Abstract
Our 2003 article clashed with conventional wisdom by concluding that high self-esteem has only a couple of benefits, notably high initiative (based on trusting one’s own judgment) and feeling good. Its high citation rate reflects not only the novel conclusions but also widespread interest in self-esteem both among researchers and in the broader society. Psychology may have lost some credibility by advocating efforts to raise self-esteem that were based on correlational evidence, which may be a salutary lesson for the field. There is still much to learn about self-esteem, but future work can improve by noting weaknesses in self-report data and correcting for confounds.
By the late 1980s, self-esteem had become a household word. California famously funded a task force to promote self-esteem in the hope that it would help solve the state’s crime, welfare, population, and economic problems. People everywhere began to think there was an epidemic of low self-esteem and that a top priority in raising and educating children should be to maximize self-esteem.
This was a tempting opportunity for psychology to shape public discourse and policy. There existed ample research evidence about self-esteem. Many of us had found significant differences, because people with high self-esteem are really different from people with low self-esteem. But are they better or just different? One could easily imagine that very low self-esteem would create problems—but then again, very high self-esteem might also create problems, such as arrogance, entitlement, pigheaded stubbornness, and overconfidence. The view that the society would be better off if everyone got a significant self-esteem boost appeared to underlie the so-called self-esteem movement, but what was the evidence that raising self-esteem would improve outcomes?
In the early 1990s, one of us undertook to write a book on the psychology of evil and violence. The causal role of low self-esteem was well known and frequently cited in the literature. But as he began to look for evidence, it was very difficult to find any original or authoritative statement of that view, or even any systematic evidence. Moreover, when he began to read studies of bullies, tyrants, violent gangs, rapists, and other aggressors, the evidence overwhelmingly indicated that these people typically thought very highly of themselves (see Baumeister, Smart, & Boden, 1996).
Hence raising self-esteem lost some of its promise of being able to reduce aggression. Work in other areas had begun to question some of the other assumed benefits of high self-esteem. Yet some lines of work had continued to uphold the intuitively appealing conclusion that it is better to have high than low self-esteem. At this point, the board of the Association for Psychological Science (APS) was launching a new journal, Psychological Science in the Public Interest (PSPI), as a vehicle for highlighting important advances in psychological research that would be useful to policy makers, politicians, practitioners, and the like. The board came up with the idea that a sweeping, rigorous evaluation of the literature on the benefits of high self-esteem would be a suitable article for this journal. That was the beginning of our articles that were later to make it onto this list of highly cited papers.
Unlike most research publications, authors for PSPI were generally not permitted to choose their own coauthors. APS sought to assemble a team of respected experts with diverse views who would collaborate in open-minded fashion to reach a consensus about the state of the literature. They invited Baumeister to spearhead the group, partly because they liked the open-mindedness apparent in the fact that he had initially been a strong proponent of the importance of self-esteem but had gradually converted into a skeptic, as evidenced by the article on aggression.
Although the authors had done work on self-esteem, most of their prior work had been aimed simply at understanding the differences in emotion, cognition, and behavior between people with high versus low self-esteem. The question this time was different: What are the advantages of high self-esteem?
What We Found
All the authors on the Baumeister, Campbell, Krueger, and Vohs (2003) article changed their views during the process of researching and writing this article. That is perhaps to be expected when one attempts to digest a massive amount of information from a giant literature review. Our conclusions, as stated in the 2003 article, were that high self-esteem does indeed confer some genuine benefits—though far fewer than many of us had believed or hoped.
We focused on multiple areas in which high self-esteem had been thought to yield benefits and advantages. These included student performance in schools and on tests, workplace performance, interpersonal interactions and friendships, romantic relationships, group participation and leadership, aggression, delinquency and antisocial behavior, happiness, coping with stress, teenage sex, alcohol and drug abuse, cigarette smoking, and pathological eating patterns. Apart from bulimia and depression, we did not delve into mental-health issues.
Two strong positive benefits of high self-esteem emerged repeatedly. First, high self-esteem increases initiative, in the sense of acting on one’s own. High self-esteem gives people confidence in their own views and in their ability to do what is right and appropriate. They are more willing than people with low self-esteem to reject advice in favor of following their own views, to speak up to criticize the group, to initiate interpersonal interactions, to experiment with sex and drugs, to initiate aggression (including standing up to bullies and other oppressors), and to respond proactively to initial failure, either by trying again harder or by switching to a different line of endeavor that seems more promising.
