Motivating individuals to adopt and maintain regular physical activity (PA) is a major challenge for health professionals, and facilitating PA via public health initiatives is considered of paramount importance. To guide these initiatives and improve the success of interventions, it has been suggested that sound behavioral theory is necessary (6). Theory is considered important to understand, explain, and make predictions about a subject matter, yet it also needs to be functional, rational, and parsimonious. Indeed, Lewin (22) defended the essential advantage of "there is nothing so practical as a good theory."
The history of behavioral research in PA has demonstrated some of this evidence. In early exercise psychology research, it was common to assess a large collection of eclectic variables and to examine each, in turn, for its correlation with exercise behavior but make no attempt to understand how the various correlates may be interrelated (12). For example, correlates of exercise behavior include age, sex, education, income, ethnicity, body weight, climate, smoking status, health status, attitudes, perceived control, self-efficacy, intentions, motivation, commitment, perceived barriers, knowledge, skills, distance from a fitness facility, esthetics, having home exercise equipment, amount of park space, time spent outdoors, spousal support, number of children at home, extraversion, neuroticism, exercise intensity, exercise type, cohesion, group norms, class size, muscle fiber type, and genetic predispositions. This approach may classify groups of correlates (demographic, personal, social, environmental, etc.), but no other model reduction or ordering structure among the correlates is made.
Theoretical models, on the other hand, specify depth and breadth among correlates in a scheme that attempts to be comprehensive yet parsimonious. Most recent evaluations of PA correlates use a theoretical model or frame (40). These generally have provided a common series of terms and isolated critical constructs that may be required to target successful PA behavior change (41).
Still, behavioral interventions focused on theories have had modest success (18), and the use of these theories, when examined in terms of mediation between an intervention and subsequent behavior change, has been limited (6,24,41). There also is recent pointed commentary that theory testing in PA may not be as practical as Lewin (22) originally surmised because of limited or continued methodological weaknesses in the research (46,48). One obvious task for researchers in the PA domain is to provide validation tests of the measures and the full theories being applied. Cafeteria-style theorizing (5), the act of simply picking and choosing constructs and measures without full attempts to validate a theory, may lead to some of the problems in the current advancement of PA research (29).
As PA theory research matures, the adaptation and augmentation of these theories with PA-specific concepts seem required to improve the explanatory power and to justify the uniqueness of the discipline. First, behavioral theories used in PA research to date have been borrowed almost entirely or applied from other parent disciplines (7). Social psychology, epidemiology, and sociology, among others, are responsible for most of our current models. Whereas this is generally good scientific practice in a young scientific discipline (29) - no use reinventing the wheel - the maturation of a specific discipline requires a more focused approach. Indeed, theories developed specifically within the domain by PA experts might well provide increased explanatory capabilities and probably are needed to justify its uniqueness from the parent disciplines (31). PA sets itself apart from other behaviors in that it is an adoption behavior (vs cessation behaviors like smoking, drinking, and drug use), where the "path of least resistance" or inertia is the absence of the desired behavior; it is not a necessary behavior (vs healthy eating); it requires a significant time commitment (vs toothbrushing, flossing, and sun-protective behavior); physiological response during PA is adaptive, whereas this is a negative sign for other behaviors (stress, alcohol, and drug use); it is not a temporary one-time decision (such as cancer screening and radon testing); and it must be performed above the metabolic equivalent of rest. Thus, there is adequate, if not overwhelming, evidence to suggest that unique theories of PA should be pursued.
Second, our understanding of PA has received the most theoretical attention within the intrapersonal/psychological domain using social cognitive theories or more recently using socioecological approaches including the environment. Whereas the move to environmental correlates of PA via a socioecological frame was sparked from the myopic preoccupation with the intrapersonal level constructs in behavior change, the socioecological model challenges researchers and practitioners with its breadth, lack of mechanistic or theoretical underpinnings, and lack of focus. The challenge is now to blend these approaches to produce models of PA that synergistically combine the two types of models.
