Attribution is the process by which someone decides why an event occurred. Gilbert (1998) explains that Heider (1958, cited in Gilbert, 1998) sees it as being as vital as the need to eat or sleep. Hewstone (1983) suggests that we use it to preserve our self-esteem and to portray ourselves in a positive light. Psychologists have approached attribution from both angles, and here we evaluate their explanations for how we make these judgements.
Heider (1958, cited in Stainton Rogers, 1995) stated that we use a common sense approach; because we believe that our own behaviour is made for a reason, other people’s behaviour must be too. We therefore, as scientists seek to understand the reason behind people’s actions in order to understand their intentions and the world around us. This idea of common-sense is a problematic one and was criticised even before Heider’s theories came to light; Lenoble (1943, cited in Stainton Rogers, 2011) said that there are as many common-senses as there are civilisations, which might suggest then that an attribution theory based on common sense would only apply in specific cultures. Nevertheless, against the backdrop of western culture, Heider makes the distinction between two categories of attribution; dispositional (relating to the actor’s intentions) and situational (relating to external factors causing the event), and this dichotomy serves as a theme against which many later theories are positioned.
Jones and Davis (1965, cited in Gilbert, 1998) attempted to explain this with their Correspondent Inference theory. They noted that people assume that an individual’s behaviour is directly linked to their core personality traits, and therefore we will be more likely to attribute behaviour to an individual’s disposition. They described how we interpret behaviour using five factors; how unusual the behaviour is (the non-common effects), the extent to which it is socially desirable, whether the behaviour was chosen, whether behaviour was intentionally directed at the observer and whether the behaviour had hedonic value to the observer. If behaviour is rare, socially unacceptable, freely acted out, purposefully aimed at the observer and brings pleasure or pain, then the observer is more likely to attribute dispositional causes.
Of course, there are situations where only some of these five factors come into play and still, we attribute dispositional causality; Hogg & Vaughan (2011) observe that some behaviours can be unintentional, but can still result in observers attributing a dispositional cause such as the actor being an inconsiderate person. Nisbett & Ross’ studies (1980, cited in Hogg & Vaughan, 2011) also criticise Jones & Davis’ theory; illustrating that people don’t effectively take into consideration the non-common effects.
Kelley’s Co-Variational Model (1967, cited in Hogg & Vaughan, 2011) aligns with previous theories, in that it agrees that we are naive scientists, analysing how behaviour varies. However, Kelley says that we analyse events to see which situational factors co-vary with the behaviour to understand the cause. He states that we look at the distinctiveness of the behaviour (whether it occurs in conjunction with every stimulus), the consistency of the behaviour (whether this behaviour happens every time this stimulus is present) and the consensuality of the behaviour (if other people behave similarly). If the behaviours aren’t distinct, are consistent and non-consensual, then dispositional attribution is made confidently.
However, Hogg & Vaughan (2011) point out that in order to make a judgement about the consistency of behaviour, we would have to see this behaviour repeated and this is unlikely. Kelley countered contemporary criticism of this kind by claiming that when we’ve only witnessed one occurrence of behaviour, we draw on causal schemata which are beliefs which have formed from previous knowledge of similar situations (Kelley, 1972, cited in Hogg & Vaughan). We then use schemata to guide our current attributional decision. However, Fiedler (1982) criticises the idea of causal schemata, saying that empirical experiments on this subject were so far removed from real life that they couldn’t be considered reflective of reality.
However, Kelley’s theories weren’t only criticised for their reference to consistency. Alloy & Tabachnik (1984) demonstrate that even if we do use this Co-Variational Model, we are actually very poor at analysing accurately. Hilton & Knibbs (1988) show that co-variation is often masked by observers’ expectations of the person or event. Hogg and Vaughan (2011) also question whether people can and do analyse co-variance in real life, away from experimental settings.
Quattrone (1982) describes the process in a simpler way, suggesting that we use a discounting principle by which we make an initial dispositional judgement, but afterwards, when we take more time to consider, we will correct and move towards a situational conclusion. Gilbert (1988) concurs, saying that it is indeed speed and attention during attributional processing which determines causality.
What followed these broad theories, however, was a range of studies which uncovered systematic errors or biases at every stage of this process.
Ichheiser (1943, cited in Gilbert, 1998) and Ross (1977, cited in Hogg & Vaughan, 2011) state that we are pre-disposed to Fundamental Attribution Error (which has also been called Correspondence Bias) where we are more likely to make dispositional judgements than situational ones. Ichheiser considers that this is because individualistic Western culture has brought us up to believe in our individual qualities rather than external causes, but other reasons for this error have also been suggested. One oft-cited cause references the apparent saliency of behaviour versus the often unnoticed aspects of the background situation (Taylor & Fiske, 1978, cited in Hogg & Vaughan, 2011). Even further, the saliency of aspects of behaviour will also contribute to this error; Efran (1974) demonstrated that the halo effect of positive traits (such as attractiveness) influences people’s attribution of guilt in simulated jury tasks, despite all other variables being equal. Another reason for Correspondence Bias lies in the English language, which has been shown to be pre-disposed to attribute dispositional cause (Nisbett & Ross, 1980, cited in Hogg & Vaughan, 2011).
The Actor-Observer Effect (Watson, 1982) refines the Fundamental Attribution Error. This bias maintains that when we’re considering others’ behaviour then we are indeed likely to blame their disposition. However it posits that when discussing our own behaviour, we’re more likely to blame situational details. This bias can be seen in groups too, where in-groups defend their actions with situational reasons and attribute dispositional blame on out-groups (Pettigrew, 1979), offering an insight into a central aspect of ethnocentrism. Reasons for this effect are closely aligned with those for the Correspondence Bias (Hogg & Vaughan, 2011).
