Athletic Insight - The Online Journal of Sport Psychology

Examining the Revised Theory of Planned
Behavior for Predicting Exercise Adherence:
A Preliminary Prospective Study

Andrew R. Levy
University of Leeds

Remco C.J. Polman & David C. Marchant
University of Hull







magazine banner


This study examined the utility of Maddux’s (1993) revised theory of planned behavior toward the prediction of exercise intentions and adherence. A prospective design was employed whereby 120 private sector health club members completed self report measures pertaining to various components of the revised theory. Adherence was measured prospectively over a sixteen week period by monitoring attendance toward prescribed exercise programs. Path analysis was used to analyze the predictions of the revised theory. Goodness of fit indices suggested an acceptable fit with the data (RMSEA <0.07; CFI > 0.94; SRMSR <0.08). However, self-efficacy was the only theoretical construct to predict intention, with the latter being the only determinant of exercise adherence. Contrary to the revised theory hypotheses, the remaining contribution of the social cognitive variables in predicting exercise intentions and adherence were minimal. The results of the present investigation provide equivocal support for the revised theory; however future research may wish to consider several of the methodological issues discussed.


       Low levels of physical activity have become a major public health problem in most western societies. For example, physical activity levels are so low that about two thirds of men and three-quarters of women currently report less than the recommended 30 minutes of moderate intensity activity a day at least five days a week (Department of Health, Physical Activity, Health Improvement and Prevention, 2004). Further to this, Dishman and Buckworth (1997) acknowledge that once sedentary individuals become active, the mean dropout rate from supervised exercise programs reported has remained at 50% during the first six months over the past 20 years. This has led researchers to examine a variety of theoretical models in order to understand the complex phenomenon of the adoption and maintenance of physical activity. However, despite the proliferation of empirical studies investigating exercise adherence, Morgan and Dishman (2001) recognized that adherence to exercise programs has not improved significantly over the last three decades. Further to this, Brawley (2002) questions our knowledge concerning exercise behavior, suggesting we know less than the volume of research portrays.

Theoretical Underpinning

       According to Biddle and Nigg (2000) an important starting point for the understanding and promotion of health related exercise and physical activity is the study of theory. Popular theoretical approaches that have provided a structured conceptual approach to understanding and predicting exercise adherence have largely been social cognitive models. These include theories of reasoned action/ planned behavior, health belief model, protection motivation theory and self-efficacy theory (for a more comprehensive review of these theories, readers are advised to consult Culos-Reed, Gyurcsik & Brawley, 2001). However, given the complexity and the vast array of determinants underlying exercise adherence it is unlikely that the adoption of one theoretical position would be able to explain complete variance. A possible way forward, therefore, may be to consider the integration of social-cognitive approaches. Indeed, Rodgers and Brawley (1993) and Weinstein (1993) echoed this over a decade ago. They suggested that the challenge for investigators was to consider ways of examining the joint contribution of social-cognitive theories for the prediction of a variety of health behaviors, in that these models are more similar to each other than different. More recently, Armitage and Conner (2000) contend that the degree of conceptual overlap between psycho-social approaches implies that such models can be usefully combined.

       An approach towards the joint contribution of various social cognitive theories was proposed by Maddux (1993), which he later termed the revised theory of planned behavior (RTPB; see figure 1). This framework combined theories of reasoned action and planned behavior (TRA; Fishbein & Ajzen, 1975; TPB; Ajzen, 1991), protection motivation theory (PMT; Maddux & Rodgers, 1983) and the health belief model (HBM; Rosenstock, 1974). According to Maddux (1993) such models are comprised of largely the same psycho-social variables, namely self-efficacy expectancy, outcome expectancy, outcome value and intention. These variables are underpinned by the value-expectancy approach which stipulates that behavior can be predicted based on the value that individuals place on outcomes and their expectations that given behaviors will lead to these outcomes. Maddux’s framework uses the TPB as a vehicle to integrate such factors. Maddux’s (1993) revised theory postulates that self-efficacy, cues to action and intention to engage in exercise are proximal antecedents of adherence. Intentions are conceptualized as motivational factors that influence behavior and they are the most immediate and powerful determinant of behavior. As such, intentions are hypothesized to be influenced by self-efficacy of the new behavior, attitude towards new behavior, attitude toward current behavior, and perceived social norms.

Figure 1 Exercise Planned

Distinction between RTPB and TPB

       Despite the apparent similarity of the revised theory compared to the TPB, three distinct differences are evident. First, with regard to attitudes, Maddux argued for the independent assessment of this construct which concerned not only attitude toward new (healthier) behavior, as proposed in the TPB, but also current (unhealthy) behavior. According to Maddux (1993) this distinction is necessary to facilitate behavior change as this would encourage an individual to engage in a comparative analysis of the benefits and costs of the current and new behavior. Specifically, attitude toward current behavior comprises of perceived vulnerability and perceived severity. Maddux (1993) regarded the former to be concerned with the degree to which individuals perceive risks if they continue their unhealthy (current) behavior and is aligned with outcome expectancy as this relates to beliefs regarding negative health consequences that may result from certain actions. The latter concerns the perceived degree of harm, discomfort or damage that will result from a health hazard and is aligned with outcome value which concerns a person’s evaluation of negative health consequences. Both of these constructs have been supported within the HBM and PMT regarding health related behavior, including exercise (Stanley & Maddux, 1986; Wurtele & Maddux, 1987).