The second benefit was that high self-esteem feels good. People with high self-esteem are happier than others, and raising self-esteem generally brings positive feelings (and lowering one’s self-esteem is typically unpleasant). Relationships with depression and stress are complex, and the data did not yield fully consistent or simple answers, but whenever differences were found, they favored people with high self-esteem. We found no studies indicating that people with low self-esteem were happier (than people with high self-esteem) under any circumstances. The well-replicated finding that people with high self-esteem bounce back from failure better than those with low self-esteem may reflect not only the initiative but also this positive emotion aspect: People with high self-esteem seem to have a stock of good feelings that they can draw on to help themselves through difficult times, whereas people with low self-esteem have much less of such a resource and therefore are more discouraged or even devastated by initial failure.
Those two advantages are substantial, but again they are far less than was widely assumed during the peak of the self-esteem movement, and they hardly seem adequate to justify widespread efforts to increase self-esteem throughout the population (or even throughout children). There was no sign that raising self-esteem would make people better students, more moral and upright citizens, better or more stable relationship partners, better leaders, or would incline them to be less aggressive members, more popular, less likely to engage in early experimentation with sex or alcohol or cigarettes, or yield other improvements.
From Where Had the Wrong Impression Come?
The 2003 article thus contradicted what at this point was the conventional wisdom, namely that high self-esteem would confer multiple, diverse, and substantial benefits. When research reaches conclusions at odds with widespread beliefs, there is some obligation on the researchers to offer at least a tentative explanation of why so many people had come to hold those false beliefs. This may be particularly important to understand in the present case, because many psychologists had encouraged educators, parents, therapists, policy makers, and others to increase efforts to raise the self-esteem of children in particular and people in general. The self-esteem movement has ended up being something of a black eye for the profession of psychology.
In retrospect, it was an honest mistake, for the most part. There was ample evidence that high self-esteem correlated with a variety of desirable outcomes. Many people saw findings indicating that students with high self-esteem had better grades in school and leapt to the highly appealing conclusion that if all students’ self-esteem could be raised, they would learn more and faster. The mistake was to confuse correlation with causation, but to refute causation while upholding correlation takes complex, time-consuming, and expensive research. One early study to do this (Bachman & O’Malley, 1986) showed that self-esteem during 10th grade failed to predict grades during 12th grade, whereas grades during 10th grade predicted self-esteem during 12th grade. High self-esteem was thus a result, not a cause, of good school performance (as well as a co-result of some third causes). Very likely some other correlational patterns contributed to a similar illusion. After all, if one has messed up one’s life substantially with drug addiction, crime, unwanted pregnancy, or other misfortunes, some degree of low self-esteem might well be expected and even warranted—but again, self-esteem would be the result rather than the cause.
Most researchers know the difference between correlation and causation, and the salutary tale of the self-esteem movement may be enough to help the field be cautious about making that mistake again. A second factor contributing to the illusion of beneficial high self-esteem may, however, be increasing, given recent methodological shifts, and so it is worth pondering.
People score high in self-esteem by rating themselves favorably on a questionnaire. Almost certainly, some of those high ratings reflect positive assessment about the self—while some of them reflect a broad tendency to give oneself favorable ratings, some of which are not justified. The well-documented “better than average effect” (e.g., Sedikides & Gregg, 2008) is evidence that some people rate themselves more favorably than the objective facts would warrant.
The problem arises when researchers rely on self-reports to test hypotheses about the benefits of high self-esteem. People who like to rate themselves favorably will score high on self-esteem but also high on whatever else is being measured: work performance, popularity, health, interpersonal skills, and leadership potential. Fortunately, we recognized this problem early in our literature search and therefore made a rule to look for objective evidence. Many studies yielded positive correlations between self-esteem and self-reported outcomes, thereby tempting researchers to conclude, incorrectly, that high self-esteem is linked to a wealth of positive outcomes. In fact, we found multiple studies that had collected both objective and self-report data, and almost invariably the benefits of high self-esteem shone brightly in the self-report data but dwindled sharply or vanished entirely in the harsh light of objective evidence. For example, people with self-esteem rate themselves as more intelligent, physically attractive, and interpersonally skilled than other people—but these ostensibly desirable traits were contradicted by IQ test scores, peer ratings of appearance, and roommates’ reports of interpersonal skills (Buhrmester, Furman, Wittenberg, & Reis, 1988; Diener, Wolsic, & Fujita, 1995; Gabriel, Critelli, & Ee, 1994).