Finally, an understanding of complex behaviors like PA undoubtedly includes both malleable and systemic determinants. As a result, and guided by epidemiological practice, it is common to assess PA theories via correlational, often cross-sectional, or very short prospective designs. It has been argued that much theory research has been caught in a quagmire because of these designs (48). As interventions and behavior change are generally the sought-after goal for this research, the move to experimental or natural longitudinal analysis of change research is critical.
Thus, the present review highlights the relevant advances of three prominent social cognitive theories applied to understand PA: theory of planned behavior (TPB), self-efficacy theory (SET), and the transtheoretical model of behavior change (TTM) based on research conducted in our laboratories during the last 15 yr. These three theories represent most of the work performed, thus far, in behavioral PA research, although several others are gaining momentum (e.g., self-determination theory). Examples and commentary are provided within the context of validating the applied theory, augmenting these theories with PA-specific variables and testing these theories with analyses of change and experimental designs. We conclude by presenting a framework for researchers to test whether a particular behavioral theory holds utility in the PA domain.
DESCRIPTIONS OF THE THEORIES
Theory of Planned Behavior
As a brief overview, TPB (2) suggests that three constructs (attitude, subjective norm, and perceived behavioral control (PBC)) predict intention, which is the proximal predictor of behavior. Attitude can be defined as the overall appraisal/evaluation that an individual has concerning a behavior; subjective norm refers to the perceived social influence/pressure that other people (i.e., family, friends, physician, etc.) place on an individual to perform a given behavior; and PBC refers to an individual's perception of the ability that they have to perform a behavior (2). PBC also is thought to predict behavior directly to the extent that it acts as a proxy measure of actual behavioral control.
A shift in theorizing from a behavioral focus to a cognitive focus sparked the development of the SET by Bandura (4). The crux of this theory is that individuals must believe that they have the capability to change behavior. This situation-specific self-confidence is known as self-efficacy, and its antecedents are past performance accomplishments, vicarious experience, social persuasion, and physiological and affective states in descending importance. Outcome expectations also are central to the SET and refer to the expectations an individual has about the outcomes of a behavior (4).
Transtheoretical Model of Behavior Change
The TTM was developed originally by Prochaska and DiClemente (34) to study behavior change associated with individuals who were quitting smoking via clinical observations. This theory now has been applied widely to a number of other health behaviors, including safe sex practices, cancer screening behavior, dietary and nutrition habits, and PA and exercise behavior. The TTM is a stage-based model that incorporates a number of factors from different theoretical frameworks. The premise of the framework is that people progress through six stages (precontemplation, contemplation, preparation, action, maintenance, and termination), and that instead of being linear (movement through the stages in only a forward fashion), the movement between stages is dynamic and can be cyclical, where individuals can progress forward or backward. As well as the six stages of change, the model also is made up of 10 processes of change, decisional balance of change (i.e., the pros and cons of behavior change), and elements of self-efficacy, including confidence and the ability to overcome temptation.
VALIDATION OF THE APPLIED THEORY
Theory of Planned Behavior
The TPB has been used extensively within the PA domain (7). Our review indicated that more than 200 studies have applied the model to predict and explain PA. A number of meta-analytic reviews also have been conducted, and their outcomes have been very consistent. For example, the most recent meta-analysis demonstrated large effect sizes between intention and PA (d = 1.01), intention and PBC (d = 0.90), and intention and attitude (d = 1.07) (47).
Subjective norm, on the other hand, is typically a very weak predictor of intention after controlling for attitude and PBC. Thus, all reviews conclude positive support for all TPB constructs but subjective norm. Although this point was made even during early reviews of the model (16), no removal of subjective norm has taken place. Certainly, no researcher observing good evidence-based practice would embark on an intervention targeting subjective norm, yet contemporary research holds steadfast to the original theoretical conceptions of the 1990s. Indeed, one wonders what the purpose of all the reviews and review updates are for if the model does not advance in application across time. Clearly, subjective norm, as currently conceived, could be dropped from future use.
