Writing

Theory of planned behaviour Cambridge Handbook of Psychology, Health and Medicine The theory of planned behaviour (TPB; Ajzen, 1991, 2002b), an extension of the theory of reasoned action (TRA; Ajzen & Fishbein, 1980; Fishbein & Ajzen, 1975), is widely used to study the cognitive determinants of health behaviours (Conner & Sparks, 2005; Sutton, 2004). It has several advantages over other ‘social cognition models’: (1) it is a general theory, and it can be argued that general theories should be preferred to health- or behaviour-specific theories for reasons of parsimony (Stroebe, 2000); (2) the constructs are clearly defined and the causal relationships between the constructs clearly specified; (3) there exist clear recommendations for how the constructs should be operationalized (Ajzen, 2002a); and (4) meta-analyses of observational studies show that the TPB accounts for a useful amount of variance in intentions and behaviour (but see the discussion of variance explained in Sutton, 2004). According to the theory, behaviour is determined by the strength of the person’s intention to perform that behaviour and the amount of actual control that the person has over performing the behaviour (Figure 1). According to Ajzen (2002b), intention is ‘the cognitive representation of a person’s readiness to perform a given behaviour, and . . . is considered to be the immediate antecedent of behaviour’, and actual behavioural control ‘. . . refers to the extent to which a person has the skills, resources and other prerequisites needed to perform a given behaviour’. Figure 1 also shows an arrow from perceived behavioural control to behaviour. ‘Perceived behavioural control’ refers to the person’s perceptions of their ability to perform the behaviour. It is similar to Bandura’s (1986) construct of self-efficacy (see ‘Perceived control’ and ‘Self-efficacy and health’). Indeed Ajzen (1991) states that the two constructs are synonymous. Perceived behavioural control is assumed to reflect actual behavioural control more or less accurately, as indicated by the arrow from actual to perceived behavioural control in Figure 1. To the extent that perceived behavioural control is an accurate reflection of actual behavioural control, it can, together with intention, be used to predict behaviour. The Theory of Planned Behaviour. The small, unlabelled arrows represent other, unspecified causes of the endogenous variables. The strength of a person’s intention is determined by three factors: (a) their attitude toward the behaviour, that is, their overall evaluation of performing the behaviour; (b) their subjective norm, that is, the extent to which they think that important others would want them to perform it; and (c) their perceived behavioural control. Attitude toward the behaviour is determined by the total set of accessible (or salient) behavioural beliefs about the personal consequences of performing the behaviour. Specifically, attitude is determined by ∑biei, where bi is belief strength and ei is outcome evaluation. Similarly, subjective norm is determined by the total set of accessible normative beliefs, that is, beliefs about the views of important others. Specifically, subjective norm is determined by ∑njmj, where nj is belief strength and mj is motivation to comply with the referent in question. Finally, perceived behavioural control is determined by accessible control beliefs, that is, beliefs about the presence of factors that may facilitate or impede performance of the behaviour. Specifically, perceived behavioural control is determined by ∑ckpk, where ck is belief strength (the perceived likelihood that a given control factor will be present) and pk is the perceived power of the control factor (the extent to which the control factor will make it easier or more difficult to perform the behaviour). The ‘principle of correspondence’ (Ajzen & Fishbein, 1977; Fishbein & Ajzen, 1975) or ‘compatibility’ (as it was renamed by Ajzen, 1988) states that, in order to maximize predictive power, all the variables in the theory should be measured at the same level of specificity or generality. This means that the measures should be matched with respect to the four components of action, target, time and context. The rationale given for the principle is a pragmatic one: it improves prediction. Presumably, however, there is also a theoretical rationale for the principle, namely that, by measuring the TPB variables at the same level of specificity, we are matching cause and effect (Sutton, 1998). Although the TPB holds that all behaviours are determined by the same limited set of variables, each behaviour is also substantively unique, in two senses (Fishbein, 2000). First, for a given population or culture, the relative importance of attitude, subjective norm and perceived behavioural control may vary across different behaviours. For example, some behaviours may be influenced mainly by attitude, whereas other behaviours may be influenced mainly by subjective norm. Ogden (2003) points out that many studies using the TPB find no role for one or other of the three putative determinants of intention and therefore