Attitudes can change the world. This is exemplified by Barack Obama’s “Yes we can!” presidential campaign. They are evaluations about people, objects, places and behaviours. Eagly and Chaiken (2007) defined ‘attitude’ as “a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavor” (p.589). Also, “attitudes are likes and dislikes” (Bem, 1970, p.14). Much research has been carried out in understanding the construction, stability, utility and malleability of attitudes. This paper endeavours to deconstruct ‘attitude’ and in turn compare and contrast its place as a construct in Festinger’s (1957) Cognitive Dissonance Theory with Ajzen’s (1991) Theory of Planned Behaviour.
Firstly, it is necessary to disentangle the components that lead to attitude formation. Fishbein and Ajzen (1975) advocated that cognition was the core element to generating attitudes. Attitudes were felt to be generated through the expectancy-value model and projection of explicit self-beliefs. In line with this school of thought, affect associated with judgements was deemed to be mediated by cognition. However, more recent research indicates that this view is exclusionary to the extent that affect is also predictive of attitudes. It is with this premise that Bodur, Brinberg and Coupey (2000) examined attitudes concerning HIV, AIDS and drink driving prevention behaviours. In predicting attitudes, affect was found to be a separate construct from cognition. For example, when evaluating attitudes about being drunk, one might revert to affective elements such as feeling disoriented as opposed to what they were thinking at the time. In turn, this may impact future behavioural judgements about alcohol intake.
The dispositional nature of attitudes remains under fire from conflicting paradigms. According to Schwarz (2007), attitudes are contextually dependent, hypothetical and unobservable constructs. Ultimately, he concedes that people don’t have pre-conceived attitudes as they are formed instantly with the aid of contextual cues. Previous evaluations are felt to be stored as semantic knowledge. They are not believed to impact current appraisals or attitudes (Schwarz, 2007).
Fazio (2007) challenges the view that attitudes are flimsy, hypothetical, latent variables. Instead they are felt to reside in memory, whereby semantic attitudes are held about people, objects, places and behaviours. This is further supported by the idea that attitudes are represented by connectionist models in memory. Stronger attitudes are deemed more easily accessible (Conrey & Smith, 2007). Stereotypical attitudes held about groups are formed from pre-conceived attitudes. These may form the basis for adequate or inadequate current appraisals.
Although Fazio (2007) emphasizes the stability of attitudes in a trait-like manner, some have also been found to be context dependent. Black faces presented in front of a church scene elicited more positive responses on an implicit attitudes test than those who were part of an urban street background (Wittenbrink, Judd, & Park, 2001). The representation of an attitude is dependent on what feature is primed. This is exemplified by the attitude construal model (Schwarz & Bohner, 2001). It outlines how variability in attitudes is due to context influences. However, this model is not resilient. It is criticized for failing to address the processes that lead to attitude stability.
Strongly held or ‘crystallized’ attitudes are felt to be more consistent over time. Fazio (2007) concludes that the constructionist nature of attitudes only take place when a novel situation or entity is involved. There are no previous evaluations to serve as benchmarks. In this scenario, situational cues serve as exemplars. Strong attitudes such as “I love my family” are less susceptible to context effects. ‘Non-attitudes’ don’t have pre-conceived associations in memory, therefore, it is theorized that they are constructed on the spot (Schwarz, 2007).
In light of this, to what extent are attitudes and behaviour predictive of each other? Wicker (1969) advocated that attitude reports and behaviour correlated “rarely above .30, and often near zero” (p.65). More recent research has concluded that this relationship has increased to r =.40 (Schwarz, 2007). Again, there are conflicting views on this issue. According to Schwarz (2007), they are only predictive of each other when similar contexts are at play between judgement formation (1) and proceeding behaviour (2). This is known as a matching principle. For example, inconsistencies in mood at time 1 and 2 may fail to replicate corresponding behaviour (Schwarz & Clore, 1983). However, it could be construed that the latter may pertain to weaker attitudes. Strength in attitude has been found to predict behaviour. For example, those who have a strong interest in a particular attitude were found to have higher correlations with carrying out the proceeding behaviour. Students who felt strongly about a proposed unfair exam were more likely to write letters of protestation and hand out fliers in opposition. For those with strong vested interests in academic achievement, attitude-behaviour consistency demonstrated correlations of r = .60 – .74. This was in direct comparison to r = .24 – .42 for those with weaker attitudes (Sivacek & Crano, 1982). It would appear that attitudes do form the basis for predicting behaviour. However, this is a highly complex process and dependent on numerous factors.
