Although Kelly’s theory has to be given recognition for being a positive development in the area of attribution theories, debate nonetheless still is evident in considering if it details the methods we use in everyday life. Research that has aimed to test the model tends to generally conclude that the model is not a correct description of how we determine the causes of behaviours. However, a study carried out during our laboratory class seemed to suggest findings in support of the ANOVA model. The aim was to compare how well Kelley’s model predicted attributions for simple opposed to complex behaviours using constrained and unconstrained responses, (full details of the experiment can be seen in the appendix). It was discovered that when responses were constrained; 84 percent responses for simple problems and 92 percent of responses for complex problems used Kelly’s categories to assess the problems. And that that when responses were unconstrained; 61 percent responses for simple problems and 64 percent of responses for complex problems used Kelly’s categories to assess the problems. Although very little statistical difference was found between the simple and complex conditions, the research nonetheless shows that a majority of people use Kelley’s ANOVA model. The research is clearly supports the notion that the model is a valid working concept.
McArthur (1972) conducted one of the first experimental tests on Kelley’s model and her results were supportive, although it was discovered that consensus information is not used comparatively as much. This study and our own laboratory class findings are examples of the small amount of data available in total support of Kelley’s ANOVA model, there is much more empirical evidence which highlights the problems with the model.
Beattie and Anderson (1995), on the basis of detailed research, concluded that consistency, distinctiveness and consensus information are not usually asked for in everyday situations, this is specifically when they are asked open-ended questions. Furthermore it has been suggested that we do not always have this information available to us when making an explanation of an event. The model is criticised for being for being overly rational and deliberative and for ignoring the importance of culture and self-esteem in decision-making. Scott and Spencer (1998) suggest that it is best to consider Kelley's model as a ‘normative model’, models of how information ought to be used but in real life actually is not. There are biases and errors in the way people actually process information. These preconceptions and inaccuracies are exemplified in numerous different problems which have been studied separately by different psychologists. Who then pose alternative explanations of how everyday attributions are made, Kelley's ANOVA model fails to provide adequate explanations in all of these areas.
Initially Kelley failed to consider what is called the fundamental attribution error (Ross et al, 1977). This context refers to the tendency to underestimate situational factors and overestimate dispositional factors in making attributions. Ross et al (1977) carried out a study including quiz masters and contestants, they found that even though the quiz masters were in a position where they could ask any questions they wanted, people thought automatically that the quiz masters were more knowledgeable. This fundamental attribution error may also serve to explain why we tend to blame road accidents on the driver not the conditions. However even the concept of fundamental attribution errors is not completely ubiquitous as it has been proven by Miller (1984) that cultural differences are evident, with individualistic societies, such as Americans, paying more attention to dispositional factors and collectivist cultures, such as Indians emphasising situational factors. Linguistic differences in the West also play a role in this; the English language contains more terms that describe a person’s dispositions than we do to describe different situations. Even though the fundamental attribution error is not a universal concept itself, it is a major flaw in Kelley’s theory that it is not considered at all.
Furthermore Kelley’s theory did not include any evidence of Actor Observer Differences, which are a tendency for actors to attribute their behaviour to the situation and for observers to attribute the observed behaviours to the individual. An example of this is seen in Jones and Nisbett’s (1972) study where they found that male students were more likely to describe their own choice of girlfriend and course at university in terms of external factors, however on describing their best friends choices these were attributed to dispositional factors. In another study it was discovered that people were more likely to say their behaviour depends on the situation than other peoples behaviours Therefore, attributions can not solely be made in the way that Kelley described, as distinctiveness information is considered differently and different attributions would be assumed. This and other evidence suggests that there is a different focus of attention for actors and observers, and actors know how their own behaviour varies across contexts. Kelley's model is also flawed in that it does not take into account any differences in perceptual focus. Any given event is usually a completely different experience for the actor and observer.
It has further been suggested that the ANOVA model fails to incorporate or acknowledge the existence of the false consensus effect. This is where we have a tendency to use our own attitudes and behaviours for deciding the consensus for a specific behaviour and furthermore to overestimate the number of people who share our beliefs and habits. The evidence available suggests that people do not, as Kelley suggested, search for consensus information, we assume all people share our own individual beliefs. Ross et al (1977) gave empirical support to this argument in their finding that when respondents were interviewed they believed over 60 percent of people, would give the same response as them. False attributions are clearly evident, one only has to think about when people who achieve high grades tell others about them and a distinct tendency to underestimate others success is apparent. The existence of the false consensus effect clearly shows that Kelley was incorrect to assume that consensus information played such a distinct role in making attributions.