       Second, the revised theory proposes that self-efficacy should replace perceived behavioral control, recognized by Ajzen’s TPB. The rationale for this provided by Maddux (1993) suggested that perceived behavioral control incorporates two separate independent constructs; self-efficacy expectancy and outcome expectancy. Thus, to include the latter in a measure of perceived behavioral control would be redundant as this is already incorporated within the revised theory in the assessment of attitudes. Further, to avoid confusion and measurement ambiguity, Maddux (1993) recommends that perceived behavioral control and self-efficacy expectancy should be given separate consideration. Indeed, Terry and O’Leary (1995) found support for this distinction suggesting that perceived behavioral control largely reflects external factors (or outcome expectancies) that interfere with the performance of a behavior, while self-efficacy expectations concern internal control factors. However, contrary to this Ajzen (2002a) suggested these constructs to be synonymous and proposed a two level hierarchical model in which self-efficacy and controllability together comprise a higher order concept, that being perceived behavioral control. As such this approach accounts for the results of previous studies that have shown the distinction between perceived behavioral control and self-efficacy, while ensuring the unitary nature of the perceived behavioral control construct within the TPB (Norman & Hoyle, 2004). In light of this debate Dawson, Gyurcsik, Culos-Reed and Brawley (2001) recommend that until the perceived behavioral control concept is assessed in a consistent manner, the use of self-efficacy measures maybe considered “best-practice” within exercise settings.

       A final distinction concerns the inclusion of past behavior within Maddux’s revised theory. A common criticism of the TPB is its failure to fully mediate the influence of past behavior, particularly given a meta-analysis conducted by Conner and Armitage (1998) revealed that past behavior accounted for an additional 13% of variance in behavior. However, few models in the exercise domain have attempted to incorporate habit as a predictor variable (Norman, Conner & Bell, 2000). The inclusion of past behavior in the revised theory is based on the theory of habit development (Ronis, Yates & Kirscht, 1989). This theory distinguishes between two types of situational cues, those being cues to decision (initiation phase) and cues to action (habit phase). Cues to decision refer to cognitions that may lead to behavioral intentions, but do not automatically prompt the new behavior itself, because social cognitive variables are postulated to be involved in the decision making of formulating intentions. However, if the behavior and decision making process are repeated frequently in the presence of the same cues, then cues to decision may become cues to action in which behavior is elicited by an automatic response superseding rational decision making. It must be recognized however that Ajzen (2002b) refutes the latter, suggesting that some behaviors (such as exercise) always require conscious control even after they have been performed many times and thereby an automatic (habitual) response can not be produced to explain behavior. Although more recent evidence documented by Jackson, Smith and Conner (2003) found habit (frequency of past behavior) to actually be predictive of exercise adherence. Maddux (1993) considers the shift from decision cues to action cues (habit) within the revised theory to correspond with Prochaska and DiClemente’s (1983) multi-stage theory, in which adherence to exercise becomes more habitual. Most notably this theoretical perspective goes against the social cognitive assumption that people make conscious decisions to exercise, rather frequently performed behaviors may become habits and may not require a decision to act.

Rationale and Purpose of Study

       To date, most research operationalizing this revised theory has been directed towards training adherence among elite athletes. A study investigating training adherence among elite netball players, found that the revised theory accounted for 77% of the variance in training adherence (Palmer, 2000). Notably, only habit contributed significantly towards this prediction. In a more recent study, investigating endurance-training adherence among elite junior netball athletes, Palmer, Burtwitz, Dyer and Spray (2005) found little support for the revised theory when compared with the TPB. In conclusion, Palmer et al. suggested it would be unwise to dismiss the utility of the revised theory based on the findings of one study. A possible reason for Palmer and colleagues’ equivocal findings may be due to the fact that elite athletes who make their living from sport are generally very motivated with regard to complying with training regimes. However, given Dishman and Buckworth’s (1997) recognition that 50% of exercisers drop out of supervised exercise programs in the first six months, adherence issues maybe more pertinent among non-elite populations. As such, applying Maddux’s revised theory among relatively sedentary individuals may be useful in elucidating the complex phenomenon of exercise adherence. Despite the burgeoning amount of research adopting social cognitive models to explore exercise behavior, no studies to date have explored a unified parsimonious framework which captures conceptual similarities across various social cognitive models. Given the absence of such research and the call of previous recommendations to explore conceptual convergence (Rodgers & Brawley, 1993; Armitage & Conner, 2000), the aim of this investigation was to assess Maddux’s (1993) integrated model for predicting exercise intentions and adherence using a prospective design. Specifically, it was hypothesized that intention, self-efficacy and habit (or cues to action) would significantly predict exercise adherence and that exercise intentions would be predicted by self-efficacy, attitude towards new behavior, attitude towards current behavior and perceived social norms.