that the theory ‘cannot be tested’. However, this represents a misunderstanding of the TPB. If at least one of the components is found to predict intention in a given study, this is consistent with the TPB. Nevertheless, it is a weakness of the theory that it does not specify the conditions under which intention will be mainly influenced by attitude, subjective norm or perceived behavioural control. The second sense in which each behaviour is substantively unique is that, for a given population or culture, the behavioural, normative and control beliefs that underlie attitude, subjective norm and perceived behavioural control respectively may also differ for different behaviours. In the same way, for a given behaviour, the relative importance of attitude, subjective norm and perceived behavioural control and the content of the underlying beliefs, may vary across different cultures or populations. The TPB is a general theory. In principle, it can be applied to any target behaviour without needing to be modified. For example, in applying the theory to a health-related behaviour, there should be no need to add a variable representing risk perceptions. If beliefs about the health risks of the behaviour (or its effect on reducing risk) are salient to a substantial proportion of the target population, this should emerge in an elicitation study that uses open-ended questions to elicit accessible beliefs (Ajzen, 2002a; Ajzen & Fishbein, 1980; for an example of an elicitation study, see Sutton et al., 2003). Like other theories of health behaviour, the TPB is a causal model and should be treated as such (Sutton, 2002a, 2004). It says, for instance, that if you hold constant a person’s subjective norm, perceived behavioural control and actual behavioural control and you change their attitude toward the behaviour, this will lead to a change in their intention (assuming that attitude is a determinant of intention for the behaviour in question in this target group), and this in turn will lead to a change in their probability of performing the behaviour (assuming that the person’s intention is stable and that the behaviour is at least partly under their control). The TPB is often depicted without actual control in the path diagram and, to date, has always been tested without measuring actual control. In this case, the direct path from perceived behavioural control to behaviour is causally ambiguous (Sutton, 2002a, 2002b). In particular, if we observe an independent predictive effect of perceived behavioural control on behaviour in an observational study in which actual control is not measured, this may be due partly to a causal effect of perceived behavioural control on behaviour and partly to a correlation induced by actual behavioural control influencing both perceived behavioural control and behaviour (Sutton, 2002a, 2002b). More generally, failing to measure and control for the effects of actual behavioural control will lead to biased estimates of the causal effects of perceived behavioural control and intention on behaviour, unless it can be assumed that perceived control is an accurate reflection of actual control (i.e. that perceived and actual control are perfectly correlated and this correlation arises from a direct causal effect of actual on perceived control). Although Figure 1 shows an arrow going directly from actual control to perceived control, this is inconsistent with the theory’s assumption that the effects of any variable on perceived control must be mediated by control beliefs. The absence of arrows, either one- or two-headed, between actual control and behavioural and normative beliefs respectively can be interpreted as indicating zero correlations and no direct causal influence in either direction. However, to date, Ajzen has not discussed these possible relationships. If actual control were related to one or both of these variables, again this would have implications for the interpretation of regression analyses from which actual control was omitted. The theory predicts an interaction between perceived behavioural control and intention on behaviour. Ajzen (2002b) states it as follows: ‘Conceptually, perceived behavioural control is expected to moderate the effect of intention on behaviour, such that a favourable intention produces the behaviour only when perceived behavioural control is strong’. He also notes that ‘In practice, intentions and perceptions of behavioural control are often found to have main effects on behaviour, but no significant interaction’ (see also Conner & Armitage, 1998). This interaction derives from an interaction between intention and actual control (and so would be predicted to occur only in situations in which perceptions of control are accurate). In particular, intention is expected to have a stronger influence on behaviour the greater the degree of actual control the person has over the behaviour. For simplicity, this interaction is not shown in Figure 1. The TPB is often described as a deliberative processing model. However, although some decisions may involve conscious deliberation and careful weighing up of pros and cons, in many cases the