How (Un)ethical Are You?

by Mahzarin R. Banaji, Max H. Bazerman, and Dolly Chugh

ANSWER TRUE OR FALSE: “I am an ethical manager.”

If you answered “true,” here’s an uncomfortable fact: You’re probably not. Most of us believe that we are ethical and unbiased. We imagine we’re good decision makers, able to objectively size up a job candidate or a venture deal and reach a fair and rational conclusion that’s in our, and our organization’s, best interests. But more than two decades of research confirms that, in reality, most of us fall woefully short of our inflated self-perception. We’re deluded by what Yale psychologist David Armor calls the illusion of objectivity, the notion that we’re free of the very biases we’re so quick to recognize in others. What’s more, these unconscious, or implicit, biases can be contrary to our consciously held, explicit beliefs. We may believe with confidence and conviction that a job candidate’s race has no bearing on our hiring decisions or that we’re immune to conflicts of interest. But psychological research routinely exposes counterintentional, unconscious biases. The prevalence of these biases suggests that even the most well-meaning person unwittingly allows unconscious thoughts and feelings to influence seemingly objective decisions. These flawed judgments are ethically problematic and undermine managers’ fundamental work—to recruit and retain superior talent, boost the performance of individuals and teams, and collaborate effectively with partners.

This article explores four related sources of unintentional unethical decision making: implicit forms of prejudice, bias that favors one’s own group, conflict of interest, and a tendency to overclaim credit. Because we are not consciously aware of these sources of bias, they often cannot be addressed by penalizing people for their bad decisions. Nor are they likely to be corrected through conventional ethics training. Rather, managers must bring a new type of vigilance to bear. To begin, this requires letting go of the notion that our conscious attitudes always represent what we think they do. It also demands that we abandon our faith in our own objectivity and our ability to be fair. In the following pages, we will offer strategies that can help managers recognize these pervasive, corrosive, unconscious biases and reduce their impact.

Implicit Prejudice: Bias That Emerges from Unconscious Beliefs

Most fair-minded people strive to judge others according to their merits, but our research shows how often people instead judge according to unconscious stereotypes and attitudes, or “implicit prejudice.” What makes implicit prejudice so common and persistent is that it is rooted in the fundamental mechanics of thought. Early on, we learn to associate things that commonly go together and expect them to inevitably coexist: thunder and rain, for instance, or gray hair and old age. This skill—to perceive and learn from associations—often serves us well.

But, of course, our associations only reflect approximations of the truth; they are rarely applicable to every encounter. Rain doesn’t always accompany thunder, and the young can also go gray. Nonetheless, because we automatically make such associations to help us organize our world, we grow to trust them, and they can blind us to those instances in which the associations are not accurate— when they don’t align with our expectations.

Because implicit prejudice arises from the ordinary and unconscious tendency to make associations, it is distinct from conscious forms of prejudice, such as overt racism or sexism. This distinction explains why people who are free from conscious prejudice may still harbor biases and act accordingly. Exposed to images that juxtapose black men and violence, portray women as sex objects, imply that the physically disabled are mentally weak and the poor are lazy, even the most consciously unbiased person is bound to make biased associations. These associations play out in the workplace just as they do anywhere else.

In the mid-1990s, Tony Greenwald, a professor of psychology at the University of Washington, developed an experimental tool called the Implicit Association Test (IAT) to study unconscious bias. A computerized version of the test requires subjects to rapidly classify words and images as “good” or “bad.” Using a keyboard, test takers must make split-second “good/bad” distinctions between words like “love,” “joy,” “pain,” and “sorrow” and at the same time sort images of faces that are (depending on the bias in question) black or white, young or old, fat or thin, and so on. The test exposes implicit biases by detecting subtle shifts in reaction time that can occur when test takers are required to pair different sets of words and faces. Subjects who consciously believe that they have no negative feelings toward, say, black Americans or the elderly are nevertheless likely to be slower to associate elderly or black faces with the “good” words than they are to associate youthful or white faces with “good” words.

Since 1998, when Greenwald, Brian Nosek, and Mahzarin Banaji put the IAT online, people from around the world have taken over 2.5 million tests, confirming and extending the findings of more traditional laboratory experiments. Both show implicit biases to be strong and pervasive. (For more information on the IAT, see “Are You Biased?”.)