The SET has been presented more recently within the Social Cognitive Theory (SCT), which represents a system of reciprocal causation, such that behavior is a complex interaction of the individual, environment, and behavior. Rather than using the entire SCT in the PA and exercise domains, the SET has been the main focus and contribution from the SCT to the literature. Because of the reciprocal nature, self-efficacy can be viewed as, and has been found to be, both a determinant and a consequence of PA participation (26). As a determinant, self-efficacy has shown an overall correlation of r = 0.35 with PA behavior in a recent meta-analysis (44); this represents one of the largest overall correlates of PA.
Unfortunately, the application of SCT outside of the use of SET is scant. Perhaps most important in future research is to test the full model of SCT proposed by Bandura (5). Our review of the literature found many claims to the examination of SCT that merely apply self-efficacy, but less than a handful of studies has examined the full SCT model. This is an obvious shortcoming in present research, and it provides an important impetus for future work. The inclusion of self-regulation, for example, has been shown to be the most reliable mediator of intervention initiatives (41), and it sits prominently within the SCT as a proximal antecedent of behavior but not present within SET. At present, it stands that validation of the SCT, outside of SET, is not well understood in the PA domain.
A number of literature reviews on the TTM and PA have been performed. The most recent review article by Spencer and colleagues (45) examined 150 studies (38 intervention studies, 70 population studies, and 42 validation studies). Overall support has been found for most TTM correlates of stage membership; that is, self-efficacy, processes of change, and pros of decisional balance show larger means as the stage of membership increases, whereas temptations and cons exhibit lower means with more advanced stages.
The initial adaptation of the stages of change mistranslated the pre-action stages from the smoking domain. Precontemplation was defined as doing no PA and no intention to start, contemplation as doing no PA but intending to start PA within the next 6 months, and preparation as doing some activity but not regularly (25). This resulted in a theoretically inconsistent classification of the preparers who were irregularly active but had no intention to become regularly active (theoretically precontemplators). Furthermore, the precontemplation stage and contemplation stage did not reference the same criterion of activity as the latter stages (intending to start vs regular), presenting an incongruence in the intention (to start) and behavior (regular activity). The implications of this never have been investigated formally; however, they are hypothesized to have misclassified individuals and thus, may be in part responsible to the lack of separation of the experiential/cognitive processes with the behavioral processes and possibly decreasing effectiveness of stage-targeted and -tailored interventions.
Staging using the Marcus definitions along with staging using the congruent definitions of intention and behavior have been validated using different indicators of PA and, to a lesser extent, intention indicators (17). However, the reliance on cross-sectional studies for validation is associated with inherent limitations, and our recent 2-yr longitudinal study indicated that the post-action stages may not be that good at predicting change in regular PA (14). Another challenge presented in the TTM applied to PA is that research does not support the theoretical separation of the two higher order levels of the processes of change (the experiential and the behavioral processes) for PA in tests of the factor structure (30). Thus, for PA, a recent development has been the suggestion of two reconceptualizations of the process structure specific to PA, both of which exhibit factorial validity across gender, age, and ethnicity (33). One conceptualization is a two-factor second-order process of change model. The experiential higher order factor is represented by first-order factors of consciousness raising, dramatic relief, environmental reevaluation, and social liberation (self-reevaluation is excluded). The behavioral higher order factor is represented by first-order factors of reinforcement management and self-liberation combined, counterconditioning, helping relationships, and stimulus control. The second conceptualization is a five-factor measurement model of the processes of change. Factor 1 is represented by the constructs of self-revaluation, reinforcement management, and self-liberation. Factor 2 combines dramatic relief and environmental reevaluation. Factors 3 to 5 are counterconditioning, helping relationships, and consciousness raising (33).
AUGMENTATION OF THE APPLIED THEORY
Theory of Planned Behavior
The initial TPB conceptualization was very open to augmentation if sound empirical evidence could be demonstrated (2). Various factors that predict PA independent of the TPB have been examined by PA researchers. In our laboratories, these include habit (38), personality (36), and the perceived environment (37).
