Understanding and Changing Health Behaviour

From Health Beliefs to Self-Regulation

Charles Abraham,Paul Norman,Mark Conner

• Publisher: Psychology Press
• Year: 2013
• ISBN:
  9781134434091
• Edition: 1
• Topic: Medicina
• Subtopic: Teoría, práctica y referencia médicas

The identification of the factors predicting health behaviour has become a major focus of research in the field of health psychology and related disciplines. This awareness not only increases our understanding but also provides important targets for interventions to change health behaviour. Understanding and Changing Health Behaviour focuses on a range of key social cognitive factors in this process, using examples from an impressive breadth of applied settings that include smoking cessation, condom use and breast examination. The book features contributions from some of the best known researchers in the field.

Section 1 –
Introduction

Chapter One
Understanding and Changing Health Behaviour: From Health Beliefs to Self-Regulation

Charles ABRAHAM & Paschal SHEERAN

A series of ‘social cognition’ models which specify modifiable cognitive antecedents of action have been proposed (Conner & Norman, 1996; Eagly & Chaiken, 1993). In this introductory chapter we review the development of these models and identify core constructs representing correspondences between them. We also discuss recent theorising which, looking beyond motivation, has focused on action control and self-regulation processes. In doing so we map progress from early attempts to model health-related beliefs (Hochbaum, 1958) to recent explorations of the cognitive underpinnings goal-related action initiation and maintenance (Gollwitzer & Bargh, 1996).

Social cognition models have been tested by using differences in self-reported cognitions to predict health-related behaviour. For example, differences in reported health beliefs have been used to predict reported preventive action such as condom use (Abraham, Sheeran, Spears & Abrams, 1992) and recorded attendance at screening appointments (King, 1982). Reviews suggest that self-report measures based on these models do reliably distinguish between those who do and do not undertake a range of health behaviours (e.g., Bandura, 1992; Godin & Kok, 1996; Sheppard, Hartwick & Warshaw, 1988). On this basis it has been argued that interventions targeting cognitions specified by these models could effectively promote health-enhancing behaviour and/or improve the outcomes of healthcare services. In general the evidence has been supportive. Since the earliest tests of the health belief model, interventions designed to change theory-specified cognition have been shown to promote health-related behaviour (Haefner & Kirscht, 1970). Moreover, in some areas, interventions based on social cognition models have been shown to be more effective than interventions without such theoretical foundations. For example, the rapid development of interventions designed to promote HIV-preventive behaviour resulted in few theoretically-based programmes or campaigns and almost none of these were found to influence sexual behaviour (Fisher & Fisher, 1992; Oakley et al., 1995). By contrast, more recent HIV-preventive interventions based on social cognition models have proved to be effective in controlled trials (Bryan, Aiken & West, 1996; Fisher et al., 1996; Kalichman, Carey & Johnson, 1996; Schaalma et al., 1996).

Thus these models appear to offer a theoretical, evidence-based foundation for health promotion activities. Moreover, the relationship between social cognition theorists and health promoters may be symbiotic because evaluative studies of theory-based interventions can provide validity tests of the models themselves. There are also promising indications that recent theorising concerning control processes, such as action initiation and maintenance, may facilitate the development of even more effective behaviour-change interventions (see e.g., Chapters 11-13 by Gollwitzer & Oettingen, Bagozzi & Edwards, and Bandura, respectively).

Development of a Model of Reasoned Motivation

The availability of many overlapping social cognition models may itself discourage consistent application to intervention design. Bandura (Chapter 13 & Bandura, 1998) is critical of the proliferation of such models and, in our view, the identification of a summary model which is both ‘content-free’ and ‘parsimonious’ (Ajzen, 1998) would be useful to those involved in intervention design. Such a model would highlight common theoretical understandings of the cognitive antecedents of motivation but could be extended to incorporate behaviour-specific antecedents in a problem-based manner (see Ajzen, 1998; Kok & Schaalma, 1998). Four key advances in this area provide a basis for such a summary model.

First, the development of the health belief model (HBM) (Rosenstock, 1974). The HBM specifies a series of subjectively rational beliefs that could account for individual differences in motivation and action. The model highlights threat perceptions as a central component of motivation and conceptualises such appraisals in terms of beliefs about the extent of perceived susceptibility to and severity of a health problem (e.g., ‘It is unlikely that I will contract lung cancer’ and ‘I would die soon after contracting lung cancer’, respectively). Susceptibility and severity beliefs are outcome expectancies, that is, beliefs about what will happen if the person does or does not perform a particular action or sequence of actions. Both have been shown to correlate with measures of health-related behaviour. However - and perhaps surprisingly - such correlations tend to be small (Janz & Becker, 1984; Sheeran & Abraham, 1996). Harrison, Mullen and Green (1992), for example, found that these measures accounted for 1–2% of the variance in behaviour across studies.