Biases are also likely to be costly. In controlled experiments, psychologists Laurie Rudman at Rutgers and Peter Glick at Lawrence University have studied how implicit biases may work to exclude qualified people from certain roles. One set of experiments examined the relationship between participants’ implicit gender stereotypes and their hiring decisions. Those holding stronger implicit biases were less likely to select a qualified woman who exhibited stereotypically “masculine” personality qualities, such as ambition or independence, for a job requiring stereotypically “feminine” qualities, such as interpersonal skills. Yet they would select a qualified man exhibiting these same qualities. The hirers’ biased perception was that the woman was less likely to be socially skilled than the man, though their qualifications were in fact the same. These results suggest that implicit biases may exact costs by subtly excluding qualified people from the very organizations that seek their talents.

Legal cases also reveal the real costs of implicit biases, both economic and social. Consider Price Waterhouse v. Hopkins. Despite logging more billable hours than her peers, bringing in $25 million to the company, and earning the praise of her clients, Ann Hopkins was turned down for partner, and she sued. The details of the case reveal that her evaluators were explicitly prejudiced in their attitudes. For example, they had commented that Ann “overcompensated for being a woman” and needed a “course at charm school.” But perhaps more damning from a legal standpoint was blunt testimony from experimental research. Testifying as an expert witness for the defense, psychology professor Susan Fiske, now at Princeton University, argued that the potential for biased decision making is inherent in a system in which a person has “solo” status—that is, a system in which the person is the only one of a kind (the only woman, the only African-American, the only person with a disability, and the like). Judge Gerhard Gesell concluded that “a far more subtle process [than the usual discriminatory intent] is involved” in the assessments made of Ann Hopkins, and she won both in a lower court and in the Supreme Court in what is now a landmark case in discrimination law.

Likewise, the 1999 case of Thomas v. Kodak demonstrates that implicit biases can be the basis for rulings. Here, the court posed the question of “whether the employer consciously intended to base the evaluations on race or simply did so because of unthinking stereotypes or bias.” The court concluded that plaintiffs can indeed challenge “subjective evaluations which could easily mask covert or unconscious race discrimination.” Although courts are careful not to assign responsibility easily for unintentional biases, these cases demonstrate the potential for corporate liability that such patterns of behavior could unwittingly create.

In-Group Favoritism: Bias That Favors Your Group

Think about some of the favors you have done in recent years, whether for a friend, a relative, or a colleague. Have you helped someone get a useful introduction, admission to a school, or a job? Most of us are glad to help out with such favors. Not surprisingly, we tend to do more favors for those we know, and those we know tend to be like ourselves: people who share our nationality, social class, and perhaps religion, race, employer, or alma mater. This all sounds rather innocent. What’s wrong with asking your neighbor, the university dean, to meet with a coworker’s son? Isn’t it just being helpful to recommend a former sorority sister for a job or to talk to your banker cousin when a friend from church gets turned down for a home loan?

Few people set out to exclude anyone through such acts of kindness. But when those in the majority or those in power allocate scarce resources (such as jobs, promotions, and mortgages) to people just like them, they effectively discriminate against those who are different from them. Such “in-group favoritism” amounts to giving extra credit for group membership. Yet while discriminating against those who are different is considered unethical, helping people close to us is often viewed favorably. Think about the number of companies that explicitly encourage this by offering hiring bonuses to employees who refer their friends for job opportunities.

But consider the finding that banks in the United States are more likely to deny a mortgage application from a black person than from a white person, even when the applicants are equally qualified. The common view has been that banks are hostile to African-Americans. While this may be true of some banks and some loan officers, social psychologist David Messick has argued that in-group favoritism is more likely to be at the root of such discriminatory lending. A white loan officer may feel hopeful or lenient toward an unqualified white applicant while following the bank’s lending standards strictly with an unqualified black applicant. In denying the black applicant’s mortgage, the loan officer may not be expressing hostility toward blacks so much as favoritism toward whites. It’s a subtle but crucial distinction.

The ethical cost is clear and should be reason enough to address the problem. But such inadvertent bias produces an additional effect: It erodes the bottom line. Lenders who discriminate in this way, for example, incur bad-debt costs they could have avoided if their lending decisions were more objective. They also may find themselves exposed to damaging publicity or discrimination lawsuits if the skewed lending pattern is publicly revealed. In a different context, companies may pay a real cost for marginal hires who wouldn’t have made the grade but for the sympathetic hiring manager swayed by in-group favoritism.

