Jones, E.E. & Nisbett, R. E., The actorand the observer: Divergent peceptions of the causes of behavior." In E.E. Jones, D. E. Kanouse, H. H. Kelly, R.E. Nisbett, S. Valins, & B. Weiner, (Eds.), Attribution: Perceiveing the causes of behavior, 1971

In explaining personal actions, "the actor's view of his behavior emphasizes the role of environmental conditions at the moment of action. The observer's view emphasizes the causal role of stable dispositional properties of the actor".

One supporting study was with a rigged IQ test with subject-accomplice pairs. The subject got 10 of 30 right, but the accomplice either got 1)15 right scattered 2)15 rignt descending in accuracy or 3) 15 right ascending in accuracy. The questions were ambiguous, and success or failure was randomly assigned independent of the subjects actual answers.

Early accomplice success biased the subject, who ranked them higher in intelligence and biased their future expectation of performance. Late accomplice success has less of a good impression. The random success accomplice was intermediate.

When the tables were turned (accomplice did 10 of 30 random) and the subject was one of the above three conditions, the subject evaluation was quite different. When the subject had early success, they attributed later failure to the fact that the questions were harder at the end. Subjects with later success thought the questions got easier. Later success thought they would do beter in subsequent tests than early success subjects.

Another study of observer's attributions to student speeches in various choice situations suggests that people attach insufficient weight to the situational constraints on other's behavior. People tend to take behavior at face value.

Another study where observers interpreted simple sentences of "Gerge mistranslated a sentence" found that most people make attributions to the person and not the situation.

Another study by Nisbett and Caputo showed that student used more situational items to explain why they chose their major and their girlfriend. In describing the same behavior of their friend, they used more person attributes.

Information Available to Actor and Observer
One can break down information into cause data and effect data. Effect data consists of (what was done) (success or failure, reaction of recipient of action, etc.) and (acto'rs experiences - pleasure, anger, embarassement). Cause data is environmental (incentives, task difficulty) and intention data (what the actor meant to do).

An observer can read inner experience of the actor only from facial and gesture cues and knowledge of how he himself would feel in similar circumstances.

One also rarely has sufficient historical data to judge another person. The "observer is characteristically normative and nomothetic: He compares the actor with other actors and judges his atributes accordingly. The actor, on the other hand, is more inclined to use an ipsative or idographic reference scale: This actionis judged with reference to his other previous actions rather than the acts of other actors."

Differences in Information Processing
Different aspects of the available information are salient for actors and observers and this differential salience affects the course and outcome of the attributiona process. The action itself -- its topography, rhythm, style, and content -- is more salient to the observer than to the actor. The actor doesn't and can't observe his actions very closely. For the observer the situational clues are ignored.

Also we tend to regard our own reactions to action as "truth" and unbiased and correct. As children we don't learn to easily distinguish between primary qualities (color, mass, etc.) and evaluative qualities (beauty, funniness, etc.), primarily because there is a high degree of consensus in many of these features.

A person with more empathy toward an actor will often make similar attributions. With less entropy they attribute the actor's behavior less to the environment stimuli.

Still, the correlation between a certain behavior and past behavior is rather low (0.3). Trait scores in questionnaires are rarely predictive of behavior in a given situation. It seems that traits exist in the eye of the beholder than the mind of the actor. We also often confuse role responses with personality traits. The fact of physical constancy may also produce the illusion of behavioral and therefore dispositional consistency.

We also seem to have implicit trait theories in our head that make quick inferences based on available data. And once we make a judgement, the opportunity for construal to support this judgement is great. Actors also see everyone else as having more stable and predictable personality traits than he himself does.