Inferring Dispositions Using Causal Attribution

We have seen that we use personality traits to help us understand and communicate about the people we know. But how do we know what traits people have? People don’t walk around with labels saying “I am generous” or “I am aggressive” on their foreheads. In some cases, thinking back to our discussions of reputation in Chapter 3, we may learn about a person indirectly, for instance, through the comments that other people make about that person. We also use the techniques of person perception to help us learn about people and their traits by observing them and interpreting their behaviors. If Zoe hits Joe, we might conclude that Zoe is aggressive. If Cejay leaves a big tip for the waitress, we might conclude that he is generous. It seems natural and reasonable to make such inferences because we can assume (often, but not always, correctly) that behavior is caused by personality. It is Zoe’s aggressiveness that causes her to hit, and it is Cejay’s generosity that led to his big tip.

Although we can sometimes infer personality by observing behavior, this is not always the case. Remember that behavior is influenced by both our personal characteristics and the social context in which we find ourselves. What this means is that the behavior we observe other people engaging in might not always be reflective of their personality; instead, the behavior might have been caused more by the situation rather than by underlying person characteristics. Perhaps Zoe hit Joe not because she is really an aggressive person but because Joe insulted or provoked her first. And perhaps Cejay left a big tip in order to impress his friends rather than because he is truly generous.

Because behavior can be influenced by both the person and the situation, we must attempt to determine which of these two causes actually more strongly determined the behavior. The process of trying to determine the causes of people’s behavior is known as causal attribution (Heider, 1958). Because we cannot see personality, we must work to infer it. When a couple we know breaks up, despite what seemed to be a match made in heaven, we are naturally curious. What could have caused the breakup? Was it something one of them said or did? Or perhaps stress from financial hardship was the culprit?

Making a causal attribution can be a bit like conducting a social psychology experiment. We carefully observe the people we are interested in, and we note how they behave in different social situations. After we have made our observations, we draw our conclusions. We make a dispositional (or internal) attribution when we decide that the behavior was caused primarily by the person. A dispositional attribution might be something like “I think they broke up because Sarah was not committed to the relationship.” At other times, we may determine that the behavior was caused primarily by the situation—we call this making a situational (or external) attribution. A situational attribution might be something like, “I think they broke up because they were under such financial stress.” At yet other times, we may decide that the behavior was caused by both the person and the situation; “I think they broke up because Sarah’s lack of commitment really became an issue once they had financial troubles.”

Making Inferences about Personality

It is easier to make dispositional attributions in some cases than in others. When a behavior is unusual or unexpected in the particular situation it occurs in, we can more easily make a dispositional attribution for it. Imagine that you go to a party and you are introduced to Selin. Selin shakes your hand and says, “Nice to meet you!” Can you readily conclude, on the basis of this behavior, that Selin is a friendly person? Probably not. Because the social context demands that people act in a friendly way (by shaking your hand and saying “Nice to meet you”), it is difficult to know whether Selin acted friendly because of the situation or because she is really friendly. Imagine, however, that instead of shaking your hand, Selin ignores you and walks away. In such cases, it is easier in this case to infer that Selin is unfriendly because her behavior is so contrary to what one would expect.

To test this idea, Edward Jones and his colleagues (Jones, Davis, & Gergen, 1961) conducted a classic experiment in which participants viewed one of four different videotapes of a man who was applying for a job. For half the participants, the video indicated that the man was interviewing for a job as a submariner, a position that required close contact with many people over a long period of time. It was clear to the man being interviewed, as well as to the research participants, that to be a good submariner you should be extroverted (i.e., you should enjoy being around others). The other half of the participants saw a video in which the man was interviewing for a job as an astronaut, which involved (remember, this study was conducted in 1961) being in a small capsule, alone, for days on end. In this case, it was clear to everyone that in order to be good astronaut, you should have an introverted personality.

During the videotape of the interview, a second variable was also manipulated. One half of the participants saw the man indicate that he was actually an introvert (he said things such as “I like to work on my own,” “I don’t go out much”), and the other half saw the man say that he was actually an extrovert (he said things such as “I would like to be a salesman,” “I always get ideas from others”). After viewing one of the four videotapes, participants were asked to indicate how introverted or extroverted they thought the applicant really was.

As you can see in Table 5.2, “Attributions to Expected and Unexpected Behaviors,” when the applicant gave responses that better matched what was required by the job (i.e., for the submariner job, the applicant said he was an extrovert, and for the astronaut job, he said he was an introvert), the participants did not think his statements were as indicative of his underlying personality as they did when the applicant said the opposite of what was expected by the job (i.e., when the job required that he be extroverted but he said he was introverted, or vice versa).

Table 5.2 Attributions to Expected and Unexpected Behaviors

The idea here is that the statements that were unusual or unexpected (on the basis of the job requirements) just seemed like they could not possibly have been caused by the situation, so the participants really thought that the interviewee was telling the truth. On the other hand, when the interviewee made statements that were consistent with what was required by the situation, it was more difficult to be sure that he was telling the truth (perhaps, thinking back to the discussion of strategic self-presentation in Chapter 3, he was just saying these things because he wanted to get the job), and the participants made weaker dispositional attributions for his behavior.

