Kunda, Z., Movitated Inference:; Self-serving generation and evauation of causal theories., JPSP, 1987, 53, 636-647.

People hold the general theory that good things will happen to them and bad things will not. They generate and evaluate caual theories in a self-serving manner. "They generate theories that view their own attributes as more predictive of desirable outcomes, and they are reluctant to believe in theories relating their own attributes to undesirable outcomes".

Most people are optimistic. They think they will have more positive outcomes than others, and fewer negative outcomes. No matter what their attributes are, they can convince themselves that their attributes will help them achieve desired outcomes and avoid feared outcomes.

This self-serving is possible because people have the ability to generate causal linkages linking almost any attribute to an outcome, and they have no way of determining correctness. Self-generated theories last longer in memory and are more resistant to change.

People also scrutinize information with negative impliations for self and accept at "face value" information with positive implications. They can apply different rules at differetn times and don't know the appropriateness of a given rule.

While many researchers argue whether this self-serving behavior is motivational or inferential, the researcher believes that the cognitive apparatus is used in the service of motivational ends. Thus the conclusions appear supported by the evidence, but the method of filtering evidence was biased.

Procedure
103 Michigan undergraduates evaluated a description of a person in terms of that person's likelihood of divorce, and then compared themselves to the person and evaluated their own chances of divorce. Half read a peson with one set of attibutes, half read with the opposite set. One third of descriptions were divorced, 1/3 maarried, and 1/3 no mention. All were supposedly U of M grads in the 60's.

Results
People rated a particular attribute as more suitable for a stable marraige when they themselves possessed the same attribute. It was true for even opposite attributes. People's causal theories are shaped by personal motivations. In this study though the average divorce rate is 50% subjects feilt their own rate is 20%.

Study 2
Was designed to test whether people with no motivation toward the topic (success in medical or law school) would still make biased predictions. The results were the same as study one: People tend to view their own attributes as more important predictors of success of other people. Thus people generate theories to bolster their own self-image.

Study 3
was designed to see if people selectively interpreted facts in a medical article to support their own self-serving theories. Subjects read an article on negative health effects of caffeine. The results confirm that the tend to disbelieve information that suggects negative effects for their behavior.

These studies suggest that people generate and evaluate causal theories ina self-serving manner only when they are motivated to do so.