Exercise habit strength, planning and the theory of planned behaviour: An action control approach

Psychology of Sport and Exercise 12 (2011) 106e114

Contents lists available at ScienceDirect

Psychology of Sport and Exercise journal homepage: www.elsevier.com/locate/psychsport

Exercise habit strength, planning and the theory of planned behaviour: An action control approach Gert-Jan de Bruijn* Amsterdam School of Communication Research ASCoR, University of Amsterdam, Kloveniersburgwal 48, 1012 CX Amsterdam, The Netherlands

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 June 2010 Received in revised form 23 September 2010 Accepted 8 October 2010 Available online 15 October 2010

Objectives: Action control refers to the successful translation of intention into behaviour. The purpose of this study was to explore the potential usefulness of extending intentioneexercise profiles with past exercise behaviour and exercise habit strength and the potential discriminative effect of action planning items and theory of planned behaviour (TPB) concepts. Design: Prospective data from 330 undergraduate students (M age ¼ 21.5; 25.5% males). Method: Measures of exercise behaviour, exercise habit strength, TPB concepts and action plans were assessed at T1; subsequent exercise behaviour was assessed again two weeks later. Profiles were created from T1 exercise behaviour, intention, habit strength and T2 exercise behaviour. Data were analyzed using chi-square analysis, discriminant function analysis and analysis of variance and interpreted using p-values and effect sizes. Results: There was considerable asymmetry in the intentioneexercise relationship, with successful exercise intenders reporting stronger exercise habits. However, more than 40% of strongly habitual exercise intenders were not following on these intentions. Measures of perceived behavioural control were the consistent predictor of action control, but could not discriminate differences between key target groups. Effect sizes for significant differences were mostly large. Planning items were generally unrelated to exercise action control. Conclusion: The extension of intentioneexercise profiles revealed noticeable distributions to allow for better exercise target group detection. Measures of controllability of exercise behaviour should be promoted in several of these target groups, but research should explore additional predictors of key target groups in order to enhance exercise levels. Ó 2010 Elsevier Ltd. All rights reserved.

Keywords: Exercise Action control Theory of planned behaviour Habit strength Action plans

Introduction Public health guidelines for (young) adults prescribe that regular bouts of exercise (i.e. three days per week for a minimum of 20 min per bout (Haskell et al., 2007)) are needed in order to accrue welldocumented health benefits, including the prevention of weight gain (Kromhout, Bloemberg, Seidell, Nissinen, & Menotti, 2001) and reduced risk for cardiovascular diseases (Aadahl et al., 2009). However, large segments of the population are currently insufficiently active to meet these guidelines and fail to achieve these health benefits. This emphasizes a need to develop interventions that target increased exercise levels. Theories and models of human behaviour are considered to be pivotal in the intervention development phase, because they outline important behavioural determinants that can be altered through the use of educational and promotional efforts (Brug,

* Tel.: þ31 205252636; fax: þ31 205253681. E-mail address: [email protected]. 1469-0292/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.psychsport.2010.10.002

Oenema, & Ferreira, 2005). Socio-cognitive behavioural theories have traditionally been utilized in this development phase and one of the most commonly used theories is the theory of planned behaviour (TPB) (Ajzen, 1991). This theory postulates that behavioural enactment (i.e. engaging in exercise) is primarily determined by the intention to act and three socio-cognitive concepts influence this intention concept. These are attitudes (positive or negative evaluations of the outcomes of behavioural performance, in which both an instrumental and affective component is postulated), subjective norms (perceived norms about whether important others in the social environments believe on should perform the behaviour) and perceived behavioural control (PBC): this latter construct reflect the extent to which performance of behaviour is easy or difficult, but also if performance is under one’s control or not. Reviews and metaanalysis of the TPB in the exercise domain have demonstrated its sufficiency (Godin, 1994; Hagger, Chatzisarantis, & Biddle, 2002; Hausenblas, Carron, Mack, & Godin, 1997). Next to the delineation of pathways linking socio-cognitive variables with intention and behaviour, the TPB is also considered to be

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114

a linear model (Ajzen, 1991) in which higher scores on, for example, attitudes are related to higher scores on intention: a similar linear relationship is also postulated between intention and behaviour. Some evidence, however, indicates that this theoretical linearity may be partly misplaced in the exercise domain. Even though Rhodes and Courneya (2005) found similar linear relationships with exercise behaviour for both PBC and instrumental attitude (r ¼ .24), threshold analysis indicated non-significant differences in exercise behaviour between groups based on either high or very high levels of exercise PBC. In another study, Rhodes, Courneya, and Jones (2003) found evidence of threshold effects of intention on exercise behaviour, with those who intended to exercise beyond this threshold being no more successful in exercise enactment than those at the intention threshold. Such findings may indicate that TPB constructs related to exercise behaviour have specific thresholds and that exercise interventions that attempt to target changes above this threshold may be misplaced (Rhodes & Courneya, 2005). Additional evidence from the exercise domain has implicitly demonstrated this threshold effect of intention on exercise behaviour, with a substantial portion of research samples holding positive exercise intentions but not translating those intentions into actual exercise behaviour (De Bruijn, De Groot, Van den Putte, & Rhodes, 2009; Rhodes, De Bruijn, & Matheson, 2010; Rhodes et al., 2003; Rhodes & Plotnikoff, 2006; Rhodes, Plotnikoff, & Courneya, 2008). As a result, linear approaches may be insufficient to fully understand the intentioneexercise gap. Categorical approaches have been suggested and applied in exercise action control research (Rhodes et al., 2003; Rhodes et al., 2010; Rhodes et al., 2008; Rhodes & Plotnikoff, 2006) by acknowledging two distinct phases, namely action planning (i.e. forming an intention) and action control (i.e. translating this intention into actual exercise behaviour) (Rhodes & Plotnikoff, 2006; Rhodes et al., 2008; Sniehotta, Nagy, Scholz, & Schwarzer, 2006). Within this type of research, four possible categories or profiles are commonly outlined. These are disinclined abstainers (those who do not intend to exercise and do not exercise), disinclined actors (those who do not intend to, but do exercise), inclined abstainers (those who intend to exercise but do not) and inclined actors (those who intend to exercise and do). Distributions of these profiles have demonstrated considerable asymmetry in the intentioneexercise relationship, with the smallest cell size (generally less than 5%) for the disinclined actors and the largest cell sizes for disinclined abstainers and inclined actors and abstainers (Rhodes et al., 2010, 2008; Rhodes & Plotnikoff, 2006). Nevertheless, of those who hold positive exercise intention, between 30% and 50% fail to act upon those intentions and an investigation of predictors of exercise action control should be able to inform exercise interventions targeted at an already motivated population. Concepts from various socio-cognitive models have been used in this line of research. For instance, applying the TPB to exercise action control, Rhodes et al. (2003) found that inclined actors had significantly higher scores on measures of exercise attitude and PBC than inclined abstainers. Similar results have been found in other studies applying the TPB to exercise action control (De Bruijn et al., 2009; Rhodes & Plotnikoff, 2006; Rhodes et al., 2008), with affective attitude measures generally outperforming instrumental attitudes measures. Similarly, using the constructs detailed in the transtheoretical model, protection motivation theory and the TPB in exercise action control research has indicated that behavioural processes, self-efficacy and PBC were strong predictors of intentione exercise relationships (Rhodes et al., 2008). Thus, socio-cognitive factors such as PBC and affective attitude should be considered as persuasive message input factors in order to facilitate the enactment of exercise behaviour amongst those who already hold positive exercise intentions. Notwithstanding the potential of PBC and affective attitude in exercise action control, self-regulatory strategies are also increasingly