Current theorists define habit as goal-directed automaticity marked by elements of repetition and low awareness (e.g., (1)). That is, the behavior once was performed via deliberative processes and subsequent motivation (i.e., intention), but it now is performed via external cues to the behavior. Our findings (38), for example, showed that habit could account for a significant 7% of the variance in PA after controlling for intention, supporting the supposition that automatic behavioral initiation may be a component of PA behavior. Still, it also is important to note that intention was the key predictor of PA; thus, PA has a strong and dominant motivational component in terms of its antecedents. The combination of these automatic and volitional factors in behavior underscores the potential complexity of regular PA performance.
Personality has numerous definitions, but most encompass the concepts that traits are enduring individual-level differences in tendencies to show consistent patterns of thoughts, feelings, and actions (28). Models like TPB suggest that personality affects behavior through its constructs. For example, one's optimism may affect the appraisal of one's attitude about a behavior, which in turn influences intention to perform the behavior and then subsequent behavior. This mediation, however, has been supported only in 2 of 14 tests in the PA domain (42). Instead, the personality traits of extraversion (disposition for positive emotions, sociability, and assertiveness) or conscientiousness (disposition to be achievement striving and self-disciplined) have had direct effects on PA independent of TPB. Like habit, the combination of motivation and personality in behavioral performance suggests that PA may be a unique behavior with a more specific model than what TPB offers.
The large move away from individual models to socioecological concepts of PA has sparked interest in how the physical and social environment may interact with or augment constructs embedded in the TPB. Our laboratory has demonstrated that regular walking can be predicted by the proximity to retail and shopping (37) independent of intention. The results demonstrate how researchers can focus broad environmental models while augmenting individual models when understanding PA. The combination of environmental and individual factors in behavior also underscores the specificity of PA modes such as walking.
Despite these advances, the largest impact to the depth of TPB structure in PA research has come from within the measurement domains of TPB constructs themselves. In particular, attitude was conceived as having affective (pleasure, enjoyment) and instrumental (utility, benefit) qualities (2), and subsequent meta-analyses show that affective attitude is clearly the largest predictor of intention (39). Thus, the affective qualities of PA are the driving factor of PA motivation, which appears distinct from several other behaviors in health and social behavior (21).
The most interesting work in PA and SET comes from the examination of multiple efficacies and their proportionate importance to behavior. An early debate among self-efficacy scholars focused on whether self-efficacy is about confidence to perform the act itself (walking involves putting one foot in front of the other) or the confidence to regulate the action (20). In PA, this has given rise to two types of self-efficacy commonly named barrier efficacy and task efficacy (8). Barrier efficacy is the confidence to overcome and navigate the possible barriers that interfere with performing repeated bouts of PA, whereas task efficacy is the confidence to perform the specific PA act itself. In general, it seems that barrier efficacy is the critical predictor of PA participation, with the exception of clinical or physically compromised populations (8). In these cases, it seems that task self-efficacy is the best predictor. This makes theoretical sense, as confidence in performing the task itself would seem necessary before one concerns oneself with confidence to regulate one's daily barriers to act. Several other self-efficacy measures have emerged in the PA literature to define the specific barriers or behaviors that contribute to regular PA participation (proxy, scheduling, coping, action, planning, etc.). Because self-efficacy is situation and behavior specific, multiple efficacies theoretically are valid and all could aid in creating unique PA models with a SET approach.
Related to multiple self-efficacies, the meaning of the construct of self-efficacy, independent of outcome expectations and motivation, has been debated (10). Cahill et al. (10) reflected on the assertions of Bandura (4) that behavioral actions were regulated and influenced by the anticipation of negative emotional consequences of those actions (i.e., anxiety and fear). Such consequences amount to an outcome expectancy, the influence of which can be assuaged given stronger self-efficacy to perform the desired behavior. Similarly, we have demonstrated that "confidence" items without controlling for a motivational qualifier (e.g., if I wanted to) effectively measure motivation more than ability (35). We argue that it was essential to discriminate whether a person was willing or able to perform a task. In other words, if a person had the requisite skills for a specific behavior (able), and they chose not to perform the behavior (willing), this could imply that outcome expectancies were driving that person's actions. Consequently, we have demonstrated (35) that positive outcome expectancies and motivation did, in part, influence responses in personal judgments of confidence. The challenge for PA scientists remains an attempt to separate motivation and outcome expectancies from self-efficacy expectations because of the overwhelming importance of motivation in the decision to enact PA.