Festinger’s (1957) Cognitive Dissonance Theory (CDT) demonstrates how conflicting cognitions cause people to strive for greater cognitive consistency. Such knowledge inconsistencies are unpleasant as people are motivated by a desire for consistency. Levels of homeostasis are implicated in well-being. This is also seen in self-concept clarity whereby the people those with a strong and consistent sense of self are deem to be well adjusted (Campbell, Trapnell, Heine, Katz, Lavallee, & Lehman, 1996). CDT posits that dissonance is experienced when two inconsistent cognitions are experienced. For example, “I like Obama” and “I’m a strong Republican” are dissonant cognitions. It leads to a negative drive like state which is characterised by psychological discomfort (Festinger, 1957). According to Aronson (1968), the magnitude of dissonance felt is related to the number of consonant and dissonant cognitions competing with the feature cognition. For example, in Festinger and Carlsmith’s (1959) iconic ‘induced compliance’ study, the magnitude of $20 (consonance) vs. $1 was much stronger in preventing dissonance. Festinger (1957) theorized that dissonance leads to mental frustration.
Levels of mental frustration associated with dissonance can be reduced by applying specific techniques such as changing attitudes towards a behaviour, changing ones perception of the behaviour, adding consonant cognitions, minimizing the importance of the conflict, reducing the perceived availability of choice and effort justification. If people are encouraged to behave in a way that is contrary to an original attitude, cognitive inconsistencies will ensue. Therefore in order to reduce dissonance, attitudes will be shifted in favour of appraising the new behaviour. This in turn serves to restore attitudes to a level of consistency or homeostasis (Festinger, 1957).
Attitudes are also fundamental to the Theory of planned behaviour (TPB; Ajzen, 1991). The theory posits that explicit intentions to act in a given manner are strong predictors of ensuing behaviour. Intentions are felt to comprise of attitudes towards engaging in a target behaviour and subjective norms (the extent to peers, family and societal norms encourage behaviour). Exclusively, the first two components were known as the Theory of reasoned action (Fishbein and Ajzen, 1975). It became known as the TPB when a third component was added. Perceived behavioural control (PBC) reflects a person’s perceived ability in carrying out a particular behaviour. While not synonymous with self-efficacy, it is very similar to it.
Both theories execute differing conceptions of attitudes. TPB radiates Fishbein and Ajzen’s (1975) premise that attitudes are derived from self-beliefs and their perceived strength. The affective component to forming attitudes seems prominent in Festinger’s (1957) Cognitive Dissonance Theory. Unpleasantness such as mental frustration and negative emotional drive states associated with dissonance may be alleviated when counter-attitudinal changes take place. Such unpleasantness is felt to be instrumental in instigating attitude change. Hence, both affect and cognition are predictors of attitudes in CDT (Festinger, 1957).
However, a third component, behaviour, may also be instrumental in forming attitudes. Previous behavioural experience may become indented in memory. From the TPB framework, outcomes associated with behaviours may lead to changes in attitudes. For example, on intending to write an essay, the task might be construed as difficult and unpleasant. However, when the behaviour is carried out, it might be re-assessed as more pleasant than previously believed. This could potentially facilitate an attitude change towards future essay-writing. Behavioural information is a key element to attitude formation in CDT. If a person has behaved in a particular manner that is inconsistent with cognitions, dissonance ensues. As a particular behaviour may be difficult to change, people may strive to alter attitudes in favour of the behaviour. Therefore, in TPB but more significantly in CDT, behaviour is also a predictor of attitudes.
Within both theories, to what extent are attitudes stable? CDT underscores the conflictual and complex nature of attitudes. Festinger and Carlsmith (1959) found that the higher the reward/ punishment people have for engaging in a counter attitudinal behaviour, the less dissonance they will feel. Being compensated $20 for partaking in a boring task serves as justification and they are in turn less likely to change their attitudes as a result of dissonance. However if rewards are low, they will need to somehow justify their participation by reporting that the task was in fact interesting. Attitude change in this scenario reduces the mental frustration (dissonance) associated with performing a boring task for unjust rewards. A level of cognitive homeostasis is re-instated. However, the level of attitudinal malleability revolves around the strength of the attitude (Eagly & Chaiken, 2007). If a strong attitude is resilient to change, then other coping techniques are implemented. For example, “I want to party tonight” and “I have an exam tomorrow” might give rise to a consonant cognition such as “partying is good for academic performance”. Thus dissonance is reduced (Festinger, 1957).
Attitudes within the TPB are contributory to intentions for behaviour. In order for a behaviour to take place, attitudes are likely to be congruent with social norms and perceived behavioural control. According to the principle of compatibility (Ajzen, 1988; Ajzen & Fishbein, 1977), when attitudes give rise to intentions that precede behaviour, the same “action, target, context and time element” is required (Ajzen, 2005, p.183). If attitudes are easily malleable, this would render behaviour difficult to predict. Hence, it is suggested that the strength of attitude in carrying out a behaviour is key to maintaining stability (Chaiken & Eagly, 2007). Both CDT and TPB signify the necessity for consistency between attitudes and behaviour, albeit for differing reasons (e.g., reducing dissonance vs. facilitating intentions towards a behaviour).