Kelley’s ANOVA model is further undermined by the existence of a self-serving bias, which is a tendency to attribute success to internal causes and failure to external causes. For example Williams et al (1979) found that success in exams was attributed to ‘intelligence and work ethic but failure was attributed to unreasonable lecturers and bad luck’, thus participants in this experiment clearly did not use Kelley’s modal of attribution, as their attributions were varied dependant on whether it made them look good or not. The ANOVA model does not take into account the existence of any self-centred bias either.
Kelley’s model did not highlight the existence of unrealistic optimism on the part of an individual. This refers to the belief that ‘I am at least slightly better than average’ and that good things will happen to oneself and bad things will not. For example Manstead et al (1992) found that even car drivers who had been hospitalised after an accident believed that they were ‘better than average’ drivers. Overall there is a mass of evidence for the presence of self serving biases, thus, when making attributions there is a strong motivational basis to protect or enhance ones self esteem; which Kelly did not integrate into his model.
The literature in general does not seem to support the use of the ANOVA model in everyday situations; although elements may be used they are not descriptions of actual processes that people normally go through. When making attributions, ‘our explanations are not affected by the information we have available, they are also influenced by the culture we live in, what seems to be most salient to us at the time, and by our need for self-esteem,’ (Scott and Spencer, 1998). The ANOVA model does not take into account the fact that we rely more on biased background information than details of the event and factors specified by the theory. The main criticism of the model is that it is overly rational and deliberative, and that it ignores the importance of culture and self-esteem, as previously explained (Hewstone, 1989 cited in Scott and Spencer, 1998). There is also evidence to suggest that we are not able to handle such complex statistical procedures as suggested by the model, as we lack cognitive resources; consequently, we often rely on cognitive heuristics instead.
Kelley’s model is ultimately flawed, although it is possibly a good account of how people should use information, it is not through enough to explain what happens in everyday life. However it deserves credit for being one of the most through and useful models devised at the time. It has also stimulated further research into the topic and models devised later by other psychologists have helped to make the theory more workable. For example, Forsterling’s (1992) formulation of Kelley’s ANOVA model was based on the statistical anova parameter of effect size, and thus accounted for discounting augmentation as well as covariation.
The results from our laboratory class and other pieces of research that have been in support of the ANOVA model cannot be ignored. However, with so much evidence to suggest that the MODEL is not an effective working model, one has to question whether supportive research evidence may have been flawed. As even with the introduction of Kelley’s causal schemas the theory is still unworkable. Overall the theory seems to complex and rigid to be part of everyday attributions, however in an ideal world it may be helpful to use Kelley’s ANOVA model to make unbiased, correct and thought through assumptions.
Bibliography
Scott, Peter and Spencer, Christopher (1998). PSYCHOLOGY: A CONTEMPORY INTRODUCTION. Oxford, Blackwell Publishing.
Appendix
EXPERIMENT 1
Compare how well Kelley’s model predicts attributions for simple versus complex behaviours using constrained responses.
Condition 1: Simple behaviour: “Sue is afraid of the dog.”
Condition 2: Complex behaviour: “The Conservative Party called Tony Blair a liar about promises of investment in the NHS.”
Condition 1 (Simple behaviour and response constrained)
Sue is afraid of the dog
- Other people are also afraid of the dog
- Sue is not afraid of any other dog
- In the past, Sue has always been afraid of this dog
Why did this happen?
- Something about Sue?
- Something about the dog?
- Something about the particular occasion?
Condition 2 (Complex behaviour, responses constrained)
The Conservative Party called Tony Blair a liar about promises of investment in the NHS.
- Other political parties have also called Tony Blair a liar about promises of investment in the NHS.
- The Conservative party have not called Tony Blair a liar about other issues.
- In the past, the Conservative Party have frequently called Tony Blair a liar about promises of investment in the NHS.
Why did this happen?
- Something about the Conservative Party?
- Something to do with the issue of investment in the NHS?
- Something about the particular occasion?
EXPERIMENT 2
Compare how well Kelley’s model predicts information search (unconstrained response) for simple versus complex behaviours.
Condition 1: Simple behaviour: “Sue is afraid of the dog.”
Condition 2: Complex behaviour: “The Conservative Party called Tony Blair a liar about promises of investment in the NHS.”
Condition 1 (Simple behaviour, unconstrained response)
Sue is afraid of the dog
What information do you need to decide that this happened because of….?
- Something about Sue?
- Something about the dog?
- Something about the particular occasion?
Condition 2 (Complex behaviour, unconstrained response)
The Conservative Party called Tony Blair a liar about promises of investment in the NHS.
What information do you need to decide that this happened because of…?
- Something about the Conservative Party?
- Something to do with the issue of investment in the NHS?
- Something about the particular occasion?