how are they related in the real world? Does competition really make us more motivated to complete tasks? To do these tasks well? To learn better? Let us turn to the evidence on achievement and productivity and then speculate on the reasons behind what it reveals.
ACHIEVEMENT AND COMPETITION
The question posed by this chapter’s title requires a point of reference: Is competition more productive than what? To ask this is to realize that it makes little sense to inquire whether competition has ever led anyone to be productive. It is far more useful to determine whether we are substantially more productive in competitive situations than in other situations—or, better yet, whether competition is sufficiently superior to other arrangements as to outweigh its costs. These other arrangements, as set out in chapter i, are cooperation (working together so that my success is linked to yours) and independent effort (working alone so that my success doesn’t affect yours). Some of the studies that are reviewed below contrast competition with one or the other, but a comparative picture of all three eventually emerges.
What we want to ask is this: Do we perform better when we are trying to beat others than when we are working with them or alone? It is necessary, of course, to specify the nature of the task, the measure of performance, the age and temperament of the subjects, the setting of the experiment, and a dozen other variables. But the evidence is so overwhelmingly clear and consistent that the answer to this question already can be reported: almost never. Superior performance not only does not require competition; it usually seems to require its absence.
This conclusion, to say nothing of the near unanimity of the data, will be astonishing to most readers—even those generally critical of competition. As the preceding chapter noted, we are carefully trained not only to compete but to believe that a competitive arrangement results in superior performance. This belief has practically attained the status of received truth in our society, so a consideration of the evidence could (or should, at any rate) have a profound impact on how our schools and workplaces are structured.
The great majority of studies cited in this chapter, as will quickly become apparent, have to do with education—or, more precisely, with learning tasks (anagrams, card games, problem solving, and so forth). Some are even set in the classroom and use standard curricular material. While subjects range in age from preschool to adult, most tend to be either undergraduates or in the elementary school range. Research measuring other kinds of performance is, unfortunately, in short supply, but what does exist along these lines supports the same conclusion.
Margaret M. Clifford assumed, as many of us would have, that a competitive game would help fifth-grade students to learn a set of vocabulary words. “However, contrary to prediction, neither performance nor retention was noticeably improved,” she reported, and while the competition did seem to spark some interest, it did so mostly among the winners.4 Morton Goldman and his associates discovered that undergraduates solved anagrams more effectively when they were cooperating rather than competing with each other.5 Abaineh Workie found “cooperation significantly more productive than competition” for high school students working on a card game.6 A well-known experiment that Morton Deutsch conducted with college students in 1948 turned up the same result, and, when he returned to the topic twenty-five years later, he was able to cite thirteen other studies that replicated his findings.7
A review of thirteen studies all showing that competition does not get results sounds impressive. But David and Roger Johnson and their colleagues published a far more ambitious meta-analysis (that is, review of others’ findings) in 1981.8 In what is surely the most conclusive survey of its kind, they reviewed 12a studies from 1924 to 1980 (only one of which overlaps with Deutsch’s list, incidentally), including every North American study they could find that considered achievement or performance data in competitive, cooperative, and/or individualistic structures. The remarkable results: 65 studies found that cooperation promotes higher achievement than competition, 8 found the reverse, and 36 found no statistically significant difference. Cooperation promoted higher achievement than independent work in 108 studies, while 6 found the reverse, and 42 found no difference. The superiority of cooperation held for all subject areas and all age groups.9
A number of qualifications about these conclusions have been proposed, some of which appear to be valid. Cooperation is more effective when the group is smaller10 and when the task is more complex (particularly if it involves sophisticated problem solving).11 Cooperation’s relative effectiveness depends on the degree to which subjects have to rely on each other in the means by which they accomplish a task. The more “means interdependent” the task, the more cooperation helps.12 In some instances, it is claimed, competition may produce better results—but only if the task is simple (such as rote decoding or carrying objects) and not interdependent at all. Even this caveat is questionable, however; the Johnsons contend that, at worst, the margin of cooperation’s superiority is reduced in certain tasks.