       The present study consisted of 120 participants (67 men, 53 women) aged between 18 and 50 years (30.5 ± 11.2 years) and were recruited from five private health clubs. In accordance with ethical guidelines, ethical approval was obtained by The University of Hull Research Ethics Committee, alongside informed consent, which was provided by all individuals.

Revised Theory of Planned Behavior Measures

       Constructs within Maddux’s revised theory were operationalized according to guidelines established by Maddux (1993). Because these constructs are similar to those postulated by the TPB, comparable procedures for questionnaire development recommended by Ajzen (2002c) were followed. This included obtaining correspondence based on target, action, context and time elements between exercise adherence (the behavior of interest) and the predictor variables postulated by the revised theory (see appendix 1 for examples). As such, all questions referred to the performance of exercise at a health club over a period of three weeks over the next four months. Item constructs were elicited from an extensive examination by Downs and Hausenblas (2005a) who examined salient exercise beliefs relating to the TPB. Sample characteristics of this elicitation study included male and female participants (80.4%) which consisted of community adults (26.1%), students (23.9%), worksite employees (15.2%) symptomatic patients (13%) and older adults (10.9%). Such a representative sample provides adequate basis for the use of an elicitation study in the present investigation.

       Intention. Intention was assessed by a single item measure: ‘I intend to exercise three times per week over the next four months.’ Responses were rated on a 7-point unipolar semantic differential scale anchored by the word pair extremely unlikely (1) and extremely likely (7). Previous research has shown the use of a single item measure in assessing this construct to be common and valid within the exercise domain (Rhodes & Courneya, 2005; Chatzisarantis, Hagger, Biddle, & Smith, 2005).

       Attitude Toward New and Current Behavior. The measurement of attitude toward new and current behavior included eleven and eight items, respectively, that consisted of belief and corresponding value statements. Belief-based measures for both scales were rated on a 7-point unipolar scale anchored by 1 (extremely disagree) and 7 (extremely agree). Value-based items for both attitude measures were rated on a 7 point bipolar scale with responses ranging from 1 (bad) to 7 (good). The attitude items for both scales were formed as a product of the belief and corresponding value item scores (Ajzen, 1991). The alpha coefficients for attitude toward new behavior concerning expectancy and value scales were 0.72 and 0.70, respectively. Attitude toward current behavior coefficients for both expectancy and value scales were 0.84 and 0.72, respectively.

       Perceived Social Norms. Perceived social norms were defined and measured as the subjective norm construct of the TPB (Ajzen, 1991). The perceived social norm measure included six items that consisted of normative beliefs and corresponding motivation to comply beliefs. The latter were rated on a 7-point unipolar scale anchored by 1 (extremely unnecessary) and 7 (extremely necessary). Motivation to comply statements were rated on a 7-point bipolar scale with responses ranging from 1 (not at all) to 7 (very much). The perceived social norm items were formed as the product of the normative belief and motivation to comply item scores (Ajzen, 1991). The alpha coefficients for the normative belief and motivation to comply scale were 0.78 and 0.70, respectively.

       Self-efficacy. In aligning with Ajzen (2002c), Bandura (1986) also recommended the application of target, action, context and time elements toward the assessment of self-efficacy. To assess self-efficacy expectancy, individuals were asked to rate their ability to overcome identified potential barriers, which they might encounter when performing an exercise regime. Responses were rated on a 7-point unipolar scale ranging from 1 (extremely not confident) to 7 (extremely confident). Responses were averaged over items. The use of barrier efficacy as a measure of self-efficacy is in accordance with previous studies within the exercise domain (DuCharme & Brawley, 1995). The alpha coefficient of this scale was 0.96.

       Past Exercise Behavior. This measure was assessed by asking participants to indicate on an average week how often had they participated in gym based exercise over the past two months. Responses were categorized from 0-1 times per week, 2-3 times per week, 4-5 times per week, more than 6 times per week. According to Ouellette and Wood (1998) frequency of past behavior has been traditionally used as an operationalization of habit. Indeed, previous studies within an exercise capacity have utilized frequency as a measure of past behavior (Norman, Conner & Bell, 2000; Rivis & Sheeran, 2003).

       Exercise Adherence. Exercise adherence was obtained by monitoring individuals’ attendance (frequency) toward their prescribed exercise program. A total attendance score was obtained, derived from computer records of attendance, in which participants had to scan membership cards in order to gain entry to the health club. Satisfactory levels of adherence were deemed to be between 48 and 80 sessions. This was based on the American College of Sport Medicine’s (2000) guidelines, whereby participants were encouraged to exercise between three to five times per week. In order to ensure sufficient health and fitness gains, exercise programs covered a period of sixteen weeks.


       All participants who took part in the study had previous experience of using gyms and had recently obtained their own membership for one year at a health club. Before attending their first exercise session, participants underwent a consultation with a qualified fitness instructor in order to establish an appropriate exercise program that focused upon improving cardio-respiratory fitness. At this stage, participants were asked to complete a questionnaire containing the study measures. Attendance was then monitored via computer records prospectively, over a sixteen-week period.

