Assessing Biases in Others versus Self

Assessing Biases in Others versus Self

In a questionnaire study, we (Pronin, Lin, & Ross, 2002) asked Stanford students the extent to which they and the “average American” displayed a variety of different inferential or judgmental biases that our conceptual analysis led us to believe would be subject to invidious self–other comparisons. The list included self-serving or ego-defensive attribution of success versus failure, dissonance reduction after free choice, the halo effect, biased assimilation of information to preexisting beliefs or preconceptions, reactive devaluation of offers and compromises received from the “other side,” unwarranted perception of media hostility to one’s group or cause, and an inclination to underestimate the role of situational as opposed to dispositional determinants of behavior. Each bias was described in a couple of simple jargon-free sentences of the following sort:

Psychologists have claimed that people show a ‘self-serving’ tendency in the way they view their academic or job performance. That is, they tend to take credit for success but deny responsibility for failure; they see their successes as the result of personal qualities, like drive or ability, but their failures as the result of external factors, like unreasonable work requirements or inadequate instruction.

As anticipated (Fig. 36.5), the students claimed that they personally displayed each of the relevant biases to a lesser degree than the “Average American,” with the average discrepancy constituting about one standard deviation on the relevant 9-point scale.

Figure 36.5. Perceptions of bias in self versus the "average American." Ratings were provided on 9-point scales (1 = no bias; 9 = strong bias). Pairs of bars are significantly different. FAE = Fundamental attribution error.

We further obtained survey evidence that the phenomenon in question was neither restricted to the perhaps atypically smug population of Stanford students nor a reflection of some highly general tendency for people to deny all personal shortcomings. This second survey used a sample of travelers awaiting flights at the San Francisco Airport. It included not only the items used in our Stanford survey, but also additional items pertaining to personal limitations or biases that would likely be at least as salient to the actor as the observer (e.g., fear of public speaking, procrastination, the planning fallacy) and therefore, according to our conceptual analysis, an unlikely source of invidious self-other comparisons. The results of this study both replicated our earlier results and supported our prediction about when the phenomenon should and should not occur. Thus, although our airport sample again made invidious self–other comparisons regarding the tendency to engage in ego-defensive attributions, commit the fundamental attribution error, act in a self-interested manner, and overestimate media hostility against the positions or causes they favor, they did not see themselves as less afraid of public speaking, less guilty of procrastination, or less susceptible to the planning fallacy.

The results of a third survey in this series further attest to the inability of people to perceive biases in themselves – even in the face of an explicit prompt about the prevalence of that bias. In this study, participants first rated themselves relative to their Stanford peers on a variety of dimensions (i.e., dependability, objectivity, consideration for others, snobbery, deceptiveness, and selfishness). As is generally the case in such studies, the great majority (87%) showed the familiar tendency to rate themselves more positively than their peers with respect to these characteristics (see Alicke et al., 1995).

After offering these ratings, however, participants were informed on the next page of the questionnaire that

Studies have shown that on the whole people show a “better than average” effect when assessing themselves relative to other members within their group. That is, 70–80% of individuals consistently rate themselves “better than average” on qualities that they perceive as positive and, conversely, evaluate themselves as having “less than average” amounts of characteristics that they believe are negative.

They were then asked to indicate what they thought the “most accurate, valid, and objective measures available” would show about them. Given this opportunity to acknowledge the likelihood that they had succumbed to the bias in question, only 24% of the participants who had shown the relevant self-enhancement bias did so. That means that 76% of the participants indicated that they expected the “most accurate and object measures available” to show either that they had rated themselves perfectly accurately, or that they had rated themselves too modestly!