Despite more than 100 studies using the TTM to understand PA, few advances that are PA specific have been documented since the original adaptation of the model from smoking behavior (25). For example, constructs from the TPB have been applied using the stage algorithm (3).
The theory itself has undergone extensive criticism because the concept of a stage may not be appropriate for repeated and lifelong health behaviors like PA (5). The thesis of this commentary is that stages need to be discrete with subsequently discrete antecedents, and that the TTM stages may be linear intention/behavior constructions rather than stages of change. However, there is good evidence to support the nonlinear distinction between stages (23). Independent of the stages of change controversy, there has not been strong evidence that the termination stage applies to PA (13). The temptation construct also was found recently to be unrelated to PA controlling for self-efficacy (32), demonstrating the importance to incorporate all TTM constructs to examine if all variables are relevant to PA.
THE APPLIED THEORY AND ANALYSIS OF CHANGE
Although cross-sectional or short longitudinal PA prediction studies can support a basic validation of a model's structure, the methodology is subject to considerable bias (48). Intervention and natural analysis of change studies have four main interrelated aspects of importance: 1) they test the application of the theory for behavior change, which is the central impetus for this line of research and eventual practice; 2) they demonstrate whether a theory's constructs are malleable and capable of change; 3) they provide temporal sequencing of theoretical determinants; and 4) they allow for a test of mediation between the intervention and behavior change, which supports or refutes the proposed structure of the theory. Indeed, intervention research demonstrates the practicality of a theory; thus, it is important that researchers move from initial validation and any augmentation to analyses of change as soon as possible.
Theory of Planned Behavior
The failure of this shift is arguably a central criticism of contemporary TPB research. Although correlational studies of TPB are plentiful, intervention/experimental evidence and analysis of change research have been lacking. For example, a recent review of mediators of PA among adults identified no studies that had used TPB in this capacity, and only three studies that manipulated TPB experimentally (41) despite more than 200 correlational studies using the TPB. Clearly, the new challenge for TPB is to demonstrate its use in PA interventions.
Our research program, centered on the affective attitude construct, has shown some promise for TPB structure in recent studies. We have focused on manipulating affective attitude through targeted messages (11) and interactive video games (43), with successful partial mediation between the intervention conditions and behavior change. The findings support the premise from initial correlational validation to augmented tests of prediction and finally to experimental evidence in interventions.
Self-efficacy has been shown in numerous studies to influence PA participation significantly (26). For example, a study of 174 older adults participating in an exercise program (randomized to an aerobic group or stretching and toning group) measured both exercise self-efficacy and physical efficacy levels across a 12-month trial (27). Results showed that a curvilinear growth of self-efficacy occurred across the trial, with declines at the follow-up stage (postexercise program). Participants, therefore, had increases in both exercise and self-efficacy throughout the program, but then declines occurred after. As a result of the interplay between PA participation and self-efficacy, it becomes essential for practitioners to develop PA opportunities that help build personal self-efficacy levels. The literature surrounding PA interventions using the entire SET (and not solely the self-efficacy construct) is limited, however, and has mixed support (41). Clearly, more studies on constructs in the model, other than self-efficacy, are needed before its overall use can be addressed.
Interventions from our laboratories and others have increased PA behavior successfully (19). However, the true test of the TTM is the application of a mismatched stage approach, that is, randomized control designs comparing a stage-matched intervention (TTM), a nonstage-based program (other theory-based approach), and the TTM program intentionally mismatched to stage (49). Only a handful of interventions applied this design with mixed results across health behaviors (3); yet only one study examined PA with no meaningful differences (9). At this point, it is premature to conclude if stage-based intervention would outperform the others based on tailoring intervention strategies that are specific to each person's readiness to change. Collecting experimental evidence of this is called for to evaluate if the conceptualizations inherent to the TTM may or may not be supported.