Data Analyses

       Means, standard deviations, and Pearson product moment correlations were conducted to examine the relationships among the central variables of the revised theory. In addition, data were screened for normality using skewness and kurtosis values. The hypotheses proposed by Maddux’s revised theory were examined using path analysis using AMOS 4.0 (Arbuckle & Wothke, 1999). Accordingly, results were presented using standardized path coefficients and also squared multiple correlations (R2) to determine the amount of explained variance among endogenous variables. The adequacy of the model was assessed by examining the χ2 goodness of fit test which involved comparing the fit between the sample covariance matrix and the estimated population covariance matrix. This included both absolute and incremental fit indices as recommended by Hu and Bentler (1999). The former assesses how well a priori model reproduces the sample data, while the latter indicates a measure of the proportionate improvement of the overall fit of the target model compared with a null model. The present study employed the root mean square error of approximation (RMSEA) as a measure of absolute fit index and comparative fit index (CFI) as a measure of incremental fit, alongside standardized root mean square residual (SRMSR). According to Hu and Bentler (1999) the acceptability of the model fit using these indices of fit are <0.07, >0.94 and <0.08 respectively.


       Normality among the RTPB variables was found to be acceptable apart from habit, which was positively skewed. Thus a square root transformation was employed; however, this did not correct normality, and so habit was not transformed. The overall means, standard deviations and Pearson product moment correlations concerning constructs related to the RTPB are provided in table 1. Descriptive data revealed that participants had little prior experience of participating in gym based exercise. However, they did display moderately satisfactory levels of exercise adherence, had relatively strong intentions to engage in exercise behavior and were moderately confident they could exercise in the face of barriers. Despite this, participants indicated a poor attitude towards their new exercise regime and did generally not feel vulnerable with regards to a lack of engagement in exercise. In addition, significant others’ perceptions regarding participants exercise behavior were not deemed to be particularly important. Analysis of bivariate correlations among predictor variables of the revised theory revealed a strong relationship between intentions and exercise adherence. Self-efficacy had a moderate relationship with adherence although habit was not found to be significant. The former also had a significant relationship with intention as did attitude towards new and current behavior, however perceived social norms did not.

Table 1 Exercise Planned

       The overall model fit of Maddux’s revised theory as indicated by the path analysis, depicted in figure 2, revealed an acceptable fit χ2 = .452, p = 0.929; RMSEA= 0.068; CFI= 0.951; SRMSR = 0.088. This model identified a significant (p < 0.05) standardized effect for self-efficacy upon intention (0.50), explaining 42% of the variance. Notably, neither attitude towards new (standardized effect = .05) and current behavior (standardized effect = .03) alongside perceived social norms (standardized effect = .05) significantly predicted intention. With regard to exercise adherence only intention had a significant (p < 0.05) standardized effect (0.65), predicting 58% of the variance. Contrary to the proposed hypotheses both self-efficacy (standardized effect = .18) and habit (standardized effect = .07) had no significant influence on exercise adherence.

Figure 2 Exercise Planned


       The present study examined the utility of Maddux’s (1993) revised theory of planned behavior to predict exercise intention and adherence. Path analysis suggested an acceptable fit of data for this model, highlighting the importance of self-efficacy in predicting exercise intentions and the importance of the latter upon predicting adherence. These findings corroborate with several previous studies examining exercise adherence (Hagger, Chatzisarantis, & Biddle, 2002). However, the results found no support for self-efficacy having a direct influence on exercise adherence. A study by Palmer et al. (2005) utilizing the revised theory of planned behavior to predict endurance training adherence also found a similar result. A possible reason for this highlighted by McAuley (1992) suggests that self-efficacy cognitions are most influential in the short-term or the adoption phase of exercise rather than the long-term or maintenance phase. Contrary to this, more recent research (Wilbur, Miller, Chandler, & McDevitt, 2003) recognized the importance of self-efficacy in the prediction of long-term exercise adherence (e.g. 24 weeks plus). Notably, Palmer et al. and the present study assessed adherence over 9 weeks and 16 weeks respectively, thus longer time periods assessing adherence may have been necessary to capture a significant relationship with self-efficacy. Furthermore, items used to measure self-efficacy may differ in the formation of intention and that of sustained adherence to an exercise program. This may have attenuated the relationship between self-efficacy and adherence in the present study. Consequently, future research may need to consider the salience of self-efficacy cognitions that may influence different stages of the exercise process.

       In a summary of findings assessing numerous meta-analyses concerning the TRA and the TPB (Sutton, 1998), results suggested variance for the prediction of the intention-behavior relationship to range between 19% and 38%. Two possible causes may have contributed to the strong correlation between intentions and exercise adherence found in the present study. First, given the prospective nature of the present study and the measure of adherence being monitored over a relatively long-period, Sutton (1998) suggested longer time intervals between the measurement of intention and behavior allow more opportunities for the desired behavior to be performed, thereby increasing the intention behavior correlation. Second, it is possible that the measurement of exercise adherence used in the present study may have been confounded with health club attendance, in which participants could have been engaging in a variety of other activities within their health club that are not exercise based. Future research should consider the use of more precise objective measures of exercise adherence. Obtaining such measures would be challenging given statistical recommendations for using sophisticated methods of statistical analysis which require a large sample size.