In-group favoritism is tenacious when membership confers clear advantages, as it does, for instance, among whites and other dominant social groups. (It may be weaker or absent among people whose group membership offers little societal advantage.) Thus for a wide array of managerial tasks—from hiring, firing, and promoting to contracting services and forming partnerships—qualified minority candidates are subtly and unconsciously discriminated against, sometimes simply because they are in the minority: There are not enough of them to counter the propensity for in-group favoritism in the majority.

Overclaiming Credit: Bias That Favors You

It’s only natural for successful people to hold positive views about themselves. But many studies show that the majority of people consider themselves above average on a host of measures, from intelligence to driving ability. Business executives are no exception. We tend to overrate our individual contribution to groups, which, bluntly put, tends to lead to an overblown sense of entitlement. We become the unabashed, repeated beneficiaries of this unconscious bias, and the more we think only of our own contributions, the less fairly we judge others with whom we work.

Lab research demonstrates this most personal of biases. At Harvard, Eugene Caruso, Nick Epley, and Max Bazerman recently asked MBA students in study groups to estimate what portion of their group’s work each had done. The sum of the contribution by all members, of course, must add up to 100%. But the researchers found that the totals for each study group averaged 139%. In a related study, Caruso and his colleagues uncovered rampant overestimates by academic authors of their contribution to shared research projects. Sadly, but not surprisingly, the more the sum of the total estimated group effort exceeded 100% (in other words, the more credit each person claimed), the less the parties wanted to collaborate in the future.

Likewise in business, claiming too much credit can destabilize alliances. When each party in a strategic partnership claims too much credit for its own contribution and becomes skeptical about whether the other is doing its fair share, they both tend to reduce their contributions to compensate. This has obvious repercussions for the joint venture’s performance.

Unconscious overclaiming can be expected to reduce the performance and longevity of groups within organizations, just as it diminished the academic authors’ willingness to collaborate. It can also take a toll on employee commitment. Think about how employees perceive raises. Most are not so different from the children at Lake Wobegon, believing that they, too, rank in the upper half of their peer group. But many necessarily get pay increases that are below the average. If an employee learns of a colleague’s greater compensation—while honestly believing that he himself is more deserving—resentment may be natural. At best, his resentment might translate into reduced commitment and performance. At worst, he may leave the organization that, it seems, doesn’t appreciate his contribution.

Conflict of Interest: Bias That Favors Those Who Can Benefit You

Everyone knows that conflict of interest can lead to intentionally corrupt behavior. But numerous psychological experiments show how powerfully such conflicts can unintentionally skew decision making. (For an examination of the evidence in one business arena, see Max Bazerman, George Loewenstein, and Don Moore’s November 2002 HBR article, “Why Good Accountants Do Bad Audits.”) These experiments suggest that the work world is rife with situations in which such conflicts lead honest, ethical professionals to unconsciously make unsound and unethical recommendations.

Physicians, for instance, face conflicts of interest when they accept payment for referring patients into clinical trials. While, surely, most physicians consciously believe that their referrals are the patient’s best clinical option, how do they know that the promise of payment did not skew their decisions? Similarly, many lawyers earn fees based on their clients’ awards or settlements. Since going to trial is expensive and uncertain, settling out of court is often an attractive option for the lawyer. Attorneys may consciously believe that settling is in their clients’ best interests. But how can they be objective, unbiased judges under these circumstances?

Research done with brokerage house analysts demonstrates how conflict of interest can unconsciously distort decision making. A survey of analysts conducted by the financial research service First Call showed that during a period in 2000 when the Nasdaq dropped 60%, fully 99% of brokerage analysts’ client recommendations remained “strong buy,” “buy,” or “hold.” What accounts for this discrepancy between what was happening and what was recommended? The answer may lie in a system that fosters conflicts of interest. A portion of analysts’ pay is based on brokerage firm revenues. Some firms even tie analysts’ compensation to the amount of business the analysts bring in from clients, giving analysts an obvious incentive to prolong and extend their relationships with clients. But to assume that during this Nasdaq free fall all brokerage house analysts were consciously corrupt, milking their clients to exploit this incentive system, defies common sense. Surely there were some bad apples. But how much more likely it is that most of these analysts believed their recommendations were sound and in their clients’ best interests. What many didn’t appreciate was that the built-in conflict of interest in their compensation incentives made it impossible for them to see the implicit bias in their own flawed recommendations.

Trying Harder Isn’t Enough

As companies keep collapsing into financial scandal and ruin, corporations are responding with ethics-training programs for managers, and many of the world’s leading business schools have created new courses and chaired professorships in ethics. Many of these efforts focus on teaching broad principles of moral philosophy to help managers understand the ethical challenges they face.

