Jones and Nisbett argue that observers are more likely to make dispositional attributions when explaining the behaviour of others whilst the actors themselves are more likely to consider the full range of causes of their own behaviour and are thus more likely to make situational or external attributions. However, Miller and Ross (1975) indicate that the soc-called actor-observer bias is most applicable to negative events and may in fact be reversed completely when actors consider their own achievements or successes. They describe two biases which have collectively been termed the self serving bias. This encompasses the self enhancing bias, where the individual explains his or her achievements in terms of internal factors or disposition and the self protecting bias where s/he ‘blames’ negative outcomes or failures upon situational factors outside of his or her own control.
This form of attributional bias is very well supported and there are many examples of self serving bias in the literature. For example Lau and Russell (1980) explored this concept by conducting a content analysis of attributions made in the sports pages of American newspapers. They compared the explanations given by sportswriters with interviews with the America Football coaches and players. They found that the coaches and players were more likely to credit their wins to internal factors such as hard work, the good shape of the team, natural talent etc and losses to external factors such as injuries, weather, foul play from the opposing team, etc.
This is a useful study as its demonstrates the use of self serving bias in qualitative data that has been gathered in the real world and not as part of a laboratory study, suggesting that the conclusions may be considered to be high in ecological validity and less likely to result from demand characteristics where participants are aiming to please the interviewer by answering in a way which they believe suits the purpose of the study. Another strength of the study is that the fact they analysed the explanations given by the sportswriters meaning that they also have some baseline data with which to compare the explanations given by the coaches and players. This means the researchers have a more accurate picture of the extent of the bias exhibited.
The reliability of this finding is further supported by work by Johnson et al (1964) in a study where teachers were asked to explain the reasons behind their pupils success and failure; when pupils did badly teachers attributed this to the pupils’ lack of effort or ability and saw it as the pupils’ responsibility (self-protecting bias) however over time as the pupils began to improve, the teachers altered their attributions and explained their pupils’ successes as down to their own efforts as teachers; they saw themselves as responsible for the pupils improvement, (self enhancing bias). This study also has its strengths in that the longitudinal design meant that it was possible to compare the same people’s explanations over time.
This study also has high ecological validity as the pupils’ attainment was varying naturally rather than being artificially manipulated. However, this means that there is a lack of control over confounding variables which could jeopardise the internal validity of the finding. This said, there are examples of self serving bias under laboratory conditions whereby participants have been given a lecture to listen to and a test to take based on the lecture. The Pps are then told that they have achieved either an A, B or a C but this is randomly assigned regardless of actual test performance. The participants are then asked to make attributions about their performance.
Those pupils who thought that they had done well tended to attribute this to their own intelligence and effort applied while those who didn’t do so well attributed it to the poor quality of the lecturer and test difficulty, (Snyder and Clair 1976). Although, this study has higher internal validity due to the random allocation of pupils to groups, the ethics of the study are questionable as participants are deceived and led to believe that they have not done well and this may cause some psychological distress.
Despite the research quoted above, Campbell and Sedikides (1999) have provided evidence which questions the validity of the self-protecting bias although the self enhancing bias seems well supported. In a meta-analysis of many studies in this area failure was not attributed to situational factors as much as success was attributed to dispositional factors. This findings has been accounted for by Duval and de Silvia (2002) who highlight the role of expectations for the future in determining how attributions are made. If participants believe that failure may be temporary, (unstable) they are more likely to attribute dispositionally than if they believe that they will not be able to improve at a later date, in which case the self protecting bias, (attributing failure externally is more common).
There are two main explanations for the self serving bias; the motivational account suggests that we are motivated to maintain high self esteem and prevent the loss of self esteem, (Greenberg, 1982). This is part of ‘impression management’ theory and the argument here is that we should attempt to present ourselves to others in the best possible light. We might assume that this serves an evolutionary function with regard to sexual selection, (mate choice) and also natural selection, as this way of thinking encourages perseverance as even when we are failing we are encouraged to continue.