The term self-serving bias is most often used to describe a pattern of biased causal inference, in which praise or blame depend on whether success or failure was achieved. For example, a student who gets a good grade on an exam might say, "I got an A because I am intelligent and I studied hard!" whereas a student who does poorly on an exam might say, "The teacher gave me an F because he does not like me!" When someone strategically strives to facilitate external causes for their poor performance (so that they will subsequently have a means to avoid blaming themselves for failure), it may be labeled self-handicapping.
Another example of self-serving bias can be found in the workplace. Victims of serious occupational accidents tend to attribute their accidents to external factors, whereas their coworkers and management tend to attribute the accidents to the victims' own action.
Several reasons have been proposed to explain the occurrence of self-serving bias. One class of explanation is motivational: people are motivated to protect their self-esteem, and so create causal explanations that serve to make them feel better. Another class of explanation focuses on strategic impression management: although people may not believe the content of a self-serving utterance, they may nevertheless offer it to others in order to create a favorable impression. Yet another class of explanation focuses on basic mechanisms of memory: people might have internal reasons for success more available in memory than external reasons (Miller & Ross, 1975; Roese & Olson, 2007).
The problem was recognised by A.V. Dicey in his Lectures on the Relation Between the Law and Public Opinion in England:
A man’s interest gives a bias to his judgment far oftener than it corrupts his heart... He overestimates and keeps constantly before his mind the strength of the arguments in favour of, and underestimates, or never considers at all, the force of the arguments against
Self-serving bias may result in bargaining impasse if each side interprets the facts of the dispute in their own favor. In this situation one or more of the parties may refuse to continue negotiating, believing the other side is either bluffing or refusing to accept a reasonable settlement and thus deserves to be ‘punished’.
There is a good deal of experimental evidence to support this hypothesis. In one experiment (described in Babcock & Lowenstein, 1997) which assigned participants to either the plaintiff or defendant in a hypothetical automotive accident tort case with a maximum potential damages payment of $100,000, the plaintiff’s prediction of the likely judicial award was on average $14,500 higher than the defendant’s. The plaintiff’s average nomination of a ‘fair’ figure was $17,700 higher than the defendant’s. When parties subsequently attempted to negotiate a settlement agreement, the discrepancy between the two sides' assessment of a fair compensation figure strongly correlated with whether or not parties reached an agreement within a set period of time. This experiment was conducted with real money with one real dollar being equal to $10,000 experimental dollars and if parties did not reach a negotiated agreement the case was decided by a third party and each side had to pay costly court and legal fees.
Group-serving bias is a similar bias on the group level.