Biases in Attribution

Are Our Attributions Accurate?

We have seen that person perception is useful in helping us successfully interact with others. In relation to our preceding discussion of attributions for success and failure, if we can determine why we did poorly on a test, we can try to prepare differently so we do better on the next one. Because successful navigation of the social world is based on being accurate, we can expect that our attributional skills will be pretty good. However, although people are often reasonably accurate in their attributions—we could say, perhaps, that they are “good enough” (Fiske, 2003)—they are far from perfect. In fact, causal attributions, including those relating to success and failure, are subject to the same types of biases that any other types of social judgments are. Let’s consider some of the ways that our attributions may go awry.

The Fundamental Attribution Error

One way that our attributions may be biased is that we are often too quick to attribute the behavior of other people to something personal about them rather than to something about their situation. This is a classic example of the tendency of underestimating how important the social situation really is in determining behavior. This bias occurs in two ways. First, we are too likely to make strong personal attributions to account for the behavior that we observe others engaging in. That is, we are more likely to say “Cejay left a big tip, so he must be generous” than “Cejay left a big tip, but perhaps that was because he was trying to impress his friends.” Second, we also tend to make more personal attributions about the behavior of others (we tend to say, “Cejay is a generous person”) than we do for ourselves (we tend to say, “I am generous in some situations but not in others”).

When we tend to overestimate the role of person factors and overlook the impact of situations, we are making a mistake that social psychologists have termed the fundamental attribution error. In one demonstration of the fundamental attribution error, Linda Skitka and her colleagues (Skitka, Mullen, Griffin, Hutchinson, & Chamberlin, 2002) had participants read a brief story about a professor who had selected two student volunteers to come up in front of a class to participate in a trivia game. The students were described as having been randomly assigned to the role of either quizmaster or contestant by drawing straws. The quizmaster was asked to generate five questions from his idiosyncratic knowledge, with the stipulation that he knew the correct answer to all five questions.

Joe (the quizmaster) subsequently posed his questions to the other student (Stan, the contestant). For example, Joe asked, “What cowboy movie actor’s sidekick is Smiley Burnette?” Stan looked puzzled and finally replied, “I really don’t know. The only movie cowboy that pops to mind for me is John Wayne.” Joe asked four additional questions, and Stan was described as answering only one of the five questions correctly. After reading the story, the students were asked to indicate their impression of both Stan’s and Joe’s intelligence.

If you think about the setup here, you’ll notice that the professor has created a situation that can have a big influence on the outcomes. Joe, the quizmaster, has a huge advantage because he got to choose the questions. As a result, the questions are hard for the contestant to answer. But did the participants realize that the situation was the cause of the outcomes? They did not. Rather, the students rated Joe as significantly more intelligent than Stan. You can imagine that Joe just seemed to be really smart to the students; after all, he knew all the answers, whereas Stan knew only one of the five. But of course this is a mistake. The difference was not at all due to person factors but completely to the situation: Joe got to use his own personal store of esoteric knowledge to create the most difficult questions he could think of. The observers committed the fundamental attribution error and did not sufficiently take the quizmaster’s situational advantage into account.

As we have explored in many places in this book, the culture that we live in has a significant impact on the way we think about and perceive our social worlds. Thus, it is not surprising that people in different cultures would tend to think about people at least somewhat differently. One difference is between people from  many Western cultures (e.g., the United States, Canada, Australia) and people from many Asian cultures (e.g., Japan, China, Taiwan, Korea, India). For instance, as we reviewed in Chapter 2 in our discussion of research about the self-concept, people from Western cultures tend to be primarily oriented toward individualism. This leads to them having an independent self-concept where they view themselves, and others, as autonomous beings who are somewhat separate from their social groups and environments. In contrast, people in many East Asian cultures take a more interdependent view of themselves and others, one that emphasizes not so much the individual but rather the relationship between individuals and the other people and things that surround them. In relation to our current discussion of attribution, an outcome of these differences is that, on average, people from individualistic cultures tend to focus their attributions more on the individual person, whereas, people from collectivistic cultures tend to focus more on the situation (Ji, Peng, & Nisbett, 2000; Lewis, Goto, & Kong, 2008; Maddux & Yuki, 2006).