The False-Consensus Effect (Ross, Greene & House, 1977) state that we erroneously conclude that our behaviours are the average, normal benchmark and expect that most people will act in the same way as us. Any differing behaviours, therefore, are unusual and reveal something more about the actor’s disposition. Hogg & Vaughan (2011) put forward three causes for this effect; the saliency of our own behaviours drown out reflections on other choices of action, we choose to be around people similar to us, so they’re likely to behave similarly, and the idea that we consider our own behaviours to be right, so assume that other people will do too.
In the same vein are Self-Serving Biases (Miller & Ross, 1975), where we attribute to protect our ego. Examples of this can be seen when we attribute blame to the victims of assault, because it is easier to live in a just world (Lerner, 1980 cited in Lerner, 1987) than one where pain is indiscriminate and could affect us too. In contrast we can even self-handicap to protect our ego; Jones & Berglas (1978) give the example of choosing alcohol abuse so as to externalise the reason for unsatisfactory outcomes, all in order to protect our precious self-esteem. As we mentioned briefly earlier, Hewstone (1983) studied the functions of attribution. He found that attribution serves to do three things; control the situation by understanding the cause and therefore enabling avoidance of problems in the future, preserve self-esteem by attributing blame externally when we have failed and finally to show our best selves to others by positioning attribution in our favour. It is perhaps inevitable then, that the selfish nature of these functions, also contributes to the source of attribution errors themselves.
Psychologists have been quick to point out where we make inferential mistakes, but are all of these errors really judgements which we get wrong? In many studies, it’s the experimenter who ultimately decides what the acceptable norm or correct attributional conclusion should be, but are they right? (Kruglanski & Ajzen, 1983).
Semin & Manstead (1983, cited in Stainton Rogers, 2011) note that in many situations people don’t even consider the cause of events. The act of scientists asking participants why something happens forces them to consider attribution, so conclusions aren’t representative.
In addition, many of these experiments are reductionist; they have been conducted in laboratory conditions which aren’t illustrative of what happens in the real world (Funder, 1987) and devoid of cultural influences (Hogg & Vaughan, 2011).
Culture makes a considerable difference to attributions; Schuster, Fersterling, & Weiner (1989) found that Indian attributions were much more fatalistic than those of western countries, and did not make the same attributional conclusions in experimental conditions. Beyond that, Evans-Pritchard (1937, cited in Hewstone, 1983) note that the example of Zande people who believe in witchcraft as a cause of events, would not fit neatly into any of the classical internal-external attribution theories.
Finally, are attribution errors important to understand? Are they even problematic? The consequences of making these errors aren’t necessarily negative (Funder, 1987); indeed, couples who blamed financial problems on the economy rather than individuals experienced higher levels of relationship satisfaction (Diamond & Hicks, 2012).
Criticisms of the classical theories and biases have been taken on board because later studies focus on more realistic test conditions and naturally occurring data. It is this research which has found that “other people’s single actions were not frequently brought up for explanation” (Antaki and Naji, 1987, cited in Edwards & Potter, 1993). Indeed, attribution was more likely to lie in root causes such as a general state of affairs which accounted for 39% of instances.
Studies have also looked in more detail at people’s motivation. Kruglanksi, Baldwin & Towson’s (1983) lay-epistemic analysis suggests that attribution is constructed according to the direction in which the individual needs causality to lie. This has been confirmed in more recent studies, (Miller & Bradbury, 1995) where factors such as whether attribution would improve or maintain problems, have shown to have significant impact on conclusions on causality.
Constructionist study of language and discourse forms the latest trend in attributional research, and, as Potter and Wetherell (1987, cited in Stainton Rogers, 2005) show, these theories position us not as naive but calculating negotiators of reality.
In the linguistic category model (Semin & Fiedler, 1988), it is only through describing the behaviour to others do individuals come to attribute causality. They create their argument in a way which presents these details rhetorically, and as facts, using verb choice and categorisation to actively define where attribution lies. Vickers (1982, cited in Stainton Rogers, 2011) even states that accountability can be entirely reversed simply with a different choice of verb.
Just a few years later Hilton (1990) presents us with the conversational model which proposes that attribution is constructed through dialogue, and that the discourse itself is an active part of the process. Individuals choose words, use rhetoric and position themselves in such a way to infer and illustrate attribution, and the respondent plays an active part of this process.
These last two theories have been criticised by Edwards & Potter (1993) for not focussing on the reason for the actual conversation, for not focussing on how much attention the people conversing are paying, and for not taking into account the level at which discourse analyses an â€œunderstood reality as opposed to a â€œreflection of realityâ€. The Discursive Action Model they propose aims to do just this; understand how discourse is actually â€œsocial action; a constructive attribution process. Edwards & Potter detail the need to look at situations such as blame negotiation and declining invitations – how do we learn this and what enables us to start on the path to this social action? Indeed, in Halverscheid & Witte (2008), when studying attributions of war and terrorism from Arabic and Western media, they concluded it was more appropriate to look at the ethical principles as the cause for attribution, rather than language or cognition of each culture. This study may not represent the further studies that Edwards & Potter had in mind but they illustrate the breadth of scope that needs to be covered in order to fully understand the attributional process.
Attributional studies have until recently taken an essentialist approach, focussing on an empirical and cognitive understanding of how we decide on causality. Critical social psychologists have since attempted to understand how personal motivation, expectation, culture and society influence attribution as well as the way we use language to actively construct attribution and to progress our understanding of attribution, this must continue using naturalistic experiments and naturally occurring data.