Underestimating Bias in the Self. As Wilson, Hodges, and La Fleur (1995) provocatively noted, “People are often unaware of their own unawareness” (p. 17). Four biases that we mentioned earlier seem to hinge at least in part on such lack of awareness. The most general of these biases is the much researched and highly robust false consensus effect (Ross, Greene, & House, 1977). As noted earlier, this bias seems to depend on the individual’s failure to recognize that his or her estimates of response consensus (and the appropriate interpretation to place on the two response alternatives) have been heavily influenced by the particular way in which those alternatives are construed. The individual who shows the false consensus effect fails to recognize that other respondents might construe those alternatives differently, and as a result make a different choice; indeed, the individual fails to recognize that any construal is involved.

Other more specific biases can be seen as special cases or by-products of the false consensus effect. One example is the third-person effect (Davidson, 1983; see also Gunther, 1995; Innes & Zeitz, 1988), which refers to the tendency for people to believe that other respondents, on average, are more influenced than they by “propaganda” or other types of information that they feel ought to be dismissed out of hand. Presumably, individuals who see a message as persuasive and reasonable are apt to assume that other recipients of the message will feel likewise (and deem acceptance of the relevant message as normative rather than indicative of susceptibility to bias). At the same time, individuals who see the message as unpersuasive and hateful may be inclined to fear that others, especially others who have a vested interest in accepting such a message, will view it uncritically and succumb to its influence. The net effect of combining estimates that others will be “appropriately” persuaded whenever we are persuaded, and estimates that at least some others will be “inappropriately” persuaded, even when we are not, should be clear – that is, an overall tendency to overestimate the message’s impact. (See Hoorens & Ruiter, 1996, for evidence that people do not see themselves as less amenable to persuasion than their peers, just as more discerning in evaluating the value of the messages.)

Research by Armor (1998) provides more direct evidence about perceived and actual objectivity of self versus others. In one study, this investigator introduced participants to a number of problems used to demonstrate normative violations in judgment and inference (for example, “base-rate neglect” or violation of the “conjunction rule”). As predicted, participants both underestimated their susceptibility to the relevant inferential shortcomings and perceived themselves to be less susceptible than their peers. Once again, this tendency for people to overestimate their own objectivity is perhaps inevitable, because people would presumably change their assessment in the appropriate direction in any instance in which they recognized that the response of a peer (or any other particular response available to them) was more objective or normative than their own. To make accurate assessments of our objectivity, we must recognize and make allowance for the fact that (notwithstanding our impressions about individual assessments we are making) we are likely in many instances to be unobjective, and in error – in fact, to be in error just as often as our peers who seem to be making less reasonable assessments than we are every time their assessments differ from our own.

Overestimating Bias in Others. There is ample evidence that people recognize that their peers are subject to a variety of biases. Researchers have demonstrated people’s awareness of others’ susceptibility to the actor versus observer bias in the attribution process (Krueger, Ham, & Linford, 1996), the false consensus effect (Krueger & Zeiger, 1993), the correspondence bias (Van Boven, Kamada, & Gilovich, 1999), the influence of self-interest (Epley & Dunning, 2000; Kruger & Gilovich, 1999; Miller & Ratner, 1998), and the motivating power of extrinsic rewards (Heath, 1999).

Not only are people aware of the judgmental biases we have listed, there is evidence that people may actually overestimate their magnitude, especially when the potentially biased judgments are being made about them. Thus, Van Boven, Kamada, and Gilovich (1999) found that actors in an experiment assumed that observers would make even more extreme dispositional inferences about them than proved to be the case. As the authors note, people assume that those evaluating their responses will be “no-holds-barred” dispositionalists, and that they alone assign appropriate weight to person and situation in their attributions.

A completely different and more motivational set of naive theories seem involved in lay assumptions and overestimations when it comes to assessing the impact of ego-defensive or self-serving biases. In one such set of studies, Kruger and Gilovich (1999) found that participants were overly “cynical” in their predictions about how others would allocate responsibility for successes and failures. That is, they believed others would claim too much credit for success and too little blame for failure, when in fact individuals tended to allocate too much personal responsibility for both types of outcomes (see also Ross & Sicoly, 1979).