In one study demonstrating this difference, Miller (1984) asked children and adults in both India (a collectivistic culture) and the United States (an individualist culture) to indicate the causes of negative actions by other people. Although the younger children (ages 8 and 11) did not differ, the older children (age 15) and the adults did—Americans made more personal attributions, whereas Indians made more situational attributions for the same behavior.

Masuda and Nisbett (2001) asked American and Japanese students to describe what they saw in images like the one shown in Figure 5.8, “Cultural Differences in Perception.” They found that while both groups talked about the most salient objects (the fish, which were brightly colored and swimming around), the Japanese students also tended to talk and remember more about the images in the background (they remembered the frog and the plants as well as the fish).

Michael Morris and his colleagues (Hong, Morris, Chiu, & Benet-Martínez, 2000) investigated the role of culture on person perception in a different way, by focusing on people who are bicultural (i.e., who have knowledge about two different cultures). In their research, they used high school students living in Hong Kong. Although traditional Chinese values are emphasized in Hong Kong, because Hong Kong was a British-administered territory for more than a century, the students there are also somewhat acculturated with Western social beliefs and values.

Morris and his colleagues first randomly assigned the students to one of three priming conditions. Participants in the American culture priming condition saw pictures of American icons (such as the U.S. Capitol building and the American flag) and then wrote 10 sentences about American culture. Participants in the Chinese culture priming condition saw eight Chinese icons (such as a Chinese dragon and the Great Wall of China) and then wrote 10 sentences about Chinese culture. Finally, participants in the control condition saw pictures of natural landscapes and wrote 10 sentences about the landscapes.

Then participants in all conditions read a story about an overweight boy who was advised by a physician not to eat food with high sugar content. One day, he and his friends went to a buffet dinner where a delicious-looking cake was offered. Despite its high sugar content, he ate it. After reading the story, the participants were asked to indicate the extent to which the boy’s weight problem was caused by his personality (personal attribution) or by the situation (situational attribution). The students who had been primed with symbols about American culture gave relatively less weight to situational (rather than personal) factors in comparison with students who had been primed with symbols of Chinese culture.

Returning to the case study at the start of this chapter, the very different explanations given in the English and Chinese language newspapers about the killings perpetrated by Gang Lu at the University of Iowa reflect these differing cultural tendencies toward internal versus external attributions. A focus on internal explanations led to an analysis of the crime primarily in terms of the individual characteristics of the perpetrator in the American newspaper, whereas there were more external attributions in the Chinese newspaper, focusing on the social conditions that led up to the tragedy. Morris and Peng (1994), in addition to their analyses of the news reports, extended their research by asking Chinese and American graduate students to weigh the importance of the potential causes outlined in the newspaper coverage. In line with predictions, the Chinese participants rated the social conditions as more important causes of the murders than the Americans, particularly stressing the role of corrupting influences and disruptive social changes. In contrast, the Americans rated internal characteristics of the perpetrator as more critical issues, particularly chronic psychological problems. Morris and Peng also found that, when asked to imagine factors that could have prevented the killings, the Chinese students focused more on the social conditions that could have been changed, whereas the Americans identified more changes in terms of the internal traits of the perpetrator.

Given these consistent differences in the weight put on internal versus external attributions, it should come as no surprise that people in collectivistic cultures tend to show the fundamental attribution error less often than those from individualistic cultures, particularly when the situational causes of behavior are made salient (Choi, Nisbett, & Norenzayan, 1999). Being more aware of these cross-cultural differences in attribution has been argued to be a critical issue facing us all on a global level, particularly in the future in a world where increased power and resource equality between Western and Eastern cultures seems likely (Nisbett, 2003). Human history is littered with tragic examples of the fatal consequences of cross-cultural misunderstandings, which can be fueled by a failure to understand these differing approaches to attribution. Maybe as the two worldviews increasingly interact on a world stage, a fusion of their two stances on attribution may become more possible, where sufficient weight is given to both the internal and external forces that drive human behavior (Nisbett, 2003).