107

suggested as promising constructs to bridge the intentioneexercise gap (Gollwitzer & Sheeran, 2006; Sniehotta, Scholz, & Schwarzer, 2005, 2006). These strategies include formulating implementation intentions (Gollwitzer, 1999; Gollwitzer & Sheeran, 2006) and implemental planning or action plans (Lippke, Ziegelmann, & Schwarzer, 2004; Sniehotta, Scholz et al., 2006; Sniehotta, Schwarzer, Scholz, & Schuz, 2005). These latter strategies include the formulation of plans in which one specifies how, when, and/or where to act. Action plans are thought to facilitate behavioural enactment by detailing how to act when a specific situational cue is encountered (Bandura, 1998; Gollwitzer & Schaal, 1998; Gollwitzer & Sheeran, 2006). Exercise determinant research applying action planning has shown that intentioneexercise relationships are generally stronger when stronger action plans have been formulated (Norman & Conner, 2005; Wiedemann, Schuz, Sniehotta, Scholz, & Schwarzer, 2009). Nevertheless, as previously argued, the asymmetric nature of the intentioneexercise relationship suggests that such linear statistical approaches may be insufficient to fully capture important predictors of exercise action control: investigation of the potential relevance of action planning in exercise action control using categorical statistical analyses may hold promise for a better delineation of relevant factors that impede or facilitate the translation of positive exercise intentions. Recent research endeavours on exercise action control have also demonstrated that the extension of intentioneexercise profiles with past exercise behaviour yields relevant knowledge for exercise interventions (Rhodes & Plotnikoff, 2006; Rhodes et al., 2008; Sniehotta, Nagy et al., 2006; Sniehotta, Scholz et al., 2005). First, these studies have demonstrated that the intentioneexercise gap exists not only in exercise adopters (those who have not engaged in previous exercise behaviour), but also in exercise maintainers (those who have previously engaged in exercise). For instance, Rhodes and Plotnikoff (2006) found not only that 65% of exercise adopters were not following up on their positive exercise intentions, but also that more than a quarter of exercise maintainers were not following up on their positive exercise intentions. Moreover, these authors found various predictors of action control that were independent of previous exercise status, but also reported predictors of action control that were more pronounced in exercise adopters as compared to exercise maintainers. For instance, control beliefs regarding exercise in bad weather and exercising in the company of other were key discriminators between successful and unsuccessful maintainers, but not between successful and unsuccessful adopters. Thus, exercise relapses in exercise maintainers could be prevented by persuasive messages emphasizing the relevance of exercise company and targeting beliefs regarding exercising in bad weather. Although practically informative, the application of past exercise behaviour holds little theoretical value for developing new or changing current behavioural models. That is, past behaviour has often been denoted as an empty construct (Eagly & Chaiken, 1993; Verplanken & Aarts, 1999) and viable mechanisms linking past behaviour with current behaviour should be explored in order to enhance behavioural models (Ajzen, 2002; Chatzisarantis & Hagger, 2007; Hagger, Chatzisarantis, & Biddle, 2001). Behavioural recurrence has often been suggested to reflect the operation of habits (Aarts, Paulussen, & Schaalma, 1997; Aarts, Verplanken, & van Knippenberg, 1998; Triandis, 1977), which are conceptualized as behaviours that are automatically set in motion by features of the environment, rather than by planned intentions. Research has indeed indicated that stronger exercise habits make exercise behaviour less intentional (De Bruijn & Rhodes, in press). Further, studies have recently begun extending intentionebehaviour profiles with a validated measure of habit strength (Verplanken & Orbell, 2003) in both dietary (De Bruijn, 2010) and physical

108

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114

activity behaviours (De Bruijn & Gardner, in press), including exercise behaviour (De Bruijn & Rhodes, in press; Rhodes et al., 2010). These studies have shown that exercise abstainers or nonintenders are mostly low habitual exercisers, whereas strongly habitual incliners largely comprise of inclined actors. For instance, Rhodes et al. (2010) found that, amongst strongly habitual exercisers, about a fifth could be categorized as exercise nonintenders, whereas amongst weakly habitual exercisers, more than three quarters could be categorized as exercise nonintenders. However, of the strongly habitual and inclined exercisers, about a quarter did not translate these intention into subsequent exercise behaviour, indicating that the intentioneexercise gap is also existent amongst those who hold strong exercise habits (De Bruijn & Rhodes, in press; Rhodes et al., 2010). An action control approach including measures of habit strength may shed light on important predictors, but no studies to date have explored predictors of these profiles. Similarly, there is at present no empirical evidence regarding the distribution and the predictors of exercise behavioural profiles that are extended with both past exercise behaviour and exercise habit strength. There is sufficient methodological and conceptual evidence that indicates that behavioural recurrence does not provide evidence for the operational of habitual processes (Ajzen, 2002). Alternatively, habits not only encompass behavioural recurrence, but also such features as automaticity and ease of performance. Thus, the extension of intentioneexercise profiles with both exercise habit strength and past exercise behaviour may shed some light on past behaviour and habit strength distributions. Likewise, the identification of predictors of profiles that are created from both past exercise behaviour and exercise habit strength may allow for the development of exercise interventions that target, for example, changing weak exercise habits into strong exercise habits in maintainers or adopters. Strongly habitual behaviours are theorized to be automatically initiated, and so a focus on developing strong exercise habits suggests not only that these promoted behavioural habits should be more likely to be maintained, but also that relapses are more likely to be prevented, after the intervention has ended (Verplanken & Wood, 2006). The present study was developed to fill the aforementioned research gaps. First, the potentially useful extension of intentione exercise profiles with past exercise behaviour and exercise habit strength was explored to find which behavioural profiles would emerge. Second, predictors of profile membership were explored in order to inform exercise interventions. Given their demonstrated usefulness in various exercise determinant studies, both TPB concepts and implemental planning items were modelled as potential predictors of these profiles. Based on previous evidence, PBC, affective attitude and planning items were hypothesized to be the most discriminant constructs in action control. Given their presumed importance for (particularly) behavioural change, planning items were also hypothesized to be the key predictors of action control in exercise adopters and those who hold weak exercise habits. Methods Participants and procedures Undergraduate students voluntarily participated in this study as a part of an introductory social psychology course, receiving course credits for their study participation. Invitations to participate were sent via email and announced during college hours and data were collected through self-administered questionnaires using an online survey tool hosted by the university. At T1, 561 students participated (mean age ¼ 21.41 (SD ¼ 2.85); 165 (27.8%) males). Two weeks later, 330 students (mean age ¼ 21.49 (SD ¼ 3.04); 84 (25.5%) males) provided additional data, indicating a 41.2% dropout.