       The weak effect of both attitudes toward new behavior and current behavior with intention, found in the present study, contradicts previous research. For instance, meta-analytic reviews of the TRA and TPB relating to physical activity highlight the importance of positive attitudes in the promotion of regular physical activity (Hagger et al., 2002; Downs & Hausenblas, 2005b). A possible reason for the present findings may be due the scoring of attitudes. Attitudes were scored using multiplicative composites which is in accordance with the expectancy-value approach (Maddux, 1993; Ajzen, 1991). However, a study by Gagne and Godin (2000), based on 16 datasets, found that in most cases the use of multiplicative composites for obtaining overall attitude scores (and also subjective norms) were less predictive than summing expectancy beliefs in isolation. Although Gagne and Godin’s study was based on a relatively small number of cases, French and Hankins (2003) suggested the implications of their findings has the potential to undermine previous studies using multiplicative composites used to determine overall scores of relevant planned behavior constructs. To counter this Ajzen (2002c) stipulate that the use of multiplicative composites to be a viable approach to assess expectancy-value variables included in planned behavior models. Given the important implications of Gagne and Godin’s findings, future research is required on a larger scale to elucidate the predictive ability of multiplicative composites regarding the proposed relationships concerning attitudes and perceived social norms put forward by Maddux’s revised theory.

       In accordance with previous research (Hausenblas, Carron, & Mack, 1997; Hagger et al., 2002; Downs & Hausenblas, 2005b) perceived social norms were not related to intention in the present study. Despite this, a meta-analysis by Carron, Hausenblas and Mack (1996) indicated social influence to have a moderate effect on exercise cognitions, including intention and behavior. More recently, Palmer et al. (2005) found perceived social norms to be predictive of training intentions, concluding that interventions concerned with perceptions of social pressure to perform a training regime would likely enhance training intentions and adoption. A possible explanation for these inconsistent findings may be due to the measurement of social variables. Indeed, measurement issues have been put forward as a common explanation for perceived social norms being a consistently weak predictor of exercise intentions (Sheeran & Orbell, 1999). Given such equivocal findings future research needs to address pertinent measurement issues and explore alternative forms of social influences upon exercise intentions and adherence.

       A key feature of the revised theory is the inclusion of habit. However, results from this study found past behavior to have no relationship with exercise adherence. Contrary to this, previous studies have provided evidence to support the predictive utility of past behavior upon exercise behavior (Hagger et al., 2002; Rhodes & Courneya, 2003; Jackson et al., 2003). A possible reason for habit being unrelated with exercise adherence in the present study may be due to the distal nature in which habit was assessed. Hence, it is possible that participants may have had trouble recalling such information due to memory decay. As an alternative, future research may want to consider using more proximal measures of habit by assessing the frequency of participation to a prescribed exercise regime. Indeed, the continuous open format of the latter would provide better scale correspondence with exercise than the self report measure of habit used in the present study which was dichotomously graded (Courneya & McAuley, 1993). It is possible that this lack of scale correspondence between habit and exercise adherence may have attenuated its relationship. Further to this, it is apparent that participants in this study had a low frequency of past exercise behavior. This positive skew and subsequent restricted range could have affected the predictive ability of habit. Notably, restrictions in range have previously been reported to attenuate the predictive capabilities of the TPB (Courneya & McAuley, 1993).

Limitations and Recommendations

       Despite the present findings, some key limitations and associated future research recommendations warrant mention. First, despite utilizing an exercise specific population, a small sample size was obtained which prevented the use of sophisticated statistical analysis such as structural equation modeling (SEM). To employ this approach future research should endeavor to obtain a large enough sample size. Although Biddle, Markland, Gilbourne, Chatzisarantis and Sparkes (2001) acknowledge it is difficult to provide a general rule regarding adequate sample size, Kline (1998) advocates a minimum sample size of 200 when utilizing SEM in order to not compromise the analysis. Second, given there appears to be no suitable gold standard for assessing exercise adherence, future research should use multiple measurements as opposed to the unitary measure adopted by the present study. This would allow for a more accurate assessment of the relationship between social-cognitive variables and exercise adherence. Such measures should include both subjective and objective measures of exercise in relation to intensity and duration alongside physiological outcomes (e.g. body mass index). Third, despite previous research acknowledging the use of single item measurement for intention, it must be noted that such measures could be affected by measurement error. Therefore, future research may benefit from employing multiple item measures of intention which are considered more reliable (Conner & Sparks, 2005). Fourth, in view of the problems associated with the use of multiplicative composites to compose attitude and PSN scores, future research may want to consider alterative scoring approaches. Although this may conceptually not entirely align with Maddux’s theory, Conner and Sparks (2005) comment no satisfactory solution has been found to alleviate problems associated with multiplicatively combined calculations and as such requires more attention by planned behavior researchers. Fifth, the present study did not achieve scale correspondence with respect to intentions and behavior. Intentions were measured with a fixed exercise frequency (i.e. more than three times per week) while behavior was measured using open frequencies. However, according to Courneya (1994) violation of scale correspondence is common in most planned behavior studies concerning health behavior, in particular with exercise due to its repeated nature. Despite this, it has been recommended that future research should achieve scale correspondence by employing open frequencies or continuous formats for both intentions and exercise behavior (Courneya & McAuley, 1993). Finally, despite previous recommendations for using elicitation studies (Downs & Hausenblas, 2005a) it is possible that this approach to determine belief-based measures of attitude and perceived social norms were not salient to the population under study. Accordingly, this may have contributed to the lack of significance these variables had upon intention. This issue becomes more apparent given the heterogeneity of the present sample in terms of age, gender and social background. In view of this, future research wishing to utilize Maddux’s revised theory may wish to consider using pilot studies to elicit belief-based components among more homogeneous samples.