There also are several longitudinal studies using the TTM in PA. Our work has found that the TTM constructs are useful for predicting maintenance of, or increases in, public health levels of PA, supporting cross-sectional findings that indicate that, contrary to theory, people seem to use both experiential and behavioral processes while they attempt to increase or maintain their PA (15).
The overwhelming evidence for the benefits of regular PA, curtailed by less positive participation rates, has sparked the necessity of behavioral intervention. Accordingly, theories of exercise and PA have seen enormous attention in the last two decades, and much of this work has focused on models applied from other disciplines. Attention to increasing the methodological rigor in theoretical research has been advocated (46,48), and we certainly concur with these suggestions. Of additional importance, however, is the movement and maturity of theoretical conceptualization in the understanding and application of PA models. If PA psychology is indeed a specific discipline, then the advancement in this domain seems essential.
PA is likely to share many characteristics with other health and social behaviors, and yet we believe it also has a unique compilation of characteristics that can set it apart. For example, it should be repeated several times a week across a lifetime, requires considerable time commitment in each bout, places the body in an aversive body state out of homeostasis, requires environmental supports, produces variable affective responses that are dependent on the load and temporal aspects of the act, and can be achieved via disparate modes (arguably, each a unique behavior). It seems highly likely that these aspects of the behavior set it apart in some capacity from other behaviors; the creation of unique PA models seems appropriate. To aid in this process, we suggest the observation of three components: 1) validity testing of the applied theory, 2) augmentation of the applied theory with PA-specific constructs, and 3) analysis of change of behavior and its constructs (Table).
Validity testing of the applied theory, the first component, is important to establish the use of any framework and its constructs for explaining PA. Our previous examples highlighted that TPB and TTM have undergone extensive validity testing, whereas SET has concentrated on self-efficacy but often neglects the determinants of self-efficacy or outcome expectations (5) and the overarching SCT proposed by Bandura (5). The other important aspect of this component is that researchers can decide to remove constructs without use to create model uniqueness and parsimony to the PA domain. Here, our discipline has shown more reticence, perhaps out of reverence for the parent theoretician and disciplines from which the theory was conceived originally. Nevertheless, the opportunity is important to establish PA theories and to advance our work. In the previous sections, we highlighted that the TPB construct of subjective norm likely falls into this category because it consistently demonstrates a weak predictive use and would not be a choice for behavioral intervention. For the TTM, the structure of the processes has come into question and may warrant a PA-specific, possibly more parsimonious, configuration, and we provided some evidence that the termination stage and the temptation construct may not be necessary in the TTM applied to PA.
The second component of advancing theory is to test augmentation of the applied model with additional PA constructs. This offers rigor to the original model if these additional constructs fail to augment behavior and additional explained variance if they succeed. The approach also can add integration of models and theoretical depth if certain constructs are conceived and tested as mediators of others. All of this testing has the potential to aid in the establishment of unique PA models. In our previous examples, we highlighted how TPB can be extended to include personality, habit, stage of change, and neighborhood factors. We also demonstrated the effectiveness of further dividing within-model constructs when explaining behavior in the context of affective and instrumental attitude in the TPB and multiple efficacies in SET. This component, however, requires the astute theoretician because the redundancy among constructs is highly probable (5); the goal is to include distinct constructs and not old wine but with new labels. Integrated models need to pay careful attention to potential construct redundancies (5).
Finally, the third component of advancing theory that we suggest is to move to analyses of change via either intervention methodologies or longitudinal studies of natural change. This suggestion has been covered in detail in previous commentaries and reviews (e.g., (6,48)) because it represents the most rigorous and applied test of the theory. It also demonstrates that the theory's constructs are modifiable and that modification of the theory's constructs will change PA and provide an effectiveness (cost and utility) assessment of the theory for future intervention campaigns. Indeed, it adds the appropriate test of the original supposition of Lewin (22) about the practicality of theory. In our examples provided in the previous section, we suggest that TTM and SET have advanced in this component more than TPB, which has seen enormous correlational/prediction research in comparison with analysis of change research. As a basic pragmatic suggestion, we believe researchers should move to this component in their theory testing as quickly as possible; an initial correlational design to evaluate the first two components is cost-effective, but the program of work and advancement of the theory likely will stall if component 3 is not observed on a short timetable.
