BEHAVIOUR AS SEEN BY THE ACTOR AND AS SEEN BY THE OBSERVER

RICHARD E. NISBETT, CRAIG CAPUTO, PATRICIA LEGANT AND JEANNE MARACEK (1973)

Actors are inclined to attribute their behaviour to situational causes. Observers are inclined to attribute the actor's behaviour to dispositional causes. College students (i) believed that the actor would behave in a similar way in the future. (ii) College students believed that their best friends were with their present girlfriend, and chose their college course, because of the best friend's disposition. The students explained their own choice of girlfriend, and choice of college course, with attributes or properties. (iii) Ascribe more personality traits to other people than to themselves.

STUDY 1

Observers watched the experimenter asking an actor to comply. It was predicted that the observer would expect the actor to comply in other situations.

METHOD

OVERVIEW

Besides the two subjects, present at each testing session, two confederates were present. They played the same role as the actors. Actors were asked to give up some of their time to work as weekend hostesses. They were to look after the wives of potential financial backers. Finance was needed. It was to be used for research. The research was into the learning disabilities of disadvantaged children. This work was to be carried out at the university. In one condition, $0.50 per hour was offered as payment. In the other condition, $1.50 was offered. After the real subject, playing the part of the actor had either accepted or refused, both the actor and observer were quizzed about their perceptions of the actor's behaviour.

INSTRUCTIONS TO THE OBSERVER

This part of the experiment was conducted by one (E1) of two experimenters. This is worth reading for yourself. Notice the observer was asked to watch just the real subject. After the decision making process, the two subjects were to meet face to face. Notice the observer is introduced as 'the experimenter's helper'.

SCENARIO

A different experimenter (E2) was used in the scene that was being observed. Again, refer to p162 in Gross and read the story about the banker's wives. Notice the purpose of the confederates was to lend some authenticity to the scene, by asking a couple of questions. The real actor seldom asked anything. Notice also that a conformity effect might have been the reason so many volunteered. Some variations were tried, in an attempt to reduce the effect of social pressure. Subjects still tended to volunteer, regardless of the level of social pressure. Because of this, the different conditions were analysed as one.

ASSESSMENT OF PERCEIVED REASONS FOR VOLUNTEERING

This assessment was conducted by the first experimenter (E1), in a separate room. The actor was asked to rate six questions on a scale of 0-8. These questions were designed to find out the reasons for volunteering (assuming she did volunteer). The observer was given the same questions. She was asked about the actors' motives for volunteering. If the subject actor had not volunteered, then the procedure was different. The observer was asked about the motives of one of the confederates.

ASSESSMENT OF PERCEIVED DISPOSITION TO VOLUNTEER

Both actor and observer were asked how likely the actor would be to volunteer for canvassing for the United Fund. A scale of 0-8 was used.

SUBJECTS

Thirty-three actor-observer pairs were used. Female students from Yale university were either paid $1.50 each or given credits towards a psychology course. Five of the observers did not record their disposition-to-volunteer data, owing to an error.

RESULTS

More subjects volunteered when offered $1.50 per hour, compared to $0.50. Even though money seems to be a strong incentive to volunteer, both actor and observer did not rate money as having that much effect. [See question 3, p163].

If the actor volunteered, the observer was more likely to believe that she would volunteer again (t=2.24, p<0.05). All p values are based on two-tailed tests). Observers thought the actor is more likely to help. This is in comparison to what the actor, herself, thought (t=2.12, p<0.05). Experimenters admit that their experiment had low ecological validity.

STUDY II

This is the 'why did your friend chose that particular girlfriend and college course?' experiment.