We can also make dispositional attributions more easily when we know that the person had a choice in the behavior. If a man chooses to be friendly, even in situations in which he might not be, this probably means that he is friendly. But if we can determine that he’s been forced to be friendly, it’s more difficult to know. If, for example, you saw a man pointing a gun at another person, and then you saw that person give his watch and wallet to the gunman, you would probably not infer that the person was generous!

Jones and Harris (1967) had student participants in a study read essays that had been written by other students. Half of the participants thought the students had chosen the essay topics, whereas the other half thought the students had been assigned the topics by their professor. The participants were more likely to make a dispositional attribution that the students really believed in the essay they were writing when they had chosen the topics rather than been assigned topics.

Sometimes a person may try to lead others to make dispositional attributions for their behavior to make themselves seem more believable. For example, when a politician makes statements supporting a cause in front of an audience that does not agree with her position, she will be seen as more committed to her beliefs and may be more persuasive than if she gave the same argument in front of an audience known to support her views. Again, the idea is based on principles of attribution: if there is an obvious situational reason for making a statement (the audience supports the politician’s views), then the dispositional attribution (that the politician really believes what she is saying) is harder to make.

Detecting the Covariation between Personality and Behavior

So far, we have considered how we make dispositional attributions when we have only limited information; that is, behavior observed at only a single point in time—a man leaving a big tip at a restaurant, a man answering questions at a job interview, or a politician giving a speech. But the process of making attributions also occurs when we are able to observe a person’s behavior in more than one situation. Certainly, we can learn more about Cejay’s generosity if he gives a big tip in many different restaurants with many different people, and we can learn more about a politician’s beliefs by observing the kinds of speeches she gives to different audiences over time.

When people have multiple sources of information about the behavior of a person, they can make attributions by assessing the relationship between a person’s behavior and the social context in which it occurs. One way of doing so is to use the covariation model, which states that a given behavior is more likely to have been caused by the situation if that behavior covaries (or changes) across situations. Our job, then, is to study the patterns of a person’s behavior across different situations in order to help us to draw inferences about the causes of that behavior (Jones et al., 1987; Kelley, 1967).

Research has found that people focus on three kinds of covariation information when they are observing the behavior of others (Cheng & Novick, 1990).

• Consistency information. Information regarding whether an actor always engages in a behaviour. For instance, if I always get my wife sick with my cooking, then this suggests high consistency.
• Distinctiveness information. Information regarding whether the target is the only one affected versus everyone else. For instance, if I only get my wife sick with my cooking – and no one else gets sick from my cooking, then this suggests high distinctiveness.
• Consensus information. Information regarding whether most people create a certain outcome. For instance, if many people get my wife sick with their cooking, then this suggests .

Imagine your friend Ravi – he really struggled with a recent creating writing assignment. Despite working really hard on the paper, he got a really low grade. So you sit there and wonder – why did the prof give Ravi such a bad grade? When focusing on the actor (i.e. the prof), was it something internal about them, or was it something external from them?

According to the covariation model, you should be able to get a cause by considering the three types of covariation information: consistency, distinctiveness, and consensus. One question you might ask is whether the prof always gives Ravi bad grades. If the answer is yes, then you have some consistency information. If you have noticed that Ravi is the only person who got a bad grade on that assignment, then you also have distinctiveness information. Finally, you might look for consensus information: do other profs also tend to give bad grades to Ravi?

Consider one more example. Imagine you’re watching a funny Japanese TV show involving Matsumoto and Hamada. During the show, Hamada smacks Matsumoto in the head. You wonder whether you should attribute why Hamada hit him internally to Hamada, externally to Matsumoto. To do this, you’d probably consider consistency information, distinctivenes information, and consistency information. As we will see later, different combinations of these pieces of information yield different types of attributions. You can also see still another example of the use of covariation information in Table 5.3, “Using Covariation Information.”

Table 5.3 Using Covariation Information

In summary, covariation models predict that we will most likely make external attributions about the target receiving the action when consensus, distinctiveness, and consistency are all high. In contrast, when consensus and distinctiveness are both low and this is accompanied by high consistency, then we are most likely to arrive at an internal attribution to the actor (Kelley, 1967). Moreover, when consistency is low in general, it may be attributed externally to the situation.

These predictions have generally been supported in studies of attribution, typically asking people to make attributions about a stranger’s behaviors in vignettes (Kassin, 1979). In studies in more naturalistic contexts, for example those we make about ourselves and others who we know well, many other factors will also affect the types of attributions that we make. These include our relationship to the person and our prior beliefs. For instance, our attributions toward our friends are often more favorable than those we make toward strangers (Campbell, Sedikides, Reeder, & Elliot, 2000). Also, in line with our discussions of schemas and social cogniton in Chapter 2, they are often consistent with the content of the schemas that are salient to us at the time (Lyon, Startup, & Bentall, 1999).

We have reviewed some of the important theory and research into how we make attributions. Another important question, that we will now turn to, is how accurately we attribute the causes of behavior. It is one thing to believe that that someone shouted at us because they have an aggressive personality, but quite another to prove that the situation, including our own behavior, was not the more important cause!