1. Aarts H, Paulussen T, Schaalma H. Physical exercise habit: on the conceptualization and formation of habitual health behaviours. Health Educ. Res
2. Ajzen I. The theory of planned behavior. Organ. Behav. Hum. Decis. Process
3. Armitage CJ. Is there utility in the transtheoretical model? Br J Health Psychol
4. Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol. Rev
5. Bandura A. Health promotion from the perspective of social cognitive theory. Psychol. Health
6. Baranowski T, Anderson C, Carmack C. Mediating variable framework in physical activity interventions: how are we doing? How might we do better? Am. J. Prev. Med
7. Biddle SJH, Nigg CR. Theories of exercise behavior. Int. J. Sport Psychol
8. Blanchard CM, Fortier MS, Sweet SN, et al. Explaining physical activity levels from a self-efficacy perspective: the physical activity counseling trial. Ann. Behav. Med
9. Blissmer B, McAuley E. Testing the requirements of stages of physical activity among adults: the comparative effectiveness of stage-matched, mismatched, standard care, and control interventions. Ann. Behav. Med
10. Cahill SP, Gallo LA, Lisman SA, Weinstein A. Willing or able? The meanings of self-efficacy. J. Soc. Clin. Psychol
11. Conner M, Rhodes RE, Morris B, McEachan R, Lawton R. Changing exercise through targeting affective or cognitive attitudes. Psychol. Health
12. Courneya KS. Antecedent correlates and theories of exercise behaviour. In: Morris T, Summers J, editors. Sport Psychology: Theories, Applications, and Issues
, Sydney, Australia: John Wiley and Sons; 2004. p. 492-512.
13. Courneya KS, Bobick TM. No evidence for a termination stage in exercise behaviour change. Avante
14. Dishman RK, Thom NJ, Rooks CR, Motl RW, Horwath CC, Nigg CR. Failure of post-action stages of the transtheoretical model to predict change in regular physical activity: a multi-ethnic cohort study. Ann. Behav. Med
15. Dishman RK, Vandenberg RJ, Motl RW, Nigg CR. Using constructs of the transtheoretical model to predict classes of change in regular physical activity: a multi-ethnic cohort study. Ann. Behav. Med
16. Godin G, Kok G. The theory of planned behavior: a review of its applications to health-related behaviors. Am. J. Health Promot
17. Haas S, Nigg CR. Construct validation of the stages of change with strenuous, moderate, and mild physical activity and sedentary behaviour among children. J. Sci. Med. Sport
18. Hillsdon M, Foster C, Thorogood M. Interventions for Promoting Physical Activity
. Cochrane Database of Systematic Reviews; 2005.
19. Johnson SS, Paiva AL, Cummins CO, et al. Transtheoretical model-based multiple behavior intervention for weight management: effectiveness on a population basis. Prev. Med
20. Kirsch I. Self-efficacy and outcome expectancy: a concluding commentary. In: Maddux JE, editor. Self-Efficacy, Adaptation, and Adjustment: Theory, Research, and Application (Plenum Series in Social/Clinical Psychology)
, New York: Plenum; 1995.
21. Lawton R, Conner M, McEachan R. Desire or reason: predicting health behaviors from affective and cognitive attitudes. Health Psychol
22. Lewin K. In: Cartwright D, editor. Field Theory in Social Science: Selected Theoretical Papers
, New York: Harper Row; 1951. p. 169.