METHOD

Yale male undergraduates were paid $1.50 each. Twenty-three provided data that could be used. They filled in personality trait questionnaires, which were to be used in study III. They had to write four brief paragraphs on why they had liked the girl they had dated most regularly, in the past year or so; and why he had chosen his college subject (major). He then had to pretend that he was his best friend, and then repeat the process. Finally, he had to pretend he was his best friend, pretending to be him, and then repeat the process, once again. The best friend had to be of the same sex and age. If there was more than one, then he was to choose the one he had known the longest. Scoring was for sentences that could be deemed as giving an 'entity' reason, or a 'dispositional' reason. 'She's a relaxing person', would be an 'entity' reason. 'I need someone I can relax with', would be a 'dispositional' reason. Coding was conducted by experimenters. As a check, the coding was checked by a 'blind' coder, after the sentences had been re-written as third person statements. There was high agreement between the experimenters and the 'blind' coders.

RESULTS

Subjects gave more than twice as many 'entity' reasons compared to 'dispositional', when giving reasons for liking their girlfriend (t=2.54, p<0.02). The interaction between self Vs friend and 'entity' Vs 'dispositional' is significant (t=2.23, p=0.05). Subjects gave four times as many 'dispositional' reasons for their friend's choice of college course, compared to 'entity' reasons (t=3.53, p<0.002). Subjects, when pretending to be their best friend commenting upon themselves, duplicated the results for when they were commenting on their best friend's choices.

It could be argued that the results are an artefact of literary conventions. A follow-up study used sentences culled from the main study. The fresh subjects were asked to rate the sentences that best explained the choices. Half were phrased as 'entity' statements, and half were phrased as 'dispositional' statements. Ratings for their friend were higher for 'dispositional wording' than for 'entity wording'. The opposite was found for their self-ratings. This interaction was significant (t=2.11, p=0.05). These results suggest that artefacts of language production can be ruled out.

STUDY III

Could this tendency to attribute dispositional causes, to the behaviour of others, be owing to a belief that others possess more personality traits than oneself?

METHOD

PROCEDURE

Twenty-four subjects filled out questionnaires for themselves and four other stimulus persons (see below). There were 20 three choice items on each questionnaire. The three choices were: the trait, its polar opposite, and 'depends on the situation'. The traits were subsequently rated for desirability on a 7-point scale (-3 to +3).

TRAIT TERMS

See Table 10.4, p170. The trait pairs were confounded by desirability. Many trait pairs had one trait that was significantly higher in desirability than the other. Two traits (tough-minded and deferential) had mean negative ratings. Trait pairs that significantly differed in desirability, were analysed separately.

STIMULUS PERSONS

The stimulus persons were self, best-friend, father, an admired acquaintance (own age and sex, but known for less than three months) and a television commentator (Walter Cronkite). It is noticed that the persons differ for age (compared with the age of the subject) and familiarity.

RESULTS

There were significant differences between the mean trait ascription's of the following: self and all others, father and Cronkite, acquaintance and Cronkite. Self had the lowest number of traits ascribed, and Cronkite had the most (see table 10.5, p171). The 'depends on the situation' choice was made more often when rating self than for any other person. This category was not used by the subject, purely to avoid ascribing undesirable traits to himself. More neutral traits were ascribed to others than to the subject. Also, more undesirable traits were ascribed to the subject than to others. Nonetheless, the mean desirability rating of words ascribed to the self were significantly higher than for those words ascribed to all others except best-friend. It could be argued that to choose either one trait or its opposite is unfair. The subject might view himself as being in-between the two extremes. A follow-up study allowed a 6 point trait continuum, between each pair, plus 'depends on the situation'. Still, subjects preferred to use the 'depends on the situation' category more often for themselves, compared to their best friend. The follow-up study also demonstrated that there was a correlation of -0.45 (p<0.01) between length of friendship and tendency to ascribe more traits to best friend compared to oneself. This means that the longer the subject had known his best friend the more he responded to him in the same way as he did to himself.

DISCUSSION

The authors admit that the evidence, presented above, does not mean that Jones and Nisbett's (1971) hypothesis will generally hold true. Research should be directed at the mechanisms on which the proposition rests. Storms (1973) found that if people see themselves on video, they then tend to attribute their actions to dispositional causes! Brehm (1966) has written about the 'reactance' motive. We like to see ourselves as free. If we admitted that traits have a lot to do with the way we behave, then we would go against the 'reactance' motive (i.e. deny that we have freedom). Our own sense of freedom would be enhanced if we were to consider others as trait bound.