This tendency for people to see others as more self-serving than they are may occur in part because lay explanations for attributions that seem congruent with the actor’s self interest are overly motivational or at least overly simplistic. (See Nisbett & Ross, 1980 for an account of nonmotivational determinants of such bias; see also Dunning, Perie, & Story, 1991, and Kunda, 1987, for more sophisticated accounts of the sometimes subtle interaction between motivational and cognitive determinants.) In any case, introspection about our own assessments regarding our actions and attributes is apt to reveal no simple motive (or at least no simple conscious motive) to think well of ourselves at all costs, and the feedback most of us get from our friends is apt to suggest that we are just as inclined to be too hard on ourselves as too easy.

Evidence that people see themselves as less guilty of self-serving attributional biases than others comes from a study by Friedrich (1996), who simply described the self-serving bias and asked participants how subject they were to it relative to the average person. Participants rated themselves as significantly less prone than others to this bias. It is important to note, however, that people are not completely unaware that their own positive self-assessments reflect idiosyncratic cognitive processes. In a study by Krueger (1998), the positive correlation between people’s assessments of the desirability of a particular trait and the extent to which they ascribed those traits to themselves (see also Dunning, Perie, & Story, 1991) was accompanied by the participants’ recognition that their self-ascriptions would be less strongly correlated with others’ ratings of trait desirability than their own ratings. Krueger’s participants, we suspect, would have denied showing motivated self-enhancement, and have claimed that they simply try to behave in a manner consistent with the values they personally deem important. Indeed, by their own criteria, they may well have felt that other people were less enlightened than themselves about which traits are important (or even that other people’s ratings of trait importance were biased by their knowledge of their own personal strengths and shortcomings).

Not only do we view others’ attributions as more self-serving than our own, we have similar views about behavior and general motives for action. In a provocative set of experiments on the “myth of self-interest,” Miller and Ratner (1998) demonstrated that individuals judge others according to a naive theory that attaches great weight to naked self-interest. For example, they found that participants overestimated other people’s tendency to be influenced by economic incentives in deciding whether or not to donate blood and overestimated other people’s tendency to be influenced by their group membership (as a woman or a smoker) in determining their views about abortion or smoking restrictions. Furthermore, they expected others to behave in a manner consistent with self-interest regardless of whether they reported that their own behavior would be similarly guided. In a related line of research, Heath (1999) has shown that people assume others’ on-the-job motivation is rooted in extrinsic financial incentives, while they report their own motives to be more rooted in intrinsic incentives such as the opportunity to learn new skills.

In a final set of studies to be considered, Epley and Dunning (2000) showed that people judge others’ pro-social behavior in given situations (e.g., cooperating in a Prisoner’s Dilemma Game, donating money to a nonprofit organization) to be more self-serving and less purely altruistic than their own. Interestingly, and in contrast to the Miller and Ratner findings cited earlier, individuals showing this “holier than thou” effect were more accurate in their predictions regarding others than in predictions regarding themselves; that is, rather than underestimating others’ altruism, they overestimated their own. Both sets of findings, however, seem to reflect the difference in insider versus outsider perspectives that we have discussed at several points in this chapter. In the Miller and Ratner study, the actors’ insider perspective makes them aware that self-interest is not the sole or even the principal motive for their own choices, but this perspective offers no corresponding insight about the private feelings and motives of others. In the Epley and Dunning study, the actors apparently give too much weight to the “holy intentions” that underlie the resolution of their dilemmas, and in so doing these actors lack the objectivity of outside observers who consider other possibilities as well.

In a sense, the actor is like a detective with more clues than the observer, and with different clues enjoying high perceptual or cognitive salience. To the extent that the clues in question are in fact probative, the insider has the advantage over the outsider. To the extent that these clues are more salient than they are probative, the insider is at a disadvantage to any outsider who relies on simpler theories and more general base-rate assumptions. But in either case, the actor feels that his or her perspective is the one that affords greatest accuracy. And in either case, the actor is apt to feel frustrated or even angry with those who dispute the authenticity and special insight of his or her view of “reality.”