The Actor-Observer Bias

The fundamental attribution error involves a bias in how easily and frequently we make personal versus situational attributions about others. Another, similar way that we overemphasize the power of the person is that we tend to make more personal attributions for the behavior of others than we do for ourselves and to make more situational attributions for our own behavior than for the behavior of others. This is known as the actor-observer bias (Nisbett, Caputo, Legant, & Marecek, 1973; Pronin, Lin, & Ross, 2002). When we are asked about the behavior of other people, we tend to quickly make trait attributions (“Oh, Sarah, she’s really shy”). On the other hand, when we think of ourselves, we are more likely to take the situation into account—we tend to say, “Well, I’m shy in my team at work, but with my close friends I’m not at all shy.” When a friend behaves in a helpful way, we naturally believe that they are a friendly person; when we behave in the same way, on the other hand, we realize that there may be a lot of other reasons why we did what we did.

You might be able to get a feel for the actor-observer difference by taking the following short quiz. First, think about a person you know, but not particularly well —a distant relation, a colleague at work. Then, for each row, circle which of the three choices best describes their personality (for instance, is the person’s personality more energetic, relaxed, or does it depend on the situation?). Then answer the questions again, but this time about yourself.

Richard Nisbett and his colleagues (Nisbett, Caputo, Legant, & Marecek, 1973) had college students complete a very similar task, which they did for themselves, for their best friend, for their father, and for a well-known TV newscaster at the time, Walter Cronkite. As you can see in Table 5.4, “The Actor-Observer Difference,” the participants checked one of the two trait terms more often for other people than they did for themselves, and checked off “depends on the situation” more frequently for themselves than they did for the other person; this is the actor-observer difference.

Table 5.4 The Actor-Observer Difference

Like the fundamental attribution error, the actor-observer difference reflects our tendency to overweigh the personal explanations of the behavior of other people. However, a recent meta-analysis (Malle, 2006) has suggested that the actor-observer difference might not be as common and strong as the fundamental attribution error and may only be likely to occur under certain conditions.

The tendency to overemphasize personal attributions in others versus ourselves seems to occur for several reasons. One is simply because other people are so salient in our social environments. When you look at someone’s behavior, you tend to focus on that person and are likely to make personal attributions about them. It’s just easy because you are looking right at the person. When you look at Cejay giving that big tip, you see him—and so you decide that he caused the action. In fact, research has shown that we tend to make more personal attributions for the people we are directly observing in our environments than for other people who are part of the situation but who we are not directly watching (Taylor & Fiske, 1975). When you think of your own behavior, however, you do not see yourself but are instead more focused on the situation. You also tend to have more memory for your own past situations than for others’. You come to realize that it is not only you but also the different situations that you are in that determine your behavior. Maybe you can remember the other times where you did not give a big tip, and so you conclude that your behavior is caused more by the situation than by your underlying personality.

This greater access to evidence about our own past behaviors can lead us to realize that our conduct varies quite a lot across situations, whereas because we have more limited memory of the behavior of others, we may see them as less changeable. This in turn leads to another, related attributional tendency, namely the trait ascription bias, which defines a tendency for people to view their own personality, beliefs, and behaviors as more variable than those of others (Kammer, 1982). We are thus more likely to caricature the behaviors of others as just reflecting the type of people we think they are, whereas we tend to depict our own conduct as more nuanced, and socially flexible.

A second reason for the tendency to make so many personal attributions is that they are simply easier to make than situational attributions. In fact, personal attributions seem to be made spontaneously, without any effort on our part, and even on the basis of only very limited behavior (Newman & Uleman, 1989; Uleman, Blader, & Todorov, 2005). Personal attributions just pop into mind before situational attributions do. One reason for this is that it is cognitively demanding to try to process all the relevant factors in someone else’s situation and to consider how all these forces may be affecting that person’s conduct. It is much more straightforward to label a behavior in terms of a personality trait.

Third, personal attributions also dominate because we need to make them in order to understand a situation. That is, we cannot make either a personal attribution (e.g., “Cejay is generous”) or a situational attribution (“Cejay is trying to impress his friends”) until we have first identified the behavior as being a generous behavior (“Leaving that big tip was a generous thing to do”). So we end up starting with the personal attribution (“generous”) and only later try to correct or adjust our judgment (“Oh,” we think, “perhaps it really was the situation that caused him to do that”).

Adjusting our judgments generally takes more effort than does making the original judgment, and the adjustment is frequently not sufficient. We are more likely to commit attributional errors—for example quickly jumping to the conclusion that behavior is caused by underlying personality—when we are tired, distracted, or busy doing other things (Geeraert, Yzerbyt, Corneille, & Wigboldus, 2004; Gilbert, 1989; Trope & Alfieri, 1997).

