The relationship between attitudes and behaviour
Once we've established people's attitudes, can we then accurately predict how they'll behave? Rosenberg & HovIand's (1960) three‑components model (The ABC model: affective – behavioural – cognitive) implies that the behavioural component will be highly correlated with the cognitive and affective components.
An early study which shows the inconsistency of attitudes and behaviour is that of LaPiere (1934).
in 1930 and for the next two years, LaPiere travelled around the
However, when each of the 251 establishments visited was sent a letter six months later asking: 'Will you accept members of the Chinese race as guests in your establishment?', 91 per cent of the 128 which responded gave an emphatic 'No'. One establishment gave an unqualified 'Yes' and the rest said 'Undecided: depends upon circumstances'.
Influences on behaviour
It's generally agreed that attitudes form only one determinant of behaviour. They represent predispositions to behave in particular ways, but how we actually act in a particular situation will depend on the immediate consequences of our behaviour, how we think others will evaluate our actions, and habitual ways of behaving in those kinds of situations. In addition, there may be specific situational factors influencing behaviour. For example, in the LaPiere study, the high quality of his Chinese friends' clothes and luggage and their politeness, together with the presence of LaPiere himself, may have made it more difficult to show overt prejudice. Thus, sometimes we experience a conflict of attitudes, and behaviour may represent a compromise between them.
Compatibility between attitudes and behaviour
The same attitude may be expressed in a variety of ways. For example, having a positive attitude towards the Labour Party doesn't necessarily mean that you actually become a member, or that you attend public meetings. But if you don't vote Labour in a general election, people may question your attitude. In other words, an attitude should predict behaviour to some extent, even if this is extremely limited and specific.
Indeed, Azjen & Fishbein (1977) argue that attitudes can predict behaviour, provided that both are assessed at the same level of generality. There needs to be a high degree of compatibility (or correspondence) between them. They argue that much of the earlier research (LaPiere's study included) suffered from either trying to predict specific behaviours from general attitudes, or vice versa, and this accounts for the generally low correlations. A study by Davidson and Jaccard tried to overcome this limitation.
Attitudes can predict behaviour if you ask the right questions (Davidson & Jaccard, 1979) Davidson and Jaccard analysed correlations between married women's attitudes towards birth control and their actual use of oral contraceptives during the two years following the study.
When 'attitude towards birth control' was used as the attitude measure, the correlation was 0.08. Clearly, the correspondence here was very low. But when 'attitudes towards oral contraceptives' were measured, the correlation rose to 0.32, and when 'attitudes towards using oral contraceptives' were measured, the correlation rose still further to 0.53. Finally, when 'attitudes towards using oral contraceptives during the next two years' was used, it rose still further, to 0.57. Clearly, in the last three cases, correspondence was much higher.
According to Ajzen and Fishbein, every single instance of behaviour involves four specific elements:
a specific action
performed with respect to a given target
in a given context
at a given point in time.
According to the principle of compatibility, measures of attitude and behaviour are compatible to the extent that the target, action, context and time element are assessed at identical levels of generality or specificity (Ajzen, 1988).
For example, a person's attitude towards a 'healthy lifestyle' only specifies the target, leaving the other three unspecified. A behavioural measure that would be compatible with this global attitude would have to aggregate a wide range of health behaviour across different contexts and times (Stroebe, 2000). Elaborating the psychological processes underlying the principle of compatibility, Ajzen (1996) suggested that to:
'... the extent that the beliefs salient at the time of attitude assessment are also salient when plans are formulated or executed, strong attitude‑behaviour correlations are expected'.
The reliability and consistency of behaviour
Many of the classic studies which failed to find an attitude‑behaviour relationship assessed just single instances of behaviour (Stroebe, 2000). As we noted earlier when discussing the LaPiere study, behaviour depends on many factors in addition to the attitude. This makes a single instance of behaviour an unreliable indicator of an attitude Jonas et al., 1995). Only by sampling many instances of the behaviour will the influence of specific factors 'cancel out'. This aggregation principle (Fishbein & Ajzen, 1974) has been demonstrated in a number of studies.
According to Hogg & Vaughan (1995), what has emerged in the 1980s and 1990s is a view that attitudes and overt behaviour aren't related in a simple one‑to‑one fashion. In order to predict someone's behaviour, it must be possible to account for the interaction between attitudes, beliefs and behavioural intentions, as well as how all of these connect with the later action. One attempt to formalise these links is the theory of reasoned action (TRA) (Ajzen & Fishbein, 1970; Fishbein & Ajzen, 1975). This is discussed in relation to health behaviour in Chapter 12.
The strength of attitudes
Most modern theories agree that attitudes are represented in memory, and that an attitude's accessibility can exert a strong influence on behaviour (Fazio, 1986: see Chapter 17). By definition, strong attitudes exert more influence over behaviour, because they can be automatically activated. One factor that seems to be important is direct experience. For example, Fazio & Zanna (1978) found that measures of students' attitudes towards psychology experiments were better predictors of their future participation if they'd already taken part in several experiments than if they'd only read about them. This can be explained by the mere exposure effect (Zajonc, 1968), according to which the more contact we have with something or somebody, the more we like them (see Chapter 28).
So attitudes don't predict behaviour: what's the problem? The so‑called attitude‑behaviour problem, that is, the failure to find a reliable relationship between attitudes and behaviour, threatened to undermine the entire study of attitudes. As we saw in the Introduction and overview, attitude research was a cornerstone of social psychology in general, and social cognition in particular, for much of their history (Stainton Rogers et al., 1995).
But from the perspective of discursive psychology, there's no reason to expect such a correlation: inconsistency between attitudes and behaviour is what we'd expect to find. Traditional, mainstream, attitude research is based on the fallacy of individualism (see Chapter 3), according to which attitudes 'belong' to individuals. This implies something fairly constant, and which is expressed and reflected in behaviour. From a discursive perspective, attitudes are versions of the world that are constructed by people in the course of their interactions with others.
Discursive psychology is concerned with action, as distinct from cognition. In saying or writing things, people are performing actions, whose nature can in revealed through a detailed study of the discourse (e.g. recordings of everyday conversations, newspaper articles, TV programmes). Social psychologists have underestimated the centrality of conflict in social life; an analysis of rhetoric highlights the point that people's versions of events, and their own mental life, are part of ongoing arguments, debates and dialogues (Billig, 1987, 1992, in Potter, 1996).
Compared with traditional attitude research, discursive psychology tries to shift the focus away from single, isolated, individuals towards interactions between individuals and groups, a more relational or distributed focus (Potter, 1996).