Logistic regression analysis did not show significant associations with T1 study variables, age and gender with dropout (p-values > .112). Measures Exercise behaviour was assessed at T1 and T2 using the relevant items from the International Physical Activity Questionnaire (IPAQ) (Hagstromer, Oja, & Sjostrom, 2006). These items reflected public health guidelines for weekly exercise and questioned respondents to indicate the number of exercise bouts per week and the usual amount of exercise time per bout in the past four weeks (T1) and the past two weeks (T2). Those who reported engaging in at least three exercise bouts per week and a minimum duration of 20 min per bout were coded as meeting the exercise norm; all other participants were coded as not meeting the exercise norm. For the TPB concepts, planning items and habit strength measures, participants were asked to consider the public health guidelines for exercise and use a time reference period of two weeks when answering (i.e. exercising on at least three bouts per week and for at least 20 min per bout in the next two weeks is something.). Exercise habit strength was assessed at T1 using the self-reported habit index (SRHI), developed by Verplanken and Orbell (2003). This measure captures key elements of habit strength, including behavioural recurrence, automaticity, uncontrollability and lack of awareness (Bargh, 1994; Verplanken & Orbell, 2003) assessed on 7-point scale (3 ¼ strongly disagree; þ3 ¼ strongly agree). Given the focus of the current study on past exercise behaviour and exercise habit strength, the two items from the SRHI that capture behavioural recurrence were omitted from scale construction in order to prevent conceptual overlap with past exercise behaviour and potential inflation of coefficients. Reliability was .97. Intention to exercise was assessed with two items on 7-point bipolar scales. These two items were ‘I intend to engage in sufficient exercise’ and ‘I am sure I will engage in sufficient exercise (þ3 ¼ yes, definitely; 3 no, definitely not). Inter-item correlation was .92. For the habit strength and intention measure, midscale values were used to separate participants into nonintenders (individual intention value below or equal to midscale) versus intenders (individual intention value above midscale) and into weakly habitual exercisers (individual habit strength value below or equal to midscale) versus strongly habitual exercises (individual habit strength value above midscale). Attitude towards exercise was assessed using 7-point bipolar items: instrumental and affective attitude were measured as separate concepts. Instrumental attitude was assessed with two items (very goodevery bad; very healthyevery unhealthy) and affective attitude with three items (very pleasantevery unpleasant; very enjoyableevery unenjoyable; very relaxingevery stressful), with higher scores indicating more positive exercise attitudes. Reliability for affective attitude was a ¼ .94 and inter-item correlation for instrumental attitude was .88. Subjective norm was measured with five items on 7-point bipolar scales, reflecting perceived pressure to exercise from parents, partner, friends, fellow students, and roommates: reliability was sufficient (a ¼ .64) and higher scores related to stronger perceptions of subjective norms. Perceived behavioural control was measured with two items on 7point bipolar scales, reflecting ease of performance (very easyevery difficult) and perceived control (definitely under my controledefinitely not under my control) (inter-item correlation ¼ .82) with higher scores indicating higher levels of PBC. Action planning was assessed with five items that questioned whether or not (þ3 ¼ totally agree; 3 ¼ totally disagree) respondents had made specific plans for the coming two weeks regarding what kind of exercise, and where, when, how often, and with whom to exercise.

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114 Table 1 Cell sizes for profiles created from T1 intention and T2 exercise behaviour (A), supplemented with T1 exercise behaviour (B), T1 exercise habit strength (C), and T1 exercise behaviour and exercise habit strength (D). Exercise habit strength T1

Exercise norm T1

Exercise norm T2

T1 intention Negative (n ¼ 167)

Positive (n ¼ 163)

No Yes

165a 2b

104c 59d

No Yes No Yes

161a 2b 4a 0

71c 12d 33c 47d

No Yes No Yes

147a 2b 18a 0b

34c 10d 70c 49d

No Yes No Yes No Yes No Yes

146a 2b 1a 0b 15a 0b 3a 0b

31c 3d 3c 7d 40c 9d 30c 40d

A

B No No Yes Yes C Weak Weak Strong Strong D Weak Weak Weak Weak Strong Strong Strong Strong a b c d

No No Yes Yes No No Yes Yes

109

that allowed for a maximum of 24 ¼ 16 profiles. Power analysis was performed in order to identify the minimum cell size per profile that allowed a medium effect size (f ¼ .25) to be detected (Cohen, 1992): a cell size of n ¼ 29 was considered sufficient at 95% power with a ¼ .05 (Erdfelder, Faul, & Buchner, 1996). Chi-square tests were performed to test sampling distributions, while profile membership was predicted using discriminant function analysis, which is commonly applied in (exercise) action control research (Araújo-Soares, McIntyre, & Sniehotta, 2009; De Bruijn & Rhodes, in press; Rhodes et al., 2003, 2008; Rhodes & Plotnikoff, 2006; Sheeran, 2002). Planning items and TPB concepts were modelled as predictors and significant effects were followed up by analyses of variance with Tukey post-hoc comparisons. Statistical significance was set at p < .005 and Cohen’s effect size d was used to aid in the interpretation of significant differences (small effect size ¼ .20; medium effect size ¼ .50; large effect size ¼ .80) (Cohen, 1992). Results Intention T1 e exercise T2

disinclined abstainer at T2. disinclined actor at T2. inclined abstainer at T2. inclined actor at T2.