There is a very important general message about perceiving others that applies here: we should not be too quick to judge other people! It is cognitively easy to think that poor people are lazy, that people who harm someone else are mean, and that people who say something harsh are rude or unfriendly. But these attributions may frequently overemphasize the role of the person. This can sometimes result in overly harsh evaluations of people who don’t really deserve them; we tend to blame the victim, even for events that they can’t really control (Lerner, 1980). Sometimes people are lazy, mean, or rude, but they may also be the victims of situations. When you find yourself making strong personal attribution for the behaviors of others, your knowledge of attribution research can help you to stop and think more carefully: Would you want other people to make personal attributions for your behavior in the same situation, or would you prefer that they more fully consider the situation surrounding your behavior? Are you perhaps making the fundamental attribution error? Ultimately, to paraphrase a well-known saying, we need to be try to be generous to others in our attributions, as everyone we meet is fighting a battle we know nothing about.

Self-Serving Bias

You may recall that the process of making causal attributions is supposed to proceed in a careful, rational, and even scientific manner. But this assumption turns out to be, at least in part, untrue. Our attributions are sometimes biased by affect—particularly the desire to enhance the self that we talked about in Chapter 3. Although we would like to think that we are always rational and accurate in our attributions, we often tend to distort them to make us feel better. A particularly common example is the self-serving bias, which is the tendency to attribute our successes to ourselves, and our failures to others and the situation.

We all make self-enhancing attributions from time to time. If a teacher’s students do well on an exam, he may make a personal attribution for their successes (“I am, after all, a great teacher!”). On the other hand, when they do poorly on an exam, the teacher may tend to make a situational attribution and blame them for their failure (“Why didn’t you all study harder?”). You can see that this process is clearly not the type of scientific, rational, and careful process that attribution theory suggests the teacher should be following. It’s unfair, although it does make him feel better about himself. If he were really acting like a scientist, however, he would determine ahead of time what causes good or poor exam scores and make the appropriate attribution, regardless of the outcome.

You might have noticed yourself making self-serving attributions too. Perhaps you have blamed another driver for an accident that you were in or blamed your partner rather than yourself for a breakup. Or perhaps you have taken credit (internal) for your successes but blamed your failures on external causes. If these judgments were somewhat less than accurate, but they did benefit you, then they were indeed self-serving.

Interestingly, we do not as often show this bias when making attributions about the successes and setbacks of others. This tendency to make more charitable attributions about ourselves than others about positive and negative outcomes often links to the actor-observer difference that we mentioned earlier in this section. It appears that the tendency to make external attributions about our own behavior and internal attributions about the conduct of others is particularly strong in situations where the behavior involves undesirable outcomes. This was dramatically illustrated in some fascinating research by Baumeister, Stillwell, and Wotman (1990). In this study, the researchers analyzed the accounts people gave of an experience they identified where they angered someone else (i.e., when they were the perpetrator of a behavior leading to an unpleasant outcome) and another one where someone else angered them (i.e., they were the victim).

The differences in attributions made in these two situations were considerable. When accounting for themselves as perpetrators, people tended to emphasize situational factors to describe their behavior as an isolated incident that was a meaningful, understandable response to the situation, and to assert that the action caused no lasting harm. When they were the victims, on the other hand, they explained the perpetrator’s behavior by focusing on the presumed character defects of the person and by describing  the behavior as an arbitrary and senseless action, taking place in an ongoing context of abusive behavior that caused lasting harm to them as victims. These sobering findings have some profound implications for many important social issues, including reconciliation between individuals and groups who have been in conflict. In a more everyday way, they perhaps remind us of the need to try to extend the same understanding we give to ourselves in making sense of our behaviors to the people around us in our communities. Too many times in human history we have failed to understand and even demonized other people because of these types of attributional biases.

Why is engaging in the self-serving bias so common? One answer, that we have already alluded to, is that they can help to maintain and enhance self-esteem. Consistent with this idea is that there are some cross-cultural differences, reflecting the different amounts of self-enhancement that were discussed in Chapter 3. Specifically, self-serving bias is less apparent in members of collectivistic than individualistic cultures (Mezulis, Abramson, Hyde, & Hankin, 2004).