Both theories allude to the impact of social factors on attitudes. According to Lashley (2009), cognitive dissonance was fundamental to the 2008 US presidential race. Spin doctors and the media were instrumental in instigating propaganda for and against their chosen candidates. Through denial of uncomplimentary facts, cherry picking positive attributes and issuing unproven testimonies, dissonance surrounding the presidential candidates and their attributes was reduced. These attitudes are in turn felt to influence the voters’ preferences and dissonance levels. In turn the voting public have a voting choice. Dissonance can be reduced by bolstering attributes of the chosen candidate. This is similar to Brehm’s (1956) research on free choice.
In the TPB, social factors are also pivotal in the attitude-behaviour relationship. From a social identity and self-categorization view, TPB would concede that social influence pressures are instrumental in guiding attitudes and behaviour. These are in turn internalized and adopted by the individual group member. However, these are only effective to the extent to which the person identifies with that particular group. High pressure and identity was associated with increased compliance (Chatzisarantis, Hagger & Smith, 2007).
This is in direct comparison with CDT as observed in Festinger and Carlsmith’s (1959) ‘forced compliance’ study. They found that the degree to which people are pressurised to change an attitude, the less likely they are to experience dissonance or concede to attitudinal-change. In this case, people can attribute the attitudinal change to social compliance pressures as opposed to choosing to adopt an attitudinal change. Hence, the less justification that people have for their dissonant behaviour the more they are motivated to change their attitude towards the dissonant behaviour. Volitional control is associated with dissonance (Chatzisarantis, Hagger, & Wang, 2008). Albeit in different ways, social influences have profound yet distinct impacts on both the TPB and CDT.
The applicability and viability of CDT and the TPB has been supported time and time again. Within the health behaviour domain, they have been pioneering. For example, Stice, Shaw, Becker, and Rohde (2008) found that dissonance based interventions were more successful in preventing eating disorders than psycho-educational appeals. Counter-attitudinal changes were also observed in students who wrote and delivered a speech advocating condom use versus those who merely prepared the speech. The former group elicited stronger attitudes towards intentions to use condoms (Aronson, Fried, & Stone, 1991).
The TPB has been instrumental in forecasting health behaviours. Through implementing the TPB, Hagger, Chatzisarantis, and Biddle (2001) found that self-efficacy and positive attitudes were central features to predicting intentions to engage in physical exercise. Another study reported that 45% of the variance associated with intention to exercise was associated with subjective norm, PBC and attitude. Interestingly, only 27% could be accounted for by intention and PBC alone (Hagger, Chatzisarantis, & Biddle, 2002). Thus it would seem that attitudes are predictive of behaviour. However, it is important to keep the principle of compatibility in mind. Attitudes are pertinent to the successful application of both theories. CDT focuses on changing attitudes to attain attitude-behaviour consistency while the TPB aims to predict courses of behaviour via attitudes, social norms and PBC.
Limitations apply to both theories. CDT has been challenged by alternative explanations. According to Bem’s (1967) Self-Perception Theory, monitoring one’s own behaviour facilitates being swayed towards self-persuasion. In this paradigm, by engaging in healthy eating, one will come to evaluate this behaviour favourably in place of eating junk food. Also, Impression Management Theory (Tedeschi, Schlenker, & Bonoma, 1971) suggests that ‘appearing’ consistent to others and not actually being consistent is the key element to cognitive levels of homeostasis. Such a theory is in line with Goffman’s ‘self-presentation’. Also CDT doesn’t measure the extent to which dissonance might be experienced differently by people. People may exude idiosyncratic tolerances.
TPB has been criticized for failing to adequately address the contribution of affect in generating attitudes. Also, it is based on self-report measures; respondents may distort the characteristics of reported attitudes. Intention stability and personality have been found to impact TPB outcomes, this renders the theory less resilient. In turn, it should be regarded as a flexible and less rigid framework (Hagger et al., 2007).
To conclude, this paper sought to define and conceptualize attitudes while investigating their place in both CDT and the TPB. If consistent contextual settings are in place, attitudes are the keys to unlocking and predicting behaviour. However, attitudes are highly complex and open to several interpretations. This was demonstrated on examining their significance to the above mentioned theories. It could be derived that Obama’s intention to become president derived from strong attitudes, societal influences and the perceived behavioural control to tackle America and make it a better place. Additionally, in order to reduce dissonance, Obama is expected to behave in a manner that is consistent with his overt attitudes toward foreign and domestic policies – otherwise, collective dissonance may ensue. Again, it is necessary to highlight the significance of attitudes and their links with behaviour. Future studies could assess the TPB by means of implicit attitude testing while also examining exactly how dissonance is reduced in CDT.