Disaggregated scales for the planning items were favoured in order to identify specific intervention targets for exercise promotion efforts (Fishbein & Cappella, 2006; Rhodes & Plotnikoff, 2006; Van den Putte & Dhondt, 2005). Analysis Participants were separated initially by T1 intention (positive vs. negative) and T2 exercise status (meeting norm vs. not meeting norm). These profiles were subsequently supplemented with past exercise status (meeting norm vs. not meeting norms) and/or exercise habit strength (weak vs. strong) (see Table 1 for details)

One hundred and sixty-three respondents could be categorized as incliners. However, only 59 (36.19%) were sufficiently vigorously active two weeks later (see Table 1A). In contrast, of the 167 disincliners, only 2 (1.20%) were sufficiently vigorously active two weeks later, indicating considerable asymmetry (c2 ¼ 67.056 (1), p < .001) in the intentioneexercise relationship in this sample. The cell size for the disinclined actors (n ¼ 2) was considered too small and excluded from the discriminant function analysis, which revealed one significant discriminant function (c2 ¼ 97.850 (8), p < .001; canonical correlation ¼ .786, Wilk’s D ¼ .337). This function correctly classified 76.73% of cases, accounted for 92.40% between-group variability and explained 66.30% of generalized variance in the dependent variables. The correlation of the predictor variables with this function are displayed in Table 2, which also has the mean scores of these predictor variables across the included profiles and the results of the post-hoc tests. With the exception of subjective norm, all TPB variables could distinguish between action planning and action control. However, only PBC was related to exercise action control, with significantly higher scores for PBC amongst inclined actors (n ¼ 59) than amongst inclined abstainers (n ¼ 104). No planning items could significantly discriminate inclined actors from inclined abstainers (p > .560). Effect sizes for significant differences ranged from .39 to 3.13, indicating small-to-large effect sizes.

Table 2 Correlations with discriminant function and mean scores and standard deviations for theory of planned behavior constructs and planning items across profiles constructed from T1 exercise intention and meeting exercise norm at T2 (¼T1 þ 2 weeks) (n ¼ 328).

T1 Intention T2 Exercise norm

Predictors Instrumental attitude Affective attitude Subjective norm PBC Planning when Planning where Planning what kind of exercise Planning how often Planning with whom to exercise ***

p < .001;

**

p < .01; *p < .05.

1 (n ¼ 59)

2 (n ¼ 104)

3 (n ¼ 165)

Positive

Positive

Negative

Yes

No

No

Correlation with discriminant function

Mean (SD)

Mean (SD)

Mean (SD)

F2,325

Post-hoc

.244 .479 .203 .896 .378 .431 .362 .430 .306

2.63 2.20 .58 2.08 1.19 1.34 1.41 1.36 .95

2.63 2.08 .65 1.18 1.29 1.63 1.71 1.40 1.25

2.30 .58 .07 1.23 .96 .53 .24 .70 .81

6.764** 64.525*** 3.175* 266.815*** 60.775*** 49.157*** 42.154*** 55.447*** 48.963

1,2 > 3 1,2 > 3 e 1 > All; 2 > 3 1,2 > 3 1,2 > 3 1,2 > 3 1,2 > 3 1,2 > 3

(.71) (.96) (.87) (1.13) (1.97) (2.01) (1.97) (1.90) (1.99)

(.67) (.95) (.91) (1.18) (1.62) (1.52) (1.38) (1.54) (1.61)

(.90) (1.46) (1.25) (1.03) (1.87) (2.04) (2.02) (1.90) (.184)

110

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114

Intention T1 e exercise T1 e exercise T2

which revealed two significant discriminant functions (Function 1:

Of the 59 inclined actors at T2, 47 (79.66%) could be categorized as successful maintainers and 12 as successful adopters. Conversely, of the 104 inclined abstainers, 71 (68.27%) were also not sufficiently vigorously active at T1. Of the 165 disinclined abstainers, only 4 (2.42%) were meeting T1 exercise norm, while no respondents could be categorized as disinclined successful maintainers and only 2 participants could be categorized as disinclined adopters (see Table 1B). The cell sizes for the disinclined actors (n ¼ 2), the disinclined abstainers who were previously sufficiently active (n ¼ 4), and the inclined actors who were previously insufficiently active (n ¼ 12) were considered too small and excluded from the discriminant function analysis, which revealed one significant discriminant function (c2 ¼ 126.818 (27), p < .001; canonical correlation ¼ .858, Wilk’s D ¼ .211). The function correctly classified 79.8% of cases, accounted for 92.10% between-group variability and explained 78.90% of generalized variance in the dependent variables. The correlation of the predictor variables with this function are displayed in Table 3, which also has the mean scores of these predictor variables across the included profiles and the results of the post-hoc tests. With the exception of subjective norm and instrumental attitude, nonintenders (n ¼ 161) had significantly lower scores on all predictor variables than the other three profiles. Also, successful (n ¼ 47) and unsuccessful maintainers (n ¼ 71) had significantly higher scores on PBC measures than did unsuccessful adopters (n ¼ 71). However, no other TPB variables or planning items could discriminate between successful and unsuccessful maintainers (p > .329). Effect sizes for significant differences ranged from .45 to 3.63, indicating small-tolarge effect sizes.

T1 intention e T1 habit strength e T2 exercise Of the 59 inclined actors at T2, 49 (83.05%) reported strong T1 exercise habits, while 89.09% (n ¼ 147) of the disinclined abstainers reported weak T1 exercise habits. Of those who reported strong T1 exercise habits (n ¼ 137), 119 (86.86%) had positive exercise intentions. Conversely, of those who reported weak T1 exercise habits (n ¼ 149), only 2 (1.34%) reported positive exercise intentions (see Table 1C). The cell sizes for the weakly habitual disinclined (n ¼ 2) and inclined (n ¼ 10) actors, and the strongly habitual disinclined abstainers (n ¼ 18) and actors (n ¼ 0) were considered too small and excluded from the discriminant function analysis,