Another important reason is that when we make attributions, we are not only interested in causality, we are often interested in responsibility. Fincham and Jaspers (1980) argued that, as well as acting like lay scientists, hunting for the causes of behavior, we are also often akin to lay lawyers, seeking to assign responsibility. We want to know not just why something happened, but also who is to blame. Indeed, it is hard to make an attribution of cause without also making a claim about responsibility. When we attribute someone’s angry outburst to an internal factor, like an aggressive personality, as opposed to an external cause, such as a stressful situation, we are, implicitly or otherwise, also placing more blame on that person in the former case than in the latter. Seeing attribution as also being about responsibility sheds some interesting further light on the self-serving bias. Perhaps we make external attributions for failure partly because it is easier to blame others or the situation than it is ourselves. In the victim-perpetrator accounts outlined by Baumeister, Stillwell, and Wotman (1990), maybe they were partly about either absolving or assigning responsibility, respectively. Indeed, there are a number of other attributional biases that are also relevant to considerations of responsibility. It is to these that we will now turn.

Group-Serving Biases

A self-serving pattern of attribution can also spill over into our attributions about the groups that we belong to. The group-serving bias, sometimes referred to as the ultimate attribution error, describes a tendency to make internal attributions about our ingroups’ successes, and external attributions about their setbacks, and to make the opposite pattern of attributions about our outgroups (Taylor & Doria, 1981). When members of our favorite sports team make illegal challenges on the field, or rink, or court, we often attribute it to their being provoked. What about when it is someone from the opposition? Their illegal conduct regularly leads us to make an internal attribution about their moral character! On a more serious note, when individuals are in a violent confrontation, the same actions on both sides are typically attributed to different causes, depending on who is making the attribution, so that reaching a common understanding can become impossible (Pinker, 2011).

Returning to the case study at the start of this chapter, could the group-serving bias be at least part of the reason for the different attributions made by the Chinese and American participants about the mass killing? How might this bias have played out in this situation? Remember that the perpetrator, Gang Lu, was Chinese. Might the American participants’ tendency to make internal attributions have reflected their desire to blame him solely, as an outgroup member, whereas the Chinese participants’ more external attributions might have related to their wish to try to mitigate some of what their fellow ingroup member had done, by invoking the social conditions that preceded the crime?

Morris and Peng (1994) sought to test out this possibility by exploring cross-cultural reactions to another, parallel tragedy, that occurred just two weeks after Gang Lu’s crimes. Thomas Mcllvane, an Irish American postal worker who had recently lost his job, unsuccessfully appealed the decision with his union. He had in the meantime failed to find a new full-time job. On November 14, he entered the Royal Oak, Michigan, post office and shot his supervisor, the person who handled his appeal, several fellow workers and bystanders, and then himself. In all, like Gang Lu, Thomas McIllvane killed himself and five other people that day. If the group-serving bias could explain much of the cross-cultural differences in attributions, then, in this case, when the perpetrator was American, the Chinese should have been more likely to make internal, blaming attributions against an outgroup member, and the Americans to make more external, mitigating ones about their ingroup member. This is not what was found. Although the Americans did make more situational attributions about McIlvane than they did about Lu, the Chinese participants were equally likely to use situational explanations for both sets of killings. As Morris and Peng (1994) point out, this finding indicated that whereas the American participants tended to show the group-serving bias, the Chinese participants did not. This has been replicated in other studies indicating a lower likelihood of this bias in people from collectivistic versus individualistic cultures (Heine & Lehman, 1997).

At first glance, this might seem like a counterintuitive finding. If people from collectivist cultures tend to see themselves and others as more embedded in their ingroups, then wouldn’t they be more likely to make group-serving attributions? A key explanation as to why they are less likely relates back to the discussion in Chapter 3 of cultural differences in self-enhancement. Like the self-serving bias, group-serving attributions can have a self-enhancing function, leading people to feel better about themselves by generating favorable explanations about their ingroups’ behaviors. Therefore, as self-enhancement is less of a priority for people in collectivistic cultures, we would indeed expect them to show less group-serving bias.