COMMENTARY

AIM AND NATURE

The experimenters were being cautious by not claiming that their results generalise. Notice they also used two-tailed tests. This is a general hypothesis, and it would be silly to accept it as such. More evidence will accumulate to support certain parts of the theory. It is wiser to accept or reject more specific hypotheses, for this reason. Do not use the term 'prove' and (to a lesser extent) 'disprove'. These terms suggest certainty. We should only concern ourselves with the probability of certain theories being correct; that is, having the power to predict that certain types of behaviour are more likely to occur than others.

CONTEXT AND BACKGROUND

This paper concerns Attribution Theory. The theories and experiments from many different psychologists have contributed to the development of the theory. Attribution Theory concerns the way we attempt to explain the actions of people. It looks at people as if we are all scientists making hypotheses about what people are like, and what they will do [Heider's (1958) common-sense psychology]. It was Heider whom first identified the dispositional and situational [entity] factors, mentioned in the present paper. Kelley's Co-variation model (1967), concerns whether it is the disposition of the actor or the environment that helps to cause a particular behaviour. The behaviour of the actor, viewed over different situations and with different groups of people, helps the observer to form an opinion, one way or another. The three types of causal information are known as consensus, distinctiveness and consistency. In other words, 'are others acting the same way?', 'is he like this in other similar situations?' and 'does this happen often over a time?'. The following is taken from the psychology program on the college computer network. I hope you are using it!

Attribution... Refers to the idea that we tend to search for the causes of peoples' behaviour - we try to attribute it to something. This can be either the person's disposition (e.g. mood, _personality_), or because of the situation they're in. We make a personal attribution about a person's behaviour when: other people don't behave in the same way (low consensus), the person behaves the same in similar _situations_ (low distinctiveness), and when their behaviour is stable over time (high consistency). When a lecturer is late for a class, then students will think this is probably due to a personal attribute (laziness), if: other lecturers turn up on time for their classes, the lecturer is late for other classes, and they are always late.

situations_ The immediate circumstances we find ourselves in are powerful determinants of our behaviour. Although this may seem obvious, we usually underestimate it. Our _roles_ and our _personality_ may modify our behaviour and make us act differently from other people. It is however simply the situation that often has the greatest effect. For example, we may be a politician and an extrovert but in a funeral we would usually sit quietly. When we judge the behaviour of others, we tend to underestimate the importance of the situation - we think it is their own fault or responsibility. When we judge our own behaviour however, we tend to think of it as a logical response to the immediate situation.

The term Fundamental Attribution Error, is when you see your own behaviour being determined by situational factors, whereas you view the behaviour of others as attributable to the person's disposition.

For example, YOU failed the exam because you did no work. I on the other hand failed because the lecturer was no good, or the questions were unfair, or the subject is boring!

EVALUATION

  1. Read at your own peril!
  2. The Storms (1973) video experiment is described in a little more detail. The rest - Read at your own peril!
  3. Read at your own peril!
  4. There is a problem with the sentences used. Do the different types of sentences used by the subjects really reflect underlying beliefs? Does a subject know the difference between 'I want to make a lot of money' (dispositional), and 'chemistry is a high paying field'? This section is worth reading through.
  5. The self-serving bias (Miller & Ross, 1975) points out that we like to attribute our success to our disposition, and our failures to situational factors. An exception to this rule can be found in the clinically depressed.
  6. Zimbardo would agree that situational factors are more relevant than dispositional factors. Don't forget the psychiatrists, who were asked by Milgram to predict his results. They opted for a dispositional explanation 'Americans won't do that sort of thing'. Is this because they overuse this dispositional attribution when treating patients?
  7. This area of research suffers from ethnocentrism. There is evidence to suggest those, in the East, attribute situational factors to others, rather than to oneself.

Return to Gary Sturt's homepage