c2 ¼ 123.978 (27), p < .001; canonical correlation ¼ .846, Wilk’s D ¼ .202; Function 2: c2 ¼ 26.636 (12), p ¼ .046; canonical correlation ¼ .432, Wilk’s D ¼ .709). Function 1 accounted for 87.0% between-group variability and explained 79.80% of generalized variance in the dependent variables associated with this function; Function 2 accounted for 7.90% between-group variance and explained 29.10% of generalized variance in the dependent variables associated with this function. The two functions correctly classified 75.3% of cases. The correlation of the predictor variables with these functions are displayed in Table 4, which also has the mean scores of these predictor variables across the included profiles and the results of the post-hoc tests. Neither instrumental attitude nor subjective norm could discriminate between the created profiles, but affective attitude was significantly lower amongst nonintenders than amongst the other three profiles. A similar pattern was observed for the differences regarding planning items relating to what kind of exercise and when, where, and with whom to exercise. The planning item relating to how often to exercise was also discriminative between strongly habitual inclined actors and abstainers as compared to weakly habitual inclined abstainers. No planning items could discriminate between strongly habitual inclined actors and strongly habitual inclined abstainers (p > .469), but significantly higher scores for PBC were found for strongly habitual inclined actors as compared to strongly habitual inclined abstainers. Effect sizes for significant differences were all >.80, indicating large effect sizes. T1 intention e T1 habit strength e T1 exercise e T2 exercise Of the 59 inclined actors at T2, three (5.08%) respondents were weakly habitual exercise adopters, seven (11.86%) were strongly habitual exercise adopters, nine (15.25%) were strongly habitual exercise adopters and 40 (67.80) were strongly habitual exercise maintainers. Furthermore, of the 165 disinclined abstainers at T2, 146 (88.48%) reported weak exercise habits and no exercise behaviour at T1, while 15 (9.09%) reported strong exercise habits but no exercise behaviour at T1. No respondents could be categorized as weakly habitual nonintenders who were currently and previously sufficiently active and strongly habitual nonintenders who were sufficiently active at T2 and/or T1 (see Table 1D). With the exception of the cell for the disinclined weakly habitual nonexercisers (n ¼ 146), cell sizes for the disincliners were considered

Table 3 Correlations with discriminant function and mean scores and standard deviations for theory of planned behaviour constructs and planning items across profiles constructed from T1 exercise intention and meeting exercise norm at T1 and T2 (¼T1 þ 2 weeks) (n ¼ 312). 1 (n ¼ 47)

2 (n ¼ 33)

3 (n ¼ 71)

4 (n ¼ 161)

T1 intention

Positive

Positive

Positive

Negative

T1: Exercise norm

Yes

Yes

No

No

T2: Exercise norm

Yes

No

No

No

Correlation with discriminant function

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

F3,308

Post-hoc

.181 .364 .154 .960 .328 .383 .341 .428 .326

2.66 2.28 .48 2.26 1.32 1.53 1.64 1.45 1.04

2.70 2.25 .82 2.12 1.55 1.82 1.94 1.85 1.72

2.59 2.00 .56 .74 1.17 1.54 1.61 1.20 1.03

2.31 .56 .026 1.29 .99 .57 .27 .73 .83

4.665** 42.318*** 2.105 238.283*** 42.791*** 34.932*** 30.896*** 39.287*** 34.643***

1,2 > 4 1,2,3 > 4 ns 1 > 3,4; 2 > 3,4; 3 > 4 1,2,3 > 4 1,2,3 > 4 1,2,3 > 4 1,2,3 > 4 1,2,3 > 4

Predictors Instrumental attitude Affective attitude Subjective norm Perceived behavioural control Planning when Planning where Planning what kind of exercise Planning how often Planning with whom to exercise ***

p < .001;

**

p < .01; *p < .0.

(.63) (.92) (1.04) (1.04) (1.88) (1.89) (1.81) (1.86) (2.00)

(.67) (.95) (.97) (.81) (1.60) (1.49) (1.41) (1.46) (1.42)

(.68) (.95) (.89) (.106) (1.63) (1.54) (1.36) (1.54) (1.65)

(.89) (1.47) (1.24) (.96) (1.86) (2.05) (2.04) (1.88) (1.84)

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114

111

Table 4 Correlations with discriminant function and mean scores and standard deviations for theory of planned behaviour constructs and planning items across profiles constructed from T1 exercise intention, T1 exercise habit strength and meeting exercise norm at T2 (¼T1 þ 2 weeks) (n ¼ 300). 1 (n ¼ 49)

2 (n ¼ 70)

3 (n ¼ 34)

4 (n ¼ 147)

T1 Intention

Positive

Positive

Positive

Negative

T1: Exercise Habit Strength

Strong

Strong

Weak

Weak

T2 Exercise Norm

Yes

No

No

No

Predictors Instrumental attitude Affective attitude Subjective norm Perceived behavioural control Planning when Planning where Planning what kind of exercise Planning how often Planning with whom to exercise ***

p < .001;

**

Correlation with discriminant function 1

Correlation with discriminant function 2

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

F3,308

Post-hoc

.161 .417 .204 .947 .322 .355 .309 .411 .304

.167 .297 .061 .057 .404 .528 .590 .537 .255

2.63 2.29 .88 2.26 1.35 1.49 1.55 1.47 1.22

2.64 2.26 .74 1.54 1.51 1.80 1.80 1.93 1.33

2.59 1.72 .44 .43 .82 1.26 1.53 .32 1.09

2.35 .45 .07 1.35 1.10 .70 .39 .85 .95

3.254* 50.129*** 2.854* 231.882*** 47.373*** 38.588*** 32.623*** 51.604*** 39.222***

e 1,2,3 > 4 e 1 > All; 2 > 3,4, 3 > 4 1,2,3 > 4 1,2,3 > 4 1,2,3 > 4 1 > 3,4; 2 > 3,4; 3 > 4 1,2,3 > 4

(.72) (.94) (.52) (1.03) (1.98) (2.00) (1.95) (1.92) (1.94)

(.70) (.92) (.84) (1.04) (1.58) (1.31) (1.21) (1.08) (1.51)

(.63) (.93) (1.08) (1.09) (1.64) (1.85) (1.67) (1.77) (1.80)

(.87) (1.46) (1.25) (.93) (1.80) (2.02) (2.01) (1.84) (1.79)

p < .01; *p < .05.

too small (range n ¼ 1 through n ¼ 15) and were not included in the discriminant function analysis. The cell sizes for weakly habitual exercise adopters (n ¼ 3) and unsuccessful maintainers (n ¼ 3), as well as the weakly habitual successful maintainers (n ¼ 7) and strongly habitual successful adopters (n ¼ 9) were also considered too small and excluded. The discriminant function analysis was thus run with the remaining five profiles (see Table 5), which revealed one significant discriminant function (c2 ¼ 137.191 (36), p < .001; canonical correlation ¼ .878, Wilk’s D ¼ .153) that correctly classified 80.2% of cases, accounted for 88.40% betweengroup variability and explained 84.70% of generalized variance in the dependent variables. The correlation of the predictor variables with these functions are displayed in Table 5, which also has the mean scores of these predictor variables across the included profiles and the results of the post-hoc tests. Again, neither instrumental attitude nor subjective could discriminate between the created profiles, but significantly lower scores for affective attitude were found for weakly nonintenders as compared to the other four profiles. A similar pattern was observed for the

differences regarding planning items relating to what kind of exercise and when, where, and with whom to exercise. The planning item relating to how often to exercise was also discriminative between those with strong exercise habits and weakly habitual inclined abstainers. Furthermore, significantly higher scores for PBC were found for strongly habitual inclined who were sufficiently active at T1 as compared to the three other profiles, but none of the predictor variables could discriminate between strongly habitual inclined successful and unsuccessful maintainers (p > .711). Effect sizes for significant differences were all >.80, indicating large effect sizes. Discussion In the present study, a categorical action control approach was adopted to identify relevant predictors of exercise action control in a young adult sample. Profiles were initially created from intention and subsequent exercise behaviour two weeks later and then supplemented with past exercise behaviour and/or exercise habit