We bring this up in the present because the recent methodological shifts are favoring quick, online studies with self-reports, as part of the general decline in observation of actual behavior (Baumeister, Vohs, & Funder, 2007). It is likely to accelerate with the current emphasis on obtaining large samples, an emphasis that makes behavioral work impractical.
In any case, the widespread tendency to accept self-reports at face value undoubtedly contributed to the evidence-based impression that high self-esteem has multiple advantages.
Why the High Impact?
Why has this article attracted so many citations? We can speculatively note multiple factors that probably contributed to its impact. Undoubtedly it helped that it addressed a topic of broad interest to both society (considering how many people are personally or professionally concerned about self-esteem) and scientific psychology (given the great many studies of self-esteem). APS also no doubt boosted our paper’s impact by publicizing the report and engaging the media.
The findings challenged conventional wisdom but also resonated with many people and that too probably contributed to interest in it. The concern with cultivating high self-esteem, in oneself and in children, was widespread but may itself have been a somewhat recent development. Previous generations of American parents had worried about their children becoming conceited or spoiled, and they endeavored to teach them humility. Traditional values in Western civilization were shaped by Christian religion, which has a long tradition of opposing egotism while cultivating humility. Many people have been gratified to learn that reckless, indiscriminate praise and unabashed self-promotion are not the optimal pathway toward good adjustment and life success.
A major review article can function both as a summary and as a challenge. Our 2003 article concluded that the benefits of high self-esteem were limited, and this probably inspired other researchers to search harder for such benefits. Such efforts would tend to increase our citation count, regardless of how they turned out. If they failed to find benefits, they would cite us as consistent with their findings, and if they succeeded, they could cite us as strong evidence for the novelty and importance of their findings.
The limited space precludes any systematic evaluation of subsequent evidence, but as an illustrative example, we cite the impressive article by Orth, Robins, and Widaman (2012), reporting a 30-year longitudinal study. The first article cited in its introduction was our 2003 article, which came up repeatedly throughout. They concluded that they found self-esteem caused significant benefits, contrary to our conclusion. Yet their actual findings fit quite well with ours. Their data were almost exclusively self-reports. Self-esteem mainly affected subjective outcomes, such as relationship satisfaction and depression. The more objective the measure was (e.g., salary, occupational attainment), the less effect self-esteem had. These findings represent an advance over what we had found in the sense that self-esteem at Time 1 did have effects on outcomes at Time 2 but mainly in terms of increasing one’s satisfaction and positive emotions. Despite their large sample, there was no effect whatsoever on occupational status. Thus, high self-esteem leads to being more satisfied with your job but not with getting a better job.
Looking Ahead
Where next for self-esteem research? Our review coincided with society’s losing some of its love for cultivating high self-esteem, which we think is probably a good thing. Still, the many effects of self-esteem (whether good or bad) justify its continued study. Indeed, there is much to learn about effects of self-esteem apart from whether high self-esteem is better than low.
One emphasis for further work should be to untangle self-esteem from its many confounding variables. Boden, Fergusson, and Harwood (2008) reported 10 significant correlations between self-esteem and mental health, substance use, and life satisfaction in early adulthood—but only three remained significant after statistically controlling for associated demographics and personality traits, leading the authors to conclude that adolescent self-esteem is a weak predictor of later life outcomes, which dovetails with our conclusions. On the other hand, Paulhus, Robins, Trzniewski, and Tracy (2004) found that controlling for narcissism increased some effects of self-esteem.
Self-esteem also correlates with self-control (Tangney, Baumeister, & Boone, 2004), and self-control appears to have much more robust and direct benefits. We suspect that many ostensible benefits of self-esteem are in fact based on self-control. We ourselves have shifted our research focus to self-control. We appreciate other researchers continuing to study self-esteem, but they might profitably include self-control measures also in order to assess what is actually responsible for positive outcomes. At present, we would speculate that schools, organizations, and society at large would benefit more by cultivating self-control than self-esteem.
Footnotes
Declaration of Conflicting Interests
The authors declared that there were no conflicts of interest with respect to the authorship or the publication of this article.