23. Lippke S, Ziegelmann JP, Schwarzer R, Velicer WF. Validity of stage assessment in the adoption and maintenance of physical activity and fruit and vegetable consumption. Health Psychol
24. Lubans DR, Foster C, Biddle SJH. A review of mediators of behavior in interventions to promote physical activity among children and adolescents. Prev. Med
25. Marcus BH, Selby VC, Niaura RS, Rossi JS. Self-efficacy and the stages of exercise behavior change. Res. Q. Exerc. Sport
26. McAuley E, Blissmer B. Self-efficacy determinants and consequences of physical activity. Exerc. Sport Sci. Rev
27. McAuley E, Katula JA, Mihalko SL, et al. Mode of physical activity and self-efficacy in older adults: a latent growth curve analysis. J. Gerontol. Psychol. Sci
28. McCrae RR, Costa PT, Ostendorf F, et al. Nature over nurture: temperament, personality, and life-span development. J. Pers. Soc. Psychol
29. Nigg CR, Jordan PJ. Commentary: it's a difference of opinion that makes a horserace. Health Educ. Res
30. Nigg CR, GJ N, Rossi JS, Benisovich SV. Processes of exercise behavior change: redeveloping the scale. Ann. Behav. Med
31. Nigg CR, Borrelli B, Maddock J, Dishman RK. A theory of physical activity maintenance. Appl. Psychol. Int. Rev
32. Nigg CR, McCurdy DK, McGee KA, et al. Relations among temptations, self-efficacy, and physical activity. Int. J. Sport Exerc. Psychol
33. Paxton RJ, Nigg CR, Motl RW, et al. Are constructs of the transtheoretical model for physical activity measured equivalently between sexes, age groups, and ethnicities? Ann. Behav. Med
34. Prochaska JO, DiClemente CC. Transtheoretical therapy: toward a more integrative model of change. Psychother. Theory Res. Pract
35. Rhodes RE, Blanchard CM. What do confidence items measure in the physical activity domain? J. Appl. Soc. Psychol
36. Rhodes RE, Courneya KS. Relationships between personality, an extended theory of planned behaviour model, and exercise behaviour. Br. J. Health Psychol
37. Rhodes RE, Courneya KS, Blanchard CM, Plotnikoff RC. Prediction of leisure-time walking: an integration of social cognitive, perceived environmental, and personality factors. Int. J. Behav. Nutr. Phys. Act
38. Rhodes RE, de Bruijn GJ, Matheson DH. Habit in the physical activity domain: integration with intention temporal stability and action control. J. Sport Exerc. Psychol
39. Rhodes RE, Fiala B, Conner M. Affective judgments and physical activity: a review and meta-analysis. Ann. Behav. Med
40. Rhodes RE, Mark R. Social cognitive theories. In: Acevedo EO, editor. The Oxford Handbook of Exercise Psychology
, New York: Oxford University Press (>in press)>.
41. Rhodes RE, Pfaeffli LA. Mediators of physical activity behaviour change among adult nonclinical populations: a review update. Int. J. Behav. Nutr. Phys. Act
42. Rhodes RE, Pfaeffli LA. Personality. In: Acevedo EO, editor. The Oxford Handbook of Exercise Psychology
, New York: Oxford University Press (in press).
43. Rhodes RE, Warburton DER, Bredin SS. Predicting the effect of interactive video bikes on exercise adherence: an efficacy trial. Psychol. Health Med
44. Spence JC, Burgess JA, Cutumisu N, et al. Self-efficacy and physical activity: a quantitative review. J. Sport Exerc. Psychol
45. Spencer L, Adams TB, Malone S, Roy L, Yost E. Applying the transtheoretical model to exercise: a systematic and comprehensive review of the literature. Health Promot. Pract
46. Sutton S. Using social cognition models to develop health behaviour interventions: problems and assumptions. In: Rutter D, Quine L, editors. Intervention Research With Social Cognition Models
, Buckingham, England: Open University Press; 2002. p. 193-208.
47. Symons Downs D, Hausenblas HA. Exercise behavior and the theories of reasoned action and planned behavior: a meta-analytic update. J. Phys. Act. Health
48. Weinstein ND. Misleading tests of health behavior theories. Ann. Behav. Med
49. Weinstein ND, Rothman AJ, Sutton SR. Stage theories of health behavior: conceptual and methodological issues. Health Psychol