Table 5 Correlations with discriminant function and mean scores and standard deviations for theory of planned behaviour constructs and planning items across profiles constructed from T1 exercise intention, T1 exercise habit strength and meeting exercise norm at T1 and T2 (¼T1 þ 2 weeks) (n ¼ 287). 1 (n ¼ 40)

2 (n ¼ 30)

3 (n ¼ 40)

4 (n ¼ 31)

5 (n ¼ 146)

T1 Intention

Positive

Positive

Positive

Positive

Negative

T1: Exercise habit strength

Strong

Strong

Strong

Weak

Weak

T1 exercise norm

Yes

Yes

No

No

No

T2 Exercise norm

Yes

No

No

No

No

Correlation with discriminant function 1

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

F4,282

Post-hoc

.145 .352 .167 .955 .319 .367 .323 .434 .351

2.70 2.40 .88 2.36 1.53 1.75 1.85 1.63 1.38

2.67 2.27 .82 2.15 1.57 1.97 2.03 2.13 1.83

2.63 2.26 .66 1.09 1.48 1.68 1.63 1.78 .95

2.55 1.68 .44 .29 .77 1.35 1.58 .45 1.13

2.34 .44 .07 1.36 1.08 .71 .40 .84 .94

2.800* 37.047*** 1.813 192.549*** 36.374*** 32.183*** 28.102*** 40.526*** 31.652***

e 1,2,3,4 > 5 e 1,2 > 3e5; 3 > 4,5; 4 > 5 1,2,3,4 > 5 1,2,3,4 > 5 1,2,3,4 > 5 1,2,3 > 4,5; 4 > 5 1,2,3,4 > 5

Predictors Instrumental attitude Affective attitude Subjective norm Perceived behavioural control Planning when Planning where Planning what kind of exercise Planning how often Planning with whom to exercise ***

p < .001;

**

p < .01; *p < .05.

(.60) (.87) (.63) (.99) (1.83) (1.75) (1.64) (1.79) (1.84)

(.70) (.97) (.97) (.80) (1.59) (1.25) (1.13) (.97) (1.29)

(.70) (.89) (.72) (.97) (1.58) (1.37) (1.25) (1.14) (1.57)

(.65) (.94) (1.08) (1.01) (1.63) (1.74) (1.50) (1.67) (1.76)

(.87) (1.47) (1.25) (.93) (1.80) (2.02) (2.01) (1.84) (1.79)

112

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114

strength in order to better detail relevant target groups for exercise behaviour change interventions. The inclusion of past exercise behaviour and exercise habit strength was motivated by the demonstrated relevance of these variables in previous research on exercise action control (De Bruijn et al., 2009; De Bruijn & Rhodes, in press; Rhodes et al., 2010, 2008; Rhodes & Plotnikoff, 2006). However, no research to date has simultaneously studied past exercise behaviour and exercise habit strength. Habit strength acknowledges various additional characteristics apart from past behaviour (Aarts et al., 1998; Eagly & Chaiken, 1993; Verplanken & Aarts, 1999; Verplanken & Orbell, 2003), and thus an exploration of whether exercise action control profiles can be better detailed by incorporating both past exercise behaviour and exercise habit strength appeared prudent. Regarding the initial profiling where intention and subsequent exercise behaviour were taken into account, results were in line with earlier research on exercise action control (Rhodes et al., 2003; Rhodes et al., 2010; Rhodes & Plotnikoff, 2006; Rhodes et al., 2008). First, only a very small part of the sample could be categorized as a disinclined actor (acting without positive intention), while a fairly large proportion of the sample had positive exercise intentions, but failed to act in accordance with those intentions. Thus, the distributions of intentioneexercise profiles found in the present and other studies indicate that intention to exercise is needed for exercise behaviour to occur, but is also insufficient for exercise behaviour in a substantial part of the population. When the initial profiles were further extended by previous exercise behaviour, the distributions revealed that about half of the sample were not previously engaging in sufficient exercise, but that about four percent could be categorized as exercise adopters (i.e. insufficiently active at T1, but sufficiently active at T2): of those exercise adopters, 85% could be categorized as inclined actors. Again, this indicates that exercise behaviour is virtually nonexistent without positive exercise intentions. Another noteworthy finding from this extension was that some 40% of the sample that were previously sufficiently active failed to act upon their positive exercise intention. Thus, in line with previous evidence(Rhodes & Plotnikoff, 2006; Rhodes et al., 2008), the intentioneexercise gap is prevalent not only amongst exercise adopters, but also amongst those who were previously engaged in sufficient exercise behaviour. In the next step, the initial intentioneexercise profiles were extended with exercise habit strength at T1 and, again, distributions of the profiles revealed various interesting findings. For instance, more than three quarters (149/193) of those with weak exercise habits could be categorized as exercise nonintenders, whereas nearly 90% (119/137) of those with strong exercise habits could be categorized as exercise intenders. Further, the proportion of incline actors was lower in the weak exercise habit group as compared to the strong exercise habit group (22.73% vs. 41.18%), indicating that the intentioneexercise gap is less prevalent when strong exercise habits have been developed. Finally, based on the criticism that past behaviour does not reflect the operation of habits (Ajzen, 2002), profiles were created that took both previous exercise behaviour and exercise habit strength into account. Regarding the distributions of these profiles, results showed that of the 119 strongly habitual exercise intenders, about a third could be categorized as exercise maintainers and about a third reported to have missed an exercise opportunity at either T1 or T2. Importantly, another third of these strongly habitual exercise intenders indicated having engaged in sufficient exercise at neither T1 nor T2. Thus, even those who reported strong exercise habits and intentions apparently miss several exercise occasions. Although it could be argued that this latter finding demonstrates apparent invalidity of the habit index measure, recent research that modelled habit development has indeed