processes involved in the formation and modification of beliefs, attitudes and intentions may be largely automatic (Ajzen & Fishbein, 2000; Fishbein & Ajzen, 1975). A person’s attitude toward a particular behaviour may be automatically updated when new information about the behaviour is received, and this attitude may be automatically elicited and guide behaviour in relevant situations. (However, although it seems plausible that automatic processes control the formation and change of beliefs, attitudes and intentions, for most health-related behaviours it seems less plausible to suggest that behaviour itself is automatically elicited.) How well does the theory perform? Table 1 summarizes the findings from meta-analyses of observational studies of the TPB in terms of the multiple correlation (R) and its square (which can be interpreted as the proportion of variance explained) for predicting intention and behaviour. With the exception of Ajzen (1991), all the meta-analyses explicitly or by implication restricted the analysis of prediction of behaviour to prospective studies in which intention and perceived behavioural control were measured at time 1 and behaviour was measured at time 2. The findings for both intention and behaviour show reasonable consistency. For intention, the multiple correlations range from 0.59 to 0.71 (between 35% and 50% of variance explained). Prediction of behaviour was lower, as expected, with the multiple correlations ranging between 0.51 and 0.59 (between 26% and 35% of the variance explained). All the effect sizes in Table 1 are ‘large’ in terms of Cohen’s (1992) guidelines. Godin and Kok (1996) found differences between different kinds of behaviours with respect to how well the theory predicted intentions and behaviour. For example, for behaviour, the theory worked better in studies of HIV/AIDS-related behaviours than in studies of ‘clinical and screening’ behaviours. However, these results were based on small numbers of studies and possible confounds such as sample characteristics and differences in how the TPB variables were measured were not examined. Godin and Kok’s review needs to be updated and extended. Extensions of the TPB There have been numerous attempts to extend the TPB by adding variables such as anticipated regret, moral norm and self-identity (Conner & Armitage, 1998). For the sake of parsimony and theoretical coherence, candidate variables should be provisionally accepted as official components of the theory only if a number of conditions are satisfied. First, there should be sound theoretical reasons for believing that a given candidate variable influences intention or behaviour independently of the existing variables, that is, that the variable has a direct causal effect on intention or behaviour. In some cases, it is possible that the proposed additional variable is already captured by one of the existing variables. Table 1. Summary of findings from meta-analyses of the theory of planned behaviour Meta-analysis Effect sizea Predicting intention (BI) from AB, SN and PBC Predicting behaviour from BI and PBC kb R R2 k R R2 aEffect sizes are given in terms of the multiple correlation (R) and R2. bk is the number of datasets. cRestricted to studies of health-related behaviours. dRestricted to studies of condom use. eRestricted to studies of physical activity. fRestricted to studies that included measures of both ‘perceived difficulty’ and ‘perceived control’. Ajzen (1991) 19 .71 .50 17 .51 .26 Godin & Kok (1996)c 76 .64 .41 35 .58 .34 Sheeran & Taylor (1999)d 10 .65 .42 – – – Albarracín et al. (2001)d 23 .71 .50 23 .53 .28 Armitage & Conner (2001) 154 .63 .39 63 .52 .27 Hagger et al. (2002)e 49 .67 .45 35 .52 .27 Trafimow et al. (2002)f PBC as Perceived difficulty 11 .66 .44 9 .59 .35 PBC as Perceived control 11 .59 .35 9 .58 .34 Second, in order to retain the existing structure of the TPB, the proposed new variable should have an expectancy-value basis like attitude, subjective norm and perceived behavioural control; in other words, the new variable should be determined by accessible beliefs that are specific to the target behaviour. This would seem to rule out some variables, for example self-identity. This also means that the expectancy-value basis of descriptive norm (the belief that significant others are or are not performing the target behaviour), which Ajzen (2002a) has proposed as a sub-component of subjective norm in the latest version of the theory, needs to be specified. This requirement, that any additional variable is homologous in structure to the existing variables, also implies that including too many additional variables in the theory would make it unwieldy to use in practice. Furthermore, additional open-ended questions for eliciting accessible beliefs would need to be devised for use in pilot studies. This has not yet been done for descriptive norm (Ajzen, 2002a). Third, measures of a proposed new variable should be shown to have discriminant validity with respect to measures of the existing components, in other words to be measuring something different from measures of the existing variables (see ‘Health cognition