Implications and Conclusion

       Implications from the present study suggest that during the initial adoption of exercise, practitioners should help individuals enhance feelings of self-efficacy to help them achieve personal control over their exercise intentions. According to Luszczynska and Schwarzer (2005) individuals who are attempting to change behavior, particularly in the decision making phase, need to have the belief that they are capable of performing their intentions. To achieve this, Bandura (1986) has outlined four main sources of self-efficacy that could be targeted by exercise practitioners to ameliorate self-efficacious beliefs among exercisers. First, establishing a sense personal mastery is deemed to be the most salient efficacy source. For example, this may be established through the use of a decisional balance sheet to identify a persons perceived barriers to exercise. This would facilitate discussion between practitioner and exerciser regarding possible solutions to overcome salient barriers, thus possibly providing an individual with a sense of mastery in being able to cope with the latter. Second, self-efficacious beliefs may be enhanced through observing others successfully perform the target behavior, particularly if a person has little prior experience of exercising. This could be achieved by training with a friend, personal trainer, or possibly arranging to attend an exercise class in which repeated observable instruction is provided. Third, verbal and social persuasion from significant others can influence perceptions of self-efficacy; however, its success is dependant upon the realistic nature of its content. As an example, Biddle and Mutrie (2001) suggest this strategy could be endorsed by ensuring frequent contact between the exerciser and fitness instructor. Finally, an individual’s physiological state may be used as a source of efficacy information. For instance, according to Norman, Boer and Seydel (2005) an individual may interpret high levels of anxiety to mean a lack of capability to perform one’s intention. As such, relevant relaxation strategies may be usefully employed.

       Given the strong relationship between intention and exercise adherence in the present study, a key implication is that exercise practitioners need to ensure that individuals actually know how to act upon their intentions in order to commence exercise behavior. A limitation of planned behavior frameworks is that they do not directly address the issue of translating intentions into action. However, according to Gollwitzer (1993), this could be addressed by helping individuals form implementation intentions, which commit individuals to a specific course of action in a particular situational context. To clarify, Sheeran, Milne, Webb & Gollwitzer (2005) comment that intentions indicate what one will do, where as an implementation intention specifies the when, where, and how of what one will do. In accordance with Sheeran et al. (2005) to form an implementation intention, practitioners need to help exercisers clearly identify a desired outcome (or goal) they want to achieve (e.g. become physically active) in addition to forming an effective goal-directed response (e.g. swimming). Secondly, it is necessary to plan how the goal-directed response can be put into action. This involves specifying a date, time, duration, intensity, and location for performing one’s desired goal. Given the volitional nature of exercise, Sheeran et al. (2005) recommend individuals should also set implementation intentions for overcoming problems of initiating and maintaining one’s goal-directed response and also overcoming contextual or situational threats (e.g. watch television). Previous exercise research has supported the view that forming implementation intentions increases the likelihood of exercise intentions being enacted (Prestwich, Lawton & Conner, 2003).

       To conclude, this preliminary study provided equivocal support for the revised theory of planned behavior. However, it must be noted that the proposed relationships for attitude towards current behavior and habit which distinguished Maddux’s revised theory from Ajzen’s original theory of planned behavior were not supported. In addition, attitude toward new behavior and perceived social norms were not related to intentions; self-efficacy was found not to be related to exercise adherence. Despite this, self-efficacy was found to influence exercise intention, while exercise adherence was deemed to be influenced by intention. Before clear conclusions regarding the efficacy of Maddux’s revised model can be drawn, it is important that future research considers pertinent methodological issues identified by the present preliminary study.


       Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Process, 50, 179-211.

       Ajzen, I. (2002a). Perceived behavioral control, self-efficacy, locus of control, and the theory of planned behavior. Journal of Applied Social Psychology, 32, 1-20.

       Ajzen, I. (2002b). Residual effects of past on later behavior: Habituation and reasoned action perspectives. Personality and Social Psychology Review, 6, 107-122.

       Ajzen I. (2002c). Constructing of a standard questionnaire for the theory of planned behavior. Retrieved March 16, 2004, from

       American College of Sports Medicine. (2000). ACSM's Guidelines for Exercise Testing and Prescription (6th ed). Philadelphia: Lippincott, Williams, & Wilkins.

       Arbuckle, J. L., & Wothke, W. (1999). AMOS 4.0 user's guide. Chicago: SmallWaters Corporation.

       Armitage, C. J., & Conner, M. (2000). Social cognition models and health behavior: A structured review. Psychology and Health, 15, 173-189.