shown that missing a limited amount of exercise opportunities does not lead to significant decreases in the level of exercise automaticity (Lally, Van Jaarsveld, Potts, & Wardle, in press). Notably, whereas several authors have suggested that behavioural patterns that have become (strongly) habitual do not require (much) attention for intervention developers (Verplanken & Wood, 2006), the findings from the present study indicate that even those with strong exercise habits and positive exercise intentions occasionally fail to exercise, indicating that exercise interventions should also continue to focus on those who are motivated exercisers with strong exercise habits. Next to the exploration of the distribution of intentioneexercise profiles and the potentially useful addition of previous exercise behaviour and/or exercise habit strength, the present study also sought to identify relevant predictors of profile membership in order to provide practical information for intervention developers. In line with the formulated hypothesis, PBC was a fairly consistent predictor of exercise action control. That is, inclined actors believed that exercising in the next two weeks was less difficult and more under their control than inclined abstainers. Similar results were also found when exercise habit strength was taken into account, with strongly habitual inclined actors having significantly higher PBC scores than strongly habitual inclined abstainers. Given that in the present study about two thirds of the sample reported having had at least one exercise relapse even though they were strongly habitual exercise intenders, these strongly habitual inclined abstainers could be considered an important exercise intervention target group. The significant PBC differences suggest that this target group should preferably be targeted with persuasive messages that focus on instilling a sense of controllability to exercise in order to prevent these exercise relapses. However, there were no significant differences for PBC scores when the profiles were deconstructed by accounting for previous exercise behaviour. Although significantly higher PBC scores were obtained for successful maintainers as compared to unsuccessful adopters, PBC was not discriminative between strongly habitual successful and unsuccessful maintainers. Thus, while a meta-analysis of the TPB in the physical activity domain (Hagger et al., 2002) has suggested that physical activity interventions should focus on fostering a sense of control, these types of interventions may be misplaced when they are designed to prevent exercise relapses in inclined and strongly habitual maintainers. Likewise, a differentiation of strongly habitual successful and unsuccessful exercise maintainers may also be a viable explanation for the relative lack of success of some exercise interventions targeting PBC or self-efficacy have been relatively unsuccessful in producing exercise changes (Rhodes & Pfaeffli, 2010). Although previous evidence has indicated medium-to-large effect sized correlations between affective attitude and exercise behaviour (Rhodes, Fiala, & Conner, 2009), between affective attitude and exercise habit strength (De Bruijn & Rhodes, in press) and small-tomedium sized differences for affective attitude and exercise action control (De Bruijn et al., 2009; Rhodes et al., 2003), findings did not provide support for the hypothesized role of affective attitude in exercise action control across the created profiles. However, while several earlier studies in undergraduate students (Rhodes et al., 2003) and young adults samples (De Bruijn et al., 2009) have suggested affective attitude to be more related to exercise action control, other research in a population-based adult sample has also indicated that affective attitude was unrelated to exercise action control (Rhodes et al., 2008). The mixed and limited evidence of affect and exercise action control therefore suggests that additional research in more diverse samples is needed to ascertain the potentially important role of affective attitude components in exercise action control. Although a tentative explanation, affective components of attitudes

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114

are also often considered to be more automatic than cognitive components due to the faster accessibility of affect (Verplanken, Hofstee, & Janssen, 1998; Zajonc, 1980). These automatic responses may not be readily captured in self-reported survey data and suggest that more automatic responses in exercise action control are needed to ascertain the role of affect in exercise action control. Indeed, a recent study (Conroy, Hyde, Doerksen, & Ribeiro, 2010) showed that an implicit measure of exercise attitude was an independent predictor of physical activity, even when statistically controlling for explicit measures of intention and PBC. More fundamental social psychological research has shown that goal-directed habits can be implicitly set in motion by positive affect (Custers & Aarts, 2005), and so an exploration of this issue in exercise promotion appears to be a fruitful endeavour. Given the premise that motivated exercise adopters or those who have weak exercise habits could be considered as those aiming to change their current exercise status, it was hypothesized that planning items would be the most discriminant factors for these profiles. There was, however, no evidence that planning items could discriminate important action control profiles. Although plans regarding how often to exercise were discriminative between strongly habitual inclined actors and weakly habitual inclined abstainers, there was no evidence that planning items could discriminate between strongly habitual inclined actors and strongly habitual inclined abstainers. A similar pattern was found when profiles were extended with past exercise behaviour, thereby rejecting the third hypothesis. The present findings thus suggest that implemental planning is unrelated to exercise action control across adopters, but also across maintainers in young adults. However, even though implemental planning is regarded as a key construct in bridging the intentioneexercise gap (Sniehotta, 2009; Sniehotta, Nagy et al., 2006; Sniehotta, Scholz et al., 2005) there is also evidence from exercise determinant and intervention studies that show that self-regulatory planning strategies have at best an only marginally effect on exercise. For instance, in a study amongst adolescents (Araújo-Soares et al., 2009) no main effects of action planning were found. Similarly, Rise, Thompson, and Verplanken (2003) found that implemental planning items did not significantly increase the amount of explained variance in exercise behaviour amongst college students, while Sniehotta and colleagues (Sniehotta, Scholz et al., 2005) found non-significant effects of exercise action plans on exercise behaviour two months later in a clinical sample. Finally, in an exercise intervention study, De Vet and colleagues (De Vet, Oenema, Sheeran, & Brug, 2009) found non-significant differences in changes in physical activity between those who formulated implementation intentions and those who did not. Given this mixed evidence, careful consideration should be given before self-regulatory planning is applied in exercise promotion interventions. This consideration may involve the extension of intentioneexercise profiles with measures of past exercise behaviour (Rhodes et al., 2008) and exercise habit strength (De Bruijn & Rhodes, in press; Rhodes et al., 2010), but also the identification of relevant critical situations in which personally relevant cues are more likely to impede on behavioural enactment (Adriaanse, De Ridder, & De Wit, 2009). The results of the present study should be viewed in terms of several limitations. The most general concern is the use of a convenience sample of undergraduate students, which prohibits the generalization of findings of this study to other populations or the population at large. Likewise, a fairly high proportion of participants dropped out after the initial assessment. Although attrition analysis revealed non-significant differences on demographics and study variables, unknown variables leading to this dropout may have biased findings. Another limitation of this study was the use of self-administered survey questionnaires to assess