We applaud these efforts, but we doubt that a well-intentioned, just-try-harder approach will fundamentally improve the quality of executives’ decision making. To do that, ethics training must be broadened to include what is now known about how our minds work and must expose managers directly to the unconscious mechanisms that underlie biased decision making. And it must provide managers with exercises and interventions that can root out the biases that lead to bad decisions.

Managers can make wiser, more ethical decisions if they become mindful of their unconscious biases. But how can we get at something outside our conscious awareness? By bringing the conscious mind to bear. Just as the driver of a misaligned car deliberately counteracts its pull, so can managers develop conscious strategies to counteract the pull of their unconscious biases. What’s required is vigilance—continual awareness of the forces that can cause decision making to veer from its intended course and continual adjustments to counteract them. Those adjustments fall into three general categories: collecting data, shaping the environment, and broadening the decision-making process.

Collect data

The first step to reducing unconscious bias is to collect data to reveal its presence. Often, the data will be counterintuitive. Consider many people’s surprise to learn of their own gender and racial biases on the IAT. Why the surprise? Because most of us trust the “statistics” our intuition provides. Better data are easily, but rarely, collected. One way to get those data is to examine our decisions in a systematic way.

Remember the MBA study groups whose participants overestimated their individual contributions to the group effort so that the totals averaged 139%? When the researchers asked group members to estimate what each of the other members’ contributions were before claiming their own, the total fell to 121%. The tendency to claim too much credit still persisted, but this strategy of “unpacking” the work reduced the magnitude of the bias. In environments characterized by “I deserve more than you’re giving me” claims, merely asking team members to unpack the contributions of others before claiming their own share of the pot usually aligns claims more closely with what’s actually deserved. As this example demonstrates, such systematic audits of both individual and group decision-making processes can occur even as the decisions are being made.

Unpacking is a simple strategy that managers should routinely use to evaluate the fairness of their own claims within the organization. But they can also apply it in any situation where team members or subordinates may be overclaiming. For example, in explaining a raise that an employee feels is inadequate, a manager should ask the subordinate not what he thinks he alone deserves but what he considers an appropriate raise after taking into account each coworker’s contribution and the pool available for pay increases. Similarly, when an individual feels she’s doing more than her fair share of a team’s work, asking her to consider other people’s efforts before estimating her own can help align her perception with reality, restore her commitment, and reduce a skewed sense of entitlement.

Taking the IAT is another valuable strategy for collecting data. We recommend that you and others in your organization use the test to expose your own implicit biases. But one word of warning: Because the test is an educational and research tool, not a selection or evaluation tool, it is critical that you consider your results and others’ to be private information. Simply knowing the magnitude and pervasiveness of your own biases can help direct your attention to areas of decision making that are in need of careful examination and reconsideration. For example, a manager whose testing reveals a bias toward certain groups ought to examine her hiring practices to see if she has indeed been disproportionately favoring those groups. But because so many people harbor such biases, they can also be generally acknowledged, and that knowledge can be used as the basis for changing the way decisions are made. It is important to guard against using pervasiveness to justify complacency and inaction: Pervasiveness of bias is not a mark of its appropriateness any more than poor eyesight is considered so ordinary a condition that it does not require corrective lenses.

Shape your environment

Research shows that implicit attitudes can be shaped by external cues in the environment. For example, Curtis Hardin and colleagues at UCLA used the IAT to study whether subjects’ implicit race bias would be affected if the test was administered by a black investigator. One group of students took the test under the guidance of a white experimenter; another group took the test with a black experimenter. The mere presence of a black experimenter, Hardin found, reduced the level of subjects’ implicit antiblack bias on the IAT. Numerous similar studies have shown similar effects with other social groups. What accounts for such shifts? We can speculate that experimenters in classrooms are assumed to be competent, in charge, and authoritative. Subjects guided by a black experimenter attribute these positive characteristics to that person, and then perhaps to the group as a whole. These findings suggest that one remedy for implicit bias is to expose oneself to images and social environments that challenge stereotypes.

We know of a judge whose court is located in a predominantly African-American neighborhood. Because of the crime and arrest patterns in the community, most people the judge sentences are black. The judge confronted a paradox. On the one hand, she took a judicial oath to be objective and egalitarian, and indeed she consciously believed that her decisions were unbiased. On the other hand, every day she was exposed to an environment that reinforced the association between black men and crime. Although she consciously rejected racial stereotypes, she suspected that she harbored unconscious prejudices merely from working in a segregated world. Immersed in this environment each day, she wondered if it was possible to give the defendants a fair hearing.