A number of explanations for these weak relationships have been considered. Perceived severity may correlate poorly with behaviour (Maddux & Rogers, 1983; Schwarzer, 1992; Schwarzer & Fuchs, 1996; Wurtele & Maddux, 1987) because perceptions of severity only influence motivation when severity exceeds a certain threshold, and, once this threshold is reached, perceived susceptibility may be a more important component (Sheeran & Abraham, 1996; Weinstein, 1988). There has also been some debate about the meaning of correlations between perceived susceptibility to a health hazard and preventive action and about the most appropriate measures of susceptibility beliefs (see e.g., Gerrard, Gibbons & Bushman, 1996; Weinstein, 1988; Weinstein & Nicolich, 1993). Two points are worth highlighting here. First, there is a difference between simple awareness of a risk (‘Have you heard of X?’) and estimates of the magnitude of the risk to oneself (‘How likely is X to happen to you over the next five years?’) (Weinstein, 1988; Weinstein & Sandman, 1992). Secondly, when previous behaviour is not controlled for, correlations may reflect the impact of behaviour on perceived susceptibility, rather than vice versa, so that those who take more precautions perceive less risk (Weinstein & Nicolich, 1993). This latter problem may be addressed by controlling for reported health behaviours and measuring anticipated personal susceptibility in the absence of taking precautions (Sheeran & Abraham, 1996; Van der Pligt, 1998). For example, Van der Velde, Hooykaas and Van der Pligt (1996) found that ‘conditional’ measures of perceived susceptibility (which specified taking no precautions) were more likely to be related to intention than ‘unconditional’ assessments. Overall then, perceived severity may be less important than perceived susceptibility and the latter may be less central to health-related motivation than is suggested by the HBM. Nevertheless, there is little doubt that ensuring people are aware of a health threat and persuading them that they are susceptible to it unless they act (i.e., reducing defensive optimism, Schwarzer, 1998) is likely to be prerequisite to the promotion of health-related action (Weinstein, 1988; Wurtele, 1988).

Perceived barriers to action and perceived effectiveness of health-related actions (i.e., ‘response efficacy’) are also included in the health belief model. The former may be a component of self-efficacy (see below) (e.g., ‘It is difficult for me to get through the day without a cigarette.’) and both refer to relationships between outcome expectancies and individual goals (e.g., ‘I will not be able to concentrate at work without cigarettes.’ and ‘Cutting down on cigarettes reduces the chances of contracting lung cancer.’). However, the specification of ‘barriers’ and ‘response efficacy’ may underestimate the importance of other outcome expectancies. For example, the perceived likelihood of negatively evaluated emotional or social consequences of health-protective actions may be more important than health-protective outcomes (e.g., Abraham, et al., 1992; Richard, van der Pligt & de Vries 1995, see below).

A second crucial advance was made by Fishbein and Ajzen who demonstrated that behavioural prediction depended upon the use of cognition measures which describe the action/s concerned at the same level of specificity as the behaviour measure (Ajzen & Fishbein, 1977; Fishbein & Ajzen, 1975; Ajzen, 1988). For example, when identifying the cognitive antecedents of a particular action, such as destroying all your remaining cigarettes when you get home, statements used to assess individual beliefs, or cognitions should refer to this particular action towards this target in this context at this time. Similarly, when measuring cognitions relevant to the achievement of a behavioural goal which involves a sequences of actions over time, such as not smoking over the next week, cognition measures should specify that particular goal e.g., not smoking over the next week.

A third important advance was made when Fishbein and Ajzen proposed that intention formation provides a mechanism by which action-relevant beliefs and outcome expectancies affect behaviour (e.g., Fishbein & Ajzen, 1975). The theory of reasoned action (TRA) specifies attitudes and subjective norms (see below) as antecedents of intention formation and has been extensively tested over a twenty year period (e.g., Fishbein & Ajzen, 1975; Van den Putte, 1991). For example, the TRA has been employed to identify beliefs correlated with safer sex intentions which, in turn, have been shown to predict reports of subsequent safer sexual behaviour amongst a variety of samples (Fisher, Fisher & Rye, 1995; Sheeran, Abraham & Orbell, 1999).