Currently there is no type of task on which cooperative efforts are less effective than are competitive or individualistic efforts, and on most tasks (and especially the more important learning tasks such as concept attainment, verbal problem solving, categorization, spatial problem solving, retention and memory, motor, guessing-judging-predicting) cooperative efforts are more effective in promoting achievement.13
Some of the older studies, including Deutsch’s, set up the cooperative condition so that the subjects were cooperating with others in their group but the groups were competing with each other. (This is similar to the arrangement in Japanese industry, which lately has attracted considerable attention in this country: employees within a given company work closely with each other and are encouraged to develop loyalty to the company, but the companies continue to compete with each other.) This cooperative/competitive arrangement led some scholars, reasonably enough, to ask whether the greater achievement levels in cooperative settings were not actually due to the intergroup competition. By now enough experimenters have controlled for this variable so that we can be quite certain the answer is no. Unequivocally, “performance benefits [from] cooperative conditions whether [they involve] additional intergroup competition or not,” as Emmy Pepitone wrote in 1980.14 This may be because “students in an intragroup cooperation/intergroup competition situation behave primarily as if the intergroup competition did not exist.”15
In recent years, Deutsch and his associates have investigated not only the way tasks are set up but the way rewards are distributed. Among the possibilities are a winner-take-all system (which is what many contests amount to), a distribution proportional to accomplishment, and an equal distribution. Much as we tend to assume that competing boosts performance, so it is often taken for granted that the first two arrangements provide a crucial incentive for working hard: reserving a desirable reward for the winner is thought to promote excellence. A series of six experiments with Columbia University students, involving tasks that ranged from decoding Japanese poetry to estimating the number of jellybeans in ajar, was devised to test this assumption. The results: When tasks could be performed independently—that is, when there was low means interdependence—the system of distributing rewards had no effect on how good a job they did. There was absolutely no evidence to suggest that people work more productively when rewards are tied to performance than when everyone gets the same reward. But for those tasks where success depends on working together, there was a clear difference. A system of equal rewards, Deutsch discovered, “gives the best results and the competitive winner-take-all system gives the poorest results.”16
Many of the studies I have been reviewing define achievement in a rather conventional, quantified way. Traditional examinations to measure what has been learned in a classroom, for example, are biased in favor of the competitive approach, and this may account for why some studies found no significant differences between competition
and cooperation.17 It is remarkable, then, that even with such measures, competition does not work well.
Once we move from such measures of achievement as speed of performance, number of problems solved, or amount of information recalled, though, and consider the quality of performance, we find that competition fares even worse. Some of the rather primitive experiments in the 1920s that found people work faster at a mechanical task when they are competing nevertheless discovered that the quality of work was poorer under competitive conditions.18 More recent research confirms that “significantly more complex products were made in the cooperative condition than in the competitive condition”19 and that “the discussion process in cooperative groups promotes the discovery and development of higher quality cognitive strategies for learning than does the individual reasoning found in competitive and individualistic learning situations.”20 Creative problem solving was similarly hampered by competition in a study of undergraduates.21 A 1983 German study found that the competitiveness of fourth graders (as measured by a 15-item picture test) correlated negatively with school achievement.22
So far from making us more productive, then, a structure that pits us against one another tends to inhibit our performance. Children simply do not learn better when education is transformed into competitive struggle. To be sure, from the teacher’s perspective it can be seductive to turn a lesson into a competitive game in order to attract and hold students’ attention. But the real appeal of this strategy is that it makes teaching easier, not more effective; it circumvents rather than solves pedagogical problems. The fact that children seem to enjoy it says virtually nothing about how well it teaches them. And even the enjoyment may not be what it appears: the fact that a game is being substituted for the usual lesson—rather than the competitive nature of this game—could account for students’ interest. Many teachers conclude that competition holds attention better even though they have never worked with cooperative alternatives. (Indeed, evidence reviewed in the last chapter shows that children tend to prefer cooperation once they have experienced it.)
Most of American education is highly individualistic, fueled by a competitive structure of evaluation. Sometimes competition (either individual or group) also finds its way into the curriculum itself. Given the unrelenting race for good grades,23 this more explicit use of competition strikes the student as perfectly natural. On the other hand, cooperation of the kind described in all these studies is unfamiliar—no more than an abstraction, really—to many of us, including teachers. It is worth considering in more detail how a cooperative classroom actually works.
Cooperation means more than putting people into groups. It suggests, rather, group participation in a project where the result is the product of common effort, the goal is shared, and each member’s success is linked with every other’s. Practically, this means that ideas and materials, too, will be shared, labor sometimes will be divided, and everyone in the group will be rewarded for successful completion of the task. Aronson, for example, conceived the “jigsaw method” of learning: When the task is to learn about the life of a well-known person, each member of a group is given information about one period of the individual’s life. Members of the group are thus dependent on each other in order to complete the assignment.24 The brothers Johnson, by contrast, offered a straightforward assignment (“How many things can our group find that make a difference in how long [a] candle burns?”) that involved less structured division of labor.25 Elsewhere they suggest setting up groups of students who represent a range of abilities and instructing them to help one another. “When everyone in their group has mastered the solutions, they go look for another group to help until everyone in the class understands how to work the problems.”26
Since “group performance in problem solving is superior to even the individual work of the most expert group members,”27 it should not be surprising that students learn better when they cooperate. But the last technique—having students help one another—raises the question of whether students with lower ability are being helped at the expense of those with higher ability. Is this true? Knowledge, happily, is not a zero-sum product. Anyone who has taught or tutored knows that doing so not only reinforces one’s own knowledge but often pulls one to a more sophisticated understanding of the material. The cliché about teachers’ learning as much as their pupils is quite true, and the tutoring that takes place in a cooperative classroom actually benefits both the helper and the helped more than a competitive or independent study arrangement.