assessment’). Finally, the new variable should be shown to predict intention and/or behaviour independently of the existing components in studies in which the latter are well measured in accordance with published recommendations. It is likely that there are many false positive findings in the literature because the existing components are not always optimally measured. Of course, if the aim is simply to improve the predictive power of the theory rather than to specify additional determinants of intention, only the last of the requirements set out above is relevant. External variables The TPB, like other social cognition models, does not rule out other causes of behaviour. Many other factors such as socio-demographic, cultural and personality factors may influence behaviour, but these are assumed to be distal factors, in other words to be farther removed from the behaviour than the proximal factors specified by the theory. Thus, the TPB divides the determinants of behaviour into two classes: a small number of proximal determinants, which are specified by the theory (i.e. are internal to the theory); and all other causes, which are left unspecified but which are assumed to be distal and to influence behaviour only via their effects on the proximal determinants. In this sense, the TPB is sometimes said to be ‘sufficient’. A strategy for guiding future research on the determinants of health behaviour is to continue to use the TPB as a model of the proximal determinants of a given behaviour and to specify external factors that are hypothesized to influence the components of the theory or to influence behaviour directly, that is to develop theories that relate external factors to the theory’s components. In effect, this is extending the causal model representing the TPB to the left, specifying the more distal causes of a particular behaviour and the mechanisms by which they influence the components of the theory and behaviour (Sutton, 2004). Intervention studies The TPB has direct implications for behaviour change interventions. According to the theory, changing behaviour requires changing either (1) the accessible beliefs that underlie attitude, subjective norm and perceived behavioural control or (2) actual behavioural control or (3) both (Sutton, 2002b). As Hobbis and Sutton (2005) point out, with the possible exception of Project RESPECT (Fishbein et al., 2001), there have so far been few successful examples of effective TRA- or TPB-based interventions. In their systematic review, Hardeman et al. (2002) identified seven studies in which the TPB was used to develop the intervention and the effectiveness of the intervention was evaluated in a randomized controlled trial with a behavioural outcome. Of these, four showed at least one positive change in the intervention group compared with the control group and the other three showed mixed effects. Although these findings seem encouraging at first glance, Hardeman et al. note a number of problems with these studies. In particular, in most cases it was not possible to identify which components of the theory were being targeted in the intervention. Furthermore, only one study (Brubaker & Fowler, 1990) reported a mediation analysis to test the extent to which the intervention effect on behaviour was mediated by the components of the theory. This study deserves further discussion as Fishbein and Ajzen (2005) cite it as a ‘success’. The study compared three conditions: a persuasive message based on the TPB that was designed to modify participants’ beliefs about the consequences of performing testicular self-examination (TSE); a persuasive message that was not based on the TPB; and a no-message control condition. There were significant differences in self-reported behaviour at one week and four weeks after the intervention. The mediation analysis (which combined the two message conditions) suggested that the effect of exposure to a message on intention and behaviour was partly mediated by the TPB variables, though the most important mediator was actually ‘TSE knowledge’, which is not a component of the theory. Conclusions The TPB provides fairly consistent prediction of intentions and, to a lesser extent, behaviour, across a range of different behaviours including health-related behaviours. However, the vast majority of studies have used observational designs; the causal predictions of the theory should be tested in randomized experiments (Sutton, 2002a). The theory has not so far fulfilled its promise as the basis for developing effective health behaviour interventions. Acknowledgement This chapter is based partly on Sutton (2004). REFERENCES Ajzen, I. (1988). Attitudes, personality, and behavior. Buckingham, UK: Open University Press. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211. Ajzen, I. (2002a). Constructing a TpB questionnaire: conceptual and methodological considerations. Retrieved September 1, 2004, from http://www.people.umass.edu/aizen. Ajzen, I. (2002b). The theory of planned behavior. Retrieved September 1, 2004, from http://www.people.umass.edu/aizen. Ajzen, I. & Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and review of empirical research. Psychological Bulletin, 84, 888-918. Ajzen, I. & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall. Ajzen, I. & Fishbein, M. (2000). Attitudes and the attitude-behavior relation: Reasoned and automatic processes. European Review of Social Psychology, 11, 1-33. Albarracín, D., Johnson, B. T., Fishbein, M. & Muellerleile, P. A. (2001). Theories of reasoned action and planned behavior as models of condom use: a meta-analysis. Psychological Bulletin, 127, 142-61. Armitage, C. J. & Conner, M. (2001). Efficacy of the theory of planned behaviour: a meta-analytic review. British Journal of Social Psychology, 40, 471-99. Bandura, A. (1986). Social foundations of thought and action: a social cognitive theory. New York: Prentice-Hall. Brubaker, R. G. & Fowler, C. (1990). Encouraging college males to perform testicular self-examination: evaluation of a persuasive message based on the revised theory of reasoned action. Journal of Applied Social Psychology, 17, 1411-22. Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-9. Conner, M. & Armitage, C. J. (1998). Extending the theory of planned behavior: a review and avenues for further research. Journal of Applied Social Psychology, 28, 1429-64. Conner, M. & Sparks, P. (2005). The theory of planned behaviour and health behaviour. In Conner, M. & Norman, P. (Eds.). Predicting Health Behaviour: Research and Practice with Social Cognition Models (2nd edn.) (pp. 170-222). Buckingham, UK: Open University Press. Fishbein, M. (2000). The role of theory in HIV prevention. AIDS Care, 12, 273-8. Fishbein, M. & Ajzen, I. (1975). Belief, attitude, intention, and behavior: an introduction to theory and research. Reading, MA: Addison-Wesley. Fishbein, M. & Ajzen, I. (2005). Theory-based behavior change interventions: comments on Hobbis and Sutton. Journal of Health Psychology, 10, 27-31. Fishbein, M., Hennessy, M., Kamb, M., Bolan, G. A., Hoxworth, T., Iatesta, M., Rhodes, F., Zenilman, J. M. & Project RESPECT Study Group (2001). Using intervention theory to model factors influencing behavior change: Project RESPECT. Evaluation and the Health Professions, 24, 363-84. Godin, G. & Kok, G. (1996). The theory of planned behavior: a review of its applications to health-related behaviors. American Journal of Health Promotion, 11, 87-98. Hagger, M. S., Chatzisarantis, N. L. D. & Biddle, S. J. H. (2002). A meta-analytic review of the theories of reasoned action and planned behavior in physical activity: predictive validity and the contribution of additional variables. Journal of Sport and Exercise Psychology, 24, 3-32. Hardeman, W., Johnston, M., Johnston, D. W. et al. (2002). Application of the theory of planned behaviour in behaviour change interventions: a systematic review. Psychology and Health, 17, 123-58. Hobbis, I. C. A. & Sutton, S. (2005). Response to invited commentaries: the opportunity for integration remains. Journal of Health Psychology, 10, 37-43. Ogden, J. (2003). Some problems with social cognition models: a pragmatic and conceptual analysis. Health Psychology, 22, 424-8. Sheeran, P. & Taylor, S. (1999). Predicting intentions to use condoms: a meta-analysis and comparison of the theories of reasoned action and planned behavior. Journal of Applied Social Psychology, 29, 1624-75. Stroebe, W. (2000). Social Psychology and Health (2nd edn.). Buckingham, UK: Open University Press. Sutton, S. (1998). Predicting and explaining intentions and behaviour: how well are we doing? Journal of Applied Social Psychology, 28, 1317-38. . Sutton, S. (2002a). Testing attitude-behaviour theories using non-experimental data: an examination of some hidden assumptions. European Review of Social Psychology, 13, 293-323. Sutton, S. (2002b). Using social cognition models to develop health behaviour interventions: problems and assumptions. In Rutter, D. & Quine, L. (Eds.). Changing health behaviour: intervention and research with social cognition models (pp. 193-208). Buckingham, UK: Open University Press. Sutton, S. (2004). Determinants of health-related behaviours: Theoretical and methodological issues. In Sutton, S., Baum, A. & Johnston, M. (Eds.). The Sage handbook of health psychology (pp. 94-126). London: Sage. Sutton, S., French, D. P., Hennings, S. J. et al. (2003). Eliciting salient beliefs in research on the theory of planned behaviour: the effect of question wording. Current Psychology, 22, 234-51. Trafimow, D., Sheeran, P., Conner, M. & Finlay, K. A. (2002). Evidence that perceived behavioural control is a multidimensional construct: Perceived control and perceived difficulty. British Journal of Social Psychology, 41, 101-21. Stephen SuttonUniversity of Cambridge © Cambridge University Press 2007 APA citation: Sutton, S. (2007). Theory of planned behaviour. In S. Ayers, A. Baum, C. McManus, & et. al. (Eds.), Cambridge handbook of psychology, health and medicine (2nd ed.). Cambridge University Press. Credo Reference: https://ezproxy.gardner-webb.edu/login?url=https://search.credoreference.com/content/entry/cupphm/theory_of_planned_behaviour/0?institutionId=5562 Need a different citation style? Find it on Credo Online Reference Service