       Bandura, A. (1986). Social Foundations of Thought and Action: A Social-Cognitive theory. Englewood Cliffs, NJ: Prentice-Hall.

       Biddle, S. J. H., Markland, D., Gilbourne, D., Chatzisarantis, N. L. D., & Sparkes, A. C. (2001). Research methods in sport and exercise psychology: Quantitative and qualitative issues. Journal of Sports Sciences, 19, 777-809.

       Biddle, S. J. H., & Mutrie, N. (2001). Psychology of physical activity: Determinants, well-being and interventions. London, UK: Routledge.

       Biddle, S. J. H., & Nigg, C. R. (2000). Theories of Exercise Behaviour. International Journal of Sport Psychology, 31, 290-304.

       Brawley, L. R. (2002). Exercise motivation and motivated behavior change: Facing the challenge of producing outcomes. Journal of Sport and Exercise Psychology, 24, 11-11.

       Carron, A. V., Hausenblas, H. A., & Mack, D. E. (1996). Social influence and exercise: A meta analysis. Journal of Sport & Exercise Psychology, 18, 1-16.

       Chatzisarantis, N. L. D., Hagger, M. S., Biddle, S. J. H., & Smith B. (2005). The stability of the attitude-intention relationship in the context of physical activity. Journal of Sports Sciences, 23, 49-61.

       Conner, M., & Sparks, P. (2005). Theory of planned behaviour and health behaviour. In M. Conner & P. Norman (Eds.), Predicting health behaviour (pp. 170-222). Berkshire, UK: Open University Press.

       Conner, M., & Armitage, C. J. (1998). Extending the theory of planned behavior: A review and avenues for future research. Journal of Applied Social Psychology, 28, 1429-1464.

       Courneya, K. S. (1994). Predicting repeated behavior from intention: The issue of scale correspondence. Journal of Applied Social Psychology, 24, 580-594.

       Courneya, K. S., & McAuley, E. (1993). Predicting physical activity from intention: Conceptual methodological issues. Journal of Sport & Exercise Psychology, 5, 50-62.

       Culos, S. N., Gyurcsik, N. C., & Brawley, L. R. (2001). Using theories of motivated behaviour to understand physical activity: Perspectives on their influence. In R. N. Singer, H. A. Hausenblas, & C. M. Janelle (Eds.), Handbook of sport psychology (pp. 695-717). New York: Wiley.

       Dawson, K. A., Gyurcsik, N. C., Culos-Reed, S. N., & Brawley, L. R. (2001). Perceived control: A construct that bridges theories of motivated behavior. In G. C. Roberts (Ed.), Advances in motivation in sport and exercise (pp. 321-356). Champaign, IL: Human Kinetics.

       Department of Health, Physical Activity, Health Improvement and Prevention. (2004). At least five a week: Evidence on the impact of physical activity and its relationship to health. London: Department of Health.

       Dishman, R. K., & Buckworth, J. (1997). Adherence to physical activity. In W. P. Morgan (Ed.), Physical activity & mental health (pp. 63-80). Washington, DC: Taylor & Francis.

       Downs, D. S., & Hausenblas, H. A. (2005a). Elicitation studies and the theory of planned behavior: A systematic review of exercise beliefs. Psychology of Sport & Exercise, 6, 1-31.

       Downs, D. S., & Hausenblas, H. A. (2005b). The theories of reasoned action and planned behavior applied to exercise: A meta-analytic update. Journal of Physical Activity and Health, 2, 76-97.

       DuCharme, K. A., & Brawley, L. R. (1995). Predicting the intentions and behavior of exercise initiates using two forms of self-efficacy. Journal of Behavioral Medicine, 18, 479-497.

       Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley.

       French, D., & Hankins, M. (2003). The expectancy-value muddle in the theory of planned behavior-and some proposed solutions. British Journal of Health Psychology, 8, 37-55.

       Gagne, C., & Godin, G. (2000). The theory of planned behavior: Some measurement issues concerning belief-based variables. Journal of Applied Social Psychology, 30, 2173-2193.

       Gollwitzer, P. M. (1993). Goal achievement: The role of intentions. In W. Strobe & M. Hewstone (Eds.), European review of social psychology, vol 4 (pp. 141-185). Chichester, UK: Wiley.

       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.

       Hausenblas, H. A., Carron, A. V., & Mack, D. E. (1997). Application of the theories of reasoned action and planned behavior to exercise behavior: A meta-analysis. Journal of Sport & Exercise Psychology, 19, 36-51.

       Hu, L., & Bentler, P. M. (1999). Cut off criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modelling, 6, 1-55.

       Jackson, C., Smith, A. R., & Conner, M. (2003). Applying an extended version of the theory of planned behavior to physical activity. Journal of Sports Sciences, 21, 119-133.

       Kline, R. B. (1998). Principles and practice of structural equation modeling. New York: Guilford Press.

       Luszczynska, A., & Schwarzer, R. (2005). Social cognitive theory. In M. Conner & P. Norman (Eds.), Predicting health behaviour (pp. 127-169). Berkshire, UK: Open University Press.

       Maddux, J. E. (1993). Social cognitive models of health and exercise behavior: An introduction and review of conceptual issues. Journal of Applied Sport Psychology, 5, 116-140.