Rather than allow her environment to reinforce a bias, the judge created an alternative environment. She spent a vacation week sitting in a fellow judge’s court in a neighborhood where the criminals being tried were predominantly white. Case after case challenged the stereotype of blacks as criminal and whites as law abiding and so challenged any bias against blacks that she might have harbored.

Think about the possibly biased associations your workplace fosters. Is there, perhaps, a “wall of fame” with pictures of high achievers all cast from the same mold? Are certain types of managers invariably promoted? Do people overuse certain analogies drawn from stereotypical or narrow domains of knowledge (sports metaphors, for instance, or cooking terms)? Managers can audit their organization to uncover such patterns or cues that unwittingly lead to stereotypical associations.

If an audit reveals that the environment may be promoting unconscious biased or unethical behavior, consider creating countervailing experiences, as the judge did. For example, if your department reinforces the stereotype of men as naturally dominant in a hierarchy (most managers are male, and most assistants are female), find a department with women in leadership positions and set up a shadow program. Both groups will benefit from the exchange of best practices, and your group will be quietly exposed to counter-stereotypical cues. Managers sending people out to spend time in clients’ organizations as a way to improve service should take care to select organizations likely to counter stereotypes reinforced in your own company.

Broaden your decision making

Imagine that you are making a decision in a meeting about an important company policy that will benefit some groups of employees more than others. A policy might, for example, provide extra vacation time for all employees but eliminate the flex time that has allowed many new parents to balance work with their family responsibilities. Another policy might lower the mandatory retirement age, eliminating some older workers but creating advancement opportunities for younger ones. Now pretend that, as you make your decisions, you don’t know which group you belong to. That is, you don’t know whether you are senior or junior, married or single, gay or straight, a parent or childless, male or female, healthy or unhealthy. You will eventually find out, but not until after the decision has been made. In this hypothetical scenario, what decision would you make? Would you be willing to risk being in the group disadvantaged by your own decision? How would your decisions differ if you could make them wearing various identities not your own?

This thought experiment is a version of philosopher John Rawls’s concept of the “veil of ignorance,” which posits that only a person ignorant of his own identity is capable of a truly ethical decision. Few of us can assume the veil completely, which is precisely why hidden biases, even when identified, are so difficult to correct. Still, applying the veil of ignorance to your next important managerial decision may offer some insight into how strongly implicit biases influence you.

Just as managers can expose bias by collecting data before acting on intuition, they can take other preemptive steps. What list of names do you start with when considering whom to send to a training program, recommend for a new assignment, or nominate for a fast-track position? Most of us can quickly and with little concentration come up with such a list. But keep in mind that your intuition is prone to implicit prejudice (which will strongly favor dominant and well-liked groups), in-group favoritism (which will favor people in your own group), overclaiming (which will favor you), and conflict of interest (which will favor people whose interests affect your own). Instead of relying on a mental short list when making personnel decisions, start with a full list of names of employees who have relevant qualifications.

Using a broad list of names has several advantages. The most obvious is that talent may surface that might otherwise be overlooked. Less obvious but equally important, the very act of considering a counter-stereotypical choice at the conscious level can reduce implicit bias. In fact, merely thinking about hypothetical, counter-stereotypical scenarios—such as what it would be like to trust a complex presentation to a female colleague or to receive a promotion from an African-American boss—can prompt less-biased and more ethical decision making. Similarly, consciously considering counterintuitive options in the face of conflicts of interest, or when there’s an opportunity to overclaim, can promote more objective and ethical decisions.

The Vigilant Manager

If you answered “true” to the question at the start of this article, you felt with some confidence that you are an ethical decision maker. How would you answer it now? It’s clear that neither simple conviction nor sincere intention is enough to ensure that you are the ethical practitioner you imagine yourself to be. Managers who aspire to be ethical must challenge the assumption that they’re always unbiased and acknowledge that vigilance, even more than good intention, is a defining characteristic of an ethical manager. They must actively collect data, shape their environments, and broaden their decision making. What’s more, an obvious redress is available. Managers should seek every opportunity to implement affirmative action policies—not because of past wrongs done to one group or another but because of the everyday wrongs that we can now document are inherent in the ordinary, everyday behavior of good, well-intentioned people. Ironically, only those who understand their own potential for unethical behavior can become the ethical decision makers that they aspire to be.

Originally published in December 2003. Reprint 5526