Covariation Model

Understanding Kelley's Covariation Theory

Kelley's Covariation theory assumes that we use the level of consensus, distinctiveness, and consistency about a person’s behavior to explain the behavior as mainly caused either by the person’s situation or by the person’s own characteristics or dispositions.

Covariation theory is an alternative theory of attribution (developed by Harold Kelley, 1967) explaining how people determine the causes of a person’s behavior by focusing on the factors present and absent when a behavior does not occur, and specifically on the role of consensus, distinctiveness, and consistency.

The covariation theory focuses on the factors that are present when a behavior occurs and the factors that are absent when it does not occur.

As shown in Exhibit I, covariation theory has three main components: consensus, distinctiveness, and consistency.

Exhibit I: Model of Covariation
According to covariation theory, we use the level of consensus, distinctiveness, and consistency about a person’s behavior to explain the behavior as mainly caused either by the person’s situation or by the person’s own characteristics or dispositions.
ConsensusDistinctivenessConsistencyAttribution
HighHighHighSituational
Other people all think your sister’s boyfriend is great.Your sister is very picky about her dating partners. It is unusual for her to like one this much, so quickly.Your sister continues to like this person over time.This boyfriend really is special.
LowLowHighDispositional
Other people think your sister’s boyfriend is horrible.Your sister quickly likes all her dating partners.Your sister continues to like this person over time.Your sister tends to fall in love quickly. This boyfriend is nothing special.
LowUncertain
Your sister quickly decides she doesn’t like this guy.You can’t tell if this boyfriend wasn’t right or if your sister just falls in and out of love quickly.

The first component of covariation theory is the consensus of the attitude or behavior, that is, whether other people generally agree or disagree with a given person. If many people agree with that person or behave in a similar manner, we are more likely to make a situational attribution than we would if few people agreed with the target individual.

  • In the case of your sister’s dating life, we will likely make a situational attribution about the characteristics of the particular dating partner she likes if other people also really like that person.
  • On the other hand, if most other people find that person rude and annoying, we are less likely to make a situational attribution about this person.
Important Hint!  

Consensus means information about whether other people generally behave in the same way toward the stimulus as the target person.

Second, we consider the distinctiveness of the person’s attitude or behavior, meaning whether the person’s attitude or behavior in this situation is relatively unique or whether the person generally reacts in a similar way across different situations.

  • Once again, your sister’s liking of a particular partner, while ridiculing others, would make her attraction to that partner quite distinctive. In turn, we are more likely to make a situational attribution in this case.
Important Hint!  

Distinctiveness refers to information about whether a person’s behavior is generally the same toward different stimuli.

Third, we consider the consistency of the person’s attitude or behavior, that is, whether the person’s attitude and/or behavior is similar over time.

  • If a person’s behavior is highly consistent over time and across situations (e.g., your sister likes a particular partner over time, even when they engage in different types of dating activities), we are likely to make a dispositional attribution.
  • On the other hand, if a given behavior is unusual for a particular person, we are likely to make a situational attribution (e.g., your sister feels very attracted to a partner after their first date, but not so much after third date).
Important Hint!  

Consistency is information about whether a person’s behavior towards a given stimulus is the same across time.

Let’s go through this process using another example. Imagine that you are trying to decide whether next semester you want to take a class in politics. You ask a friend, Joan, whether she would recommend the politics class she took last year. If Joan raves about this class, do you believe her and sign up?

If you are smart, your decision about whether to take the politics class Joan recommends will be influenced by the three main components of covariation theory.

  1. Consensus — Do her opinions have high consensus? Do many people like this class, or does only Joan like it? If everyone says it is a great class, then you can make a situational attribution (Joan liked the class because it was good), whereas if others say it is a really boring class, you should make a dispositional attribution for her attitude (Joan likes boring classes).
  2. Distinctiveness — Next, consider the distinctiveness of Joan’s attitude about this class. Does Joan rave about all of the classes she takes? If so, that doesn’t tell you much about this particular class because you should make a dispositional attribution for her attitude (it is just Joan, who likes all classes). But if Joan hates most of her classes, then her liking for this particular class should be attributed to the situation (the class).
  3. Consistency — Finally, consider whether Joan’s liking for this class is consistent over time. Maybe you asked Joan about the class on a day when she was in a particularly good mood, and later on she’ll report a different opinion. To make a strong dispositional attribution for her attitude toward the class, you need to ask Joan about the class on more than one occasion to make sure that her attitude is consistently positive.

In sum, according to the covariation model, we make different attributions Opens in new window depending on the consensus, distinctiveness, and consistency of a person’s attitude and/or behavior (Fiedlerr, Walther, & Nicke, 1999).

If consensus and distinctive are low and consistency is high, we make an internal or dispositional attribution (Joan just loves this class). In contrast, if consensus, distinctiveness, and consistency are all high, we make a situational attribution (this class is really great).

Finally, in any case in which a person’s attitude or behavior is low in consistency, we can’t make a dispositional or situational attribution.