       Maddux, J. E., & Rodgers, R.W. (1983). Protection motivation and self-efficacy: A revised theory of fear appeals and attitude change. Journal of Experimental Social Psychology, 19, 469-479.

       McAuley, E. (1992). The role of efficacy cognitions in the prediction of exercise behavior in middle-aged adults. Journal of Behavioral Medicine, 15, 65-88.

       Morgan, W. P., & Dishman, R. K. (2001). Adherence to exercise and physical activity. QUEST, 53, 277-278.

       Norman, P., Boer, H., & Seydel, E. R. (2005). Protection motivation theory. In M. Conner & P. Norman (Eds.), Predicting health behaviour (pp. 170-222). Berkshire, UK: Open University Press.

       Norman, P., Conner, M., & Bell, R. (2000). The theory of planned behavior and exercise evidence for the moderating role of past behavior. British Journal of Health Psychology, 5, 249-261.

       Norman, P., & Hoyle, S. (2004). The theory of planned behavior and breast self-examination: Distinguishing between perceived control and self-efficacy. Journal of Applied Social Psychology, 34, 694-708.

       Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life: The multiple processes by which past behavior predicts future behavior. Psychological Bulletin, 124, 54-74.

       Palmer, C. L. (2000). Fitness training adherence of elite netball players. Unpublished doctoral dissertation, Manchester Metropolitan University, UK.

       Palmer, C. L., Burtwitz, L., Dyer, A. N., & Spray, C. M. (2005). Endurance training adherence in elite junior netball athletes: A test of the theory of planned behavior and a revised theory of planned behavior. Journal of Sports Sciences, 23, 277-288.

       Prestwich, A., Lawton, R., & Conner, M. (2003). Use of implementation intentions and the decision balance sheet in promoting exercise behaviour. Psychology and Health, 18, 707-721.

       Prochaska, J. O., & DiClemente, C. C. (1983). Stages and processes of self-change of smoking: Toward an integrative model of change. Journal of Consulting and Clinical Psychology, 51, 390-395.

       Rhodes, R. E., & Courneya, K. S. (2003). Modelling the theory of planned behavior and past behavior. Psychology, Health & Medicine, 8, 57-69.

       Rhodes, R. E., & Courneya, K. S. (2005). Threshold assessment of attitude, subjective norm, and perceived behavioural control for predicting exercise intention and behavior. Psychology of Sport and Exercise, 6, 349-361.

       Rivis, A., & Sheeran, P. (2003). Social influences and the theory of planned behaviour: Evidence for a direct relationship between prototypes and young people’s exercise behaviour. Psychology and Health, 18, 567-583.

       Rodgers, R. W., & Brawley, L. R. (1993). Using both self-efficacy theory and the theory of planned behavior to discriminate adherers and dropouts from structured programs. Journal of Applied Sport Psychology, 5, 195-206.

       Ronis, D. L., Yates, J. F., & Kirscht, J. P. (1989). Attitudes, decisions, and habits as determinants of behavior. In A. R. Pratkanis, S. J. Breckler & A. G. Greenwald (Eds.), Attitude structure and function (pp. 213-239). New York: Erlbaum.

       Rosenstock, I. M. (1974). The health belief model and preventative health behavior. Health Education Monographs, 2, 354-386.

       Sheeran, P., Milne, S., Webb, T. L., & Gollwitzer, P. M. (2005). Implementation intentions and health behaviour. In M. Conner & P. Norman (Eds.), Predicting health behaviour (pp. 276-323). Berkshire, UK: Open University Press.

       Sheeran, P., & Orbell, S. (1999). Augmenting the theory of planned behavior: Roles for anticipated regret and descriptive norms. Journal of Applied Social Psychology, 29, 2107-2142.

       Stanley, M. A., & Maddux, J. E. (1986). Cognitive processes in health enhancement: Investigation of a combined protection motivation and self-efficacy model. Basic and Applied Social Psychology, 7, 101-113.

       Sutton, S. (1998). Predicting and explaining intentions and behavior: How well are we doing. Journal of Applied Social Psychology, 28, 1317-1338.

       Terry, D. J., & O'Leary, J. E. (1995). The theory of planned behavior: The effects of perceived behavioural control and self-efficacy. British Journal of Social Psychology, 34, 199-220.

       Weinstein, N. D. (1993). Testing four competing theories of health-protective behaviour. Health Psychology, 12, 324-333.

       Wilbur, J., Miller, A. M., Chandler, P., & McDevitt, J. (2003). Determinants of physical activity and adherence to a 24-week home based walking program in african american and caucasian women. Research in Nursing & Health, 26, 213-224.

       Wurtele, S. K., & Maddux, J. E. (1987). Relative contributions of protection motivation theory components in predicting exercise intentions and behavior. Health Psychology, 6, 453-466.

Appendix 1 Exercise Planned

We look forward to your comments and feedback.  Simply e-mail Athletic Insight.
Athletic Insight - The Online Journal of Sport Psychology: Line Mental Health Net Award Winner
Copyright © 2008 Athletic Insight, Inc.
ISSN 1536-0431