The evidence substantiates this view. As the Johnsons conclude:
There can be little doubt that the low and medium ability students especially benefit from working collaboratively with peers from the full range of ability differences. There is also evidence that high ability students are better off academically when they collaborate with medium and low ability peers than when they work alone; at the worst . . . [they] are not hurt.28
Returning to the question in 1985, the Johnsons cited three studies that found gifted students are helped by such collaboration, one that found no difference, and none that found they were disadvantaged.29 A study of 75 Midwestern second graders published that same year also reported that “high-, medium-, and low-achieving students all academically benefited from participation in heterogeneous cooperative learning groups.”30 Moreover, earlier research discovered that tutoring occurred spontaneously when the consequences for learning were shared by the group—and that this “produced no long-term complaints from the participants in this experiment.” Once again, “even the gifted [benefited,] possibly more than they would have under individual consequences.”31 (The advantages of a cooperative structure in education, incidentally, extend beyond mere performance; chapter 6 will consider the interpersonal consequences of cooperation and competition.)
Most of these studies assess performance in terms of learning, with the schoolroom as the prototype if not the actual setting. But what of the rest of the world? Don’t competitive conditions—or personal competitiveness—lead to better performance?
Research on productivity and competition in traditional work settings has been nowhere near as plentiful as the classroom data. The findings, however, are remarkably consistent with those on learning. One early study, which has become something of a classic in sociological investigation, was conducted by Peter Blau in 1954. Blau compared two groups of interviewers in an employment agency. In one, there was fierce competition to fill job openings. The other group worked cooperatively. Members of the first group, who were personally ambitious and extremely concerned about productivity, hoarded job notifications rather than posting them so everyone could see them, as they were supposed to do. This practice eventually was used defensively and so became self-perpetuating. Members of the second group, by contrast, told each other about vacancies. And it was this second group that ended up filling significantly more jobs—the clear index of performance.32
A quarter century later, Robert L. Helmreich of the University of Texas and his colleagues decided to investigate the relationship between achievement, on the one hand, and such traits as the orientation toward work, mastery (preference for challenging tasks), and competitiveness, on the other. A sample of 103 male Ph.D. scientists were rated on these three factors based on a questionnaire. Achievement, meanwhile, was defined in terms of the number of times their work was cited by colleagues. The result was that “the most citations were obtained by those high on the Work and Mastery but low on the Competitiveness scale.”
This startled Helmreich, who did not expect that competitiveness would have a deleterious effect. Could the result be a fluke? He conducted another study, this one involving academic psychologists. The result was the same. He did two more studies, one involving male businessmen, measuring achievement by their salaries, and the other with 1300 male and female undergraduates, using grade point average as the attainment criterion. In both cases he again found a significant negative correlation between competitiveness and achievement
.33 Of the four studies, the one involving businessmen was particularly “exciting” to him because “the stereotype of the very successful businessman is of someone who is . . . highly competitive”—a stereotype called into serious question by these findings.34 In fact, as he and his colleague Janet Spence later observed, the data “dramatically refute the contention that competitiveness is vital to a successful business career.”35
But Helmreich did not stop there. As of 1985, he had conducted three more studies. The first compared the standardized achievement tests of fifth- and sixth-graders to their competitiveness. They were negatively related.36 The second examined the relationship between performance of airline pilots and competitiveness. The relationship between the latter and superior performance was negative.37 The third looked at airline reservation agents and again found a negative correlation between performance and competitiveness.38 Seven different studies, then, with vastly different populations and measures of success, have all determined that intentional competition is associated with lower performance.
Helmreich’s research is particularly important because it does not rely on a compound personality measure such as the “motive for success.” Rather, it teases out competitiveness from the other components of this variable. Before this was done, researchers simply assumed that all of these components were associated with better performance. Now it appears that competitiveness in particular is not. Other researchers who have used a comparable methodology have found the same thing. In the German study of fourth graders mentioned earlier, just as in Helmreich’s, competitive students did not get better grades. And psychologist Georgia Sassen, in the course of looking for differences between male and female students, also found a slight negative relationship between competitiveness and achievement.39