There is considerable variability in the strength of the intention-behaviour relationship across different health behaviours but reported intentions are reliably and moderately correlated with a range of health actions (Armitage & Conner, in press; Godin & Kok, 1996; Randall & Wolff, 1994; Sheppard, Hartwick & Warshaw, 1988). For example, in a review of studies of health behaviours, Godin and Kok (1996) report intention-behaviour correlations of 0.35 across six studies of screening attendance, 0.52 across eight applications to exercise behaviour, 0.56 across five applications to addictive behaviour and an overall correlation of .46 (across twenty six applications). Comparisons of intention-behaviour correlations using self-report and observational behavioural measures suggest that correlations are somewhat lower when observational behavioural measures are employed (Armitage & Conner, in press). However, across studies, we can expect intention measures to account for 20%-25% of the variance in health behaviour measures. This indicates that other cognitive antecedents of health behaviour need to be considered but, nonetheless, establishes reported intention strength as a key indicator of cognitive preparedness for action. This is underlined by Sutton (1998) who identifies nine reasons, including inherent methodological limitations, why better behavioural prediction has not been achieved. In addition Sutton points out that percentage of variance explained is a ‘pessimistic’ effect size measure. He shows, for example, that, even when only 16% of the variance in behaviour is explained, this can correspond to a 40% success rate difference between an intervention and control group and an odds ratio of 5.4.

A fourth key advance is described by Ajzen (1998) as follows: ‘If one had to point to one profound insight produced by work on self-regulation, it is probably the tremendous importance of self-efficacy beliefs or perceived behavioural control’ (p. 738). The addition of perceived behavioural control to the theory of reasoned action, created the theory of planned behaviour (Ajzen, 1991; Ajzen & Madden, 1986).

There is some debate about of the definition of perceived behavioural control and its relationship to self-efficacy beliefs (see e.g., Sparks, Guthrie & Shepherd, 1997; Terry & O’Leary, 1995). Self-efficacy has been typically defined in terms of perceived personal competence or confidence (e.g., ‘I believe I can do X successfully.’) while perceived behavioural control also includes measures of perceived barriers and difficulties (e.g., ‘Doing X would be difficult.’). Conner and Sparks (1996) note that some studies have found low reliabilities for multi-item scales designed to incorporate both these aspects of perceived control and Sparks et al., (1997) present evidence from two studies showing that perceived control and perceived difficulty have different relationships to intentions to change eating behaviours. By contrast, it has been argued (e.g., Schwarzer, 1992) that self-efficacy is related to a variety of perceptions of the self, social context and task demands and that multi-item measures which assess personal confidence in relation to perceived barriers can take account of these various components (e.g., ‘I believe I can successfully do X even when difficulty Y is present.’) This alternative view suggests that self-efficacy and perceived behavioural control can be regarded as synonyms (see Ajzen, 1998; Bandura, 1998) and, in the interests of conceptual simplification, we shall use the term ‘self-efficacy’ to mean an overall sense of control taking account of both personal resources and perceived barriers (in the HBM sense).

Bandura and others have demonstrated that self-efficacy to successfully perform an action is predictive of actual success (e.g., Bandura, 1992; Bandura, 1997; Schwarzer & Fuchs, 1996). Those who believe they will succeed are, in general, more likely to: formulate intentions to act (e.g., De Vries & Backbier, 1994), set themselves higher goals, exert greater effort, regard errors as learning experiences and persevere for longer. They are also less likely to be distracted by anxiety and self-doubt during performance (Bandura, 1992). The TPB suggests that self-efficacy promotes action primarily through bolstering strength of intention and this view is supported by evidence showing that that, while self-efficacy is correlated with behaviour, it does not necessarily add to the prediction of behaviour achieved by intention measures (Godin & Kok, 1996). This is illustrated in Chapter 2 by Morrison, Baker and Gillmore (see also Morrison, Baker & Gillmore, 1998). In a longitudinal study of study of high-risk heterosexual teenagers Morrison and colleagues found that self-efficacy did not add to the variance explained in condom use once strength of intention had been taken into account. Note, however, that in their study of safety helmet use amongst schoolboy cyclists, Quine, Rutter and Arnold report (in Chapter 4) that self-efficacy did add to the model’s capacity to predict helmet use, although this ceased to be the case once previous behaviour was included (see als...