Victim-Blaming Biases

We saw earlier how the fundamental attribution error, by causing us to place too much weight on the person and not enough on the situation, can lead to us to make attributions of blame toward others, even victims, for their behaviors. Another bias that increases the likelihood of victim-blaming is termed the belief in a just world, which is a tendency to make attributions based on the belief that the world is fundamentally just. In other words, that the outcomes people experience are fair.

Lerner (1965), in a classic experimental study of these beliefs, instructed participants to watch two people working together on an anagrams task. They were informed that one of the workers was selected by chance to be paid a large amount of money, whereas the other was to get nothing. Participants also learned that both workers, though ignorant of their fate, had agreed to do their best. In addition, the attractiveness of the two workers was set up so that participants would perceive one as more attractive. Consistent with the idea of having a belief in a just world, once the outcome was known to the observers, they persuaded themselves that the person who had been awarded the money by chance had really earned it after all. Also, when the less attractive worker was selected for payment, the performance of the entire group was devalued.

As with many of the attributional biases that have been identified, there are some positive aspects to these beliefs when they are applied to ourselves. Fox, Elder, Gater, & Johnson (2010), for instance, found that stronger endorsement of just world beliefs in relation to the self was related to higher self-esteem. Intuitively this makes sense: if we believe that the world is fair, and will give us back what we put in, this can be uplifting. On the other hand, though, as in the Lerner (1965) study above, there can be a downside, too. If we believe that the world is fair, this can also lead to a belief that good things happen to good people and bad things happen to bad people. In other words, people get what they deserve. When people are in difficult positions, believing in a just world can cause others to make internal attributions about the causes of these difficulties and to end up blaming them for their problems (Rubin & Peplau, 1973). Consistent with this, Fox and colleagues  found that greater agreement with just world beliefs about others was linked to harsher social attitudes and greater victim derogation.

People’s belief in a just world is often at work when they react to news of a particular crime by blaming the victim, or when they apportion responsibility to members of marginalized groups (e.g. those who are homeless) for the predicaments they face. Degree of endorsement of just world attributions also relates to more stigmatizing attitudes toward people who have mental illnesses (Rüsch, Todd, Bodenhausen, & Corrigan, 2010). These views, in turn, can act as a barrier to empathy and to an understanding of the social conditions that can create these challenges. Belief in a just world has also been shown to correlate with meritocratic attitudes, which assert that people achieve their social positions on the basis of merit alone. For example, people who endorse just world statements are also more likely to rate high-status individuals as more competent than low-status individuals. Such beliefs are in turn used by some individuals to justify and sustain inequality and oppression (Oldmeadow & Fiske, 2007). Here, then, we see important links between attributional biases held by individuals and the wider social inequities in their communities that these biases help to sustain.

Attributions that blame victims don’t only have the potential to help to reinforce people’s general sense that the world is a fair place, they also help them to feel more safe from being victimized themselves. If, according to the logic of the belief in a just world, victims are bad people who get what they deserve, then those who see themselves as good people do not have to confront the threatening possibility that they, too, could be the victims of similar misfortunes. Accordingly, defensive attribution (e.g., Shaver, 1970) occurs when we make attributions which defend ourselves from the notion that we could be the victim of an unfortunate outcome, and often also that we could be held responsible as the victim. Put another way, people’s attributions about the victims are motivated by both harm avoidance (this is unlikely to happen to me) and blame avoidance (if it did happen to me, I would not be to blame). If we see ourselves as more similar to the victim, therefore, we are less likely to attribute the blame to them. If, on the other hand, we identify more with the perpetrator, then our attributions of responsibility to the victim will increase (Burger, 1981).

This pattern of attribution clearly has significant repercussions in legal contexts. For example, attributions about the victims of  rape are related to the amount that people identify with the victim versus the perpetrator, which could have some important implications for jury selection procedures (Grubb & Harrower, 2009). Furthermore, men are less likely to make defensive attributions about the victims of sexual harassment than women, regardless of the gender of the victim and perpetrator (e.g., Smirles, 2004). Defensive attributions can also shape industrial disputes, such as damages claims for work-related injuries. The victims of serious occupational accidents tend to attribute the accidents to external factors. In contrast, their coworkers and supervisors are more likely to attribute the accidents to internal factors in the victim (Salminen, 1992). Again, the role of responsibility attributions are clear here. It is in the victims’ interests to not be held accountable, just as it may well be for the colleagues or managers who might instead be in the firing line.