113

exercise behaviour. Even though doubly labelled water techniques have demonstrated good agreement between the IPAQ and these techniques, there is generally an underestimation of questionnairederived energy expenditure at higher intensity levels of activity (Maddison et al., 2007). A final limitation relates the time span between intention, planning and subsequent exercise behaviour. Although the TPB favours shorter time periods for more accurate prediction (Ajzen, 1991; Ajzen & Fishbein, 1980), experimental research on implemental planning often have shown more pronounced effects on behavioural change across longer time periods (Sniehotta, Nagy et al., 2006; Sniehotta, Scholz et al., 2005) than at shorter intervals (De Nooijer, De Vet, Brug, & de Vries, 2006; Kellar & Abraham, 2005). Thus, research covering longer time spans should be employed to ascertain the role of implemental planning in exercise action control. References Aadahl, M., von Huth Smith, L., Pisinger, C., Toft, U. N., Glumer, C., Borch-Johnsen, K., et al. (2009). Five-year change in physical activity is associated with changes in cardiovascular disease risk factors: the Inter99 study. Preventive Medicine, 48(4), 326e331. Aarts, H., Paulussen, T., & Schaalma, H. (1997). Physical exercise habit: on the conceptualization and formation of habitual health behaviours. Health Education Research, 12(3), 363e374. Aarts, H., Verplanken, B., & van Knippenberg, A. (1998). Predicting behavior from actions in the past: repeated decision making or a matter of habit? Journal of Applied Social Psychology, 28(15), 1355e1374. Adriaanse, M. A., De Ridder, D. T. D., & De Wit, J. B. F. (2009). Finding the critical cue: implementation intentions to change one’s diet work best when tailored to personally relevant reasons for unhealthy eating. Personality and Social Psychology Bulletin, 35(1), 60e71. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179e211. Ajzen, I. (2002). Residual effects of past on later behaviour: habituation and reasoned action perspectives. Personality and Social Psychology Review, 6(2), 107e122. Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall. Araújo-Soares, V., McIntyre, T., & Sniehotta, F. F. (2009). Predicting changes in physical activity among adolescents: the role of self-efficacy, intention, action planning and coping planning. Health Education Research, 24(1), 128e139. Bandura, A. (1998). Health promotion from the perspective of social cognitive theory. Psychology and Health, 13, 623e649. Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, efficiency, intention, and control in social cognition (2nd ed.). Hillsdale, NJ: Erlbaum. Brug, J., Oenema, A., & Ferreira, A. (2005). Theory, evidence and intervention mapping to improve behavior nutrition and physical activity interventions. International Journal of Behavioral Nutrition and Physical Activity, 2(2). Chatzisarantis, N. L. D., & Hagger, M. S. (2007). Mindfulness and the intentionebehavior relationship within the theory of planned behavior. Personality and Social Psychology Bulletin, 33(5), 663e676. doi:10.1177/0146167206297401. Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155e159. Conroy, D., Hyde, A., Doerksen, S., & Ribeiro, N. (2010). Implicit attitudes and explicit motivation prospectively predict physical activity. Annals of Behavioral Medicine, 39(2), 112e118. doi:10.1007/s12160-010-9161-0. Custers, R., & Aarts, H. (2005). Positive affect as implicit motivator: on the nonconscious operation of behavioral goals. Journal of Personality and Social Psychology, 89(2), 129e142. De Bruijn, G. J. (2010). Understanding college students’ fruit consumption: integrating habit strength in the theory of planned behaviour. Appetite, 54, 16e22. De Bruijn, G. J., De Groot, R., Van den Putte, B., & Rhodes, R. E. (2009). Conscientiousness, extroversion, and action control: comparing moderate and vigorous physical activity. Journal of Sport and Exercise Psychology, 31(6), 724e742. De Bruijn, G. J., & Gardner, B. Active commuting and habit strength: an interactive and discriminant analyses approach. American Journal of Health Promotion, in press. De Bruijn, G. J., & Rhodes, R. E. Exercise, intention and habit strength relationships. Scandinavian Journal of Medicine and Science in Sports, in press. De Nooijer, J., De Vet, E., Brug, J., & de Vries, N. K. (2006). Do implementation intentions help to turn good intentions into higher fruit intakes? Journal of Nutrition Education and Behavior, 38(1), 25e29. De Vet, E., Oenema, A., Sheeran, P., & Brug, J. (2009). Should implementation intention interventions be implemented in obesity prevention: the impact of ifthen plans on daily physical activity in Dutch adults. International Journal of Behavioral Nutrition and Physical Activity, 6, 11. Eagly, A., & Chaiken, S. (1993). The psychology of attitudes. Fort Worth, TX. Erdfelder, E., Faul, F., & Buchner, A. (1996). GPower: a general power analysis program. Behavior Research Methods, Instruments, & Computers, 28, 1e11. Fishbein, M., & Cappella, J. N. (2006). The role of theory in developing effective health communications. Journal of Communication, 56, s1es17.

114

G.-J. de Bruijn / Psychology of Sport and Exercise 12 (2011) 106e114

Godin, G. (1994). Theories of reasoned action and planned behavior: usefulness for exercise promotion. Medicine and Science in Sports and Exercise, 26(11), 1391e1394. Gollwitzer, P. M. (1999). Implementation intentions: strong effects of simple plans. American Psychologist, 54(7), 493e503. Gollwitzer, P. M., & Schaal, B. (1998). Meta-cognition in action: the importance of implementation intentions. Personality and Social Psychology Review, 2, 124e136. Gollwitzer, P., & Sheeran, P. (2006). Implementation intentions and goal achievement: a meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69e119. Hagger, M. S., Chatzisarantis, N., & Biddle, S. J. (2001). The influence of self-efficacy and past behaviour on the physical activity intentions of young people. Journal of Sports Sciences, 19(9), 711e725. 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(1), 3e32. Hagstromer, M., Oja, P., & Sjostrom, M. (2006). The international physical activity questionnaire (IPAQ): a study of concurrent and construct validity. Public Health Nutrition, 9(6), 755e762. Haskell, W. L., Lee, I. M., Pate, R. R., Powell, K. E., Blair, S. N., Franklin, B. A., et al. (2007). Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Medicine and Science in Sports and Excercise, 39(8), 1423e1434. Hausenblas, H. A., Carron, A. V., Mack, D. E., & Godin, G. (1997). Application of the theories of reasoned action and planned behavior to exercise behavior: a metaanalysis. Journal of Sport and Exercise Psychology, 19(1), 36e51. Kellar, I., & Abraham, C. (2005). Randomized controlled trial of a brief researchbased intervention promoting fruit and vegetable consumption. British Journal of Health Psychology, 10(4), 543e558. Kromhout, D., Bloemberg, B., Seidell, J. C., Nissinen, A., & Menotti, A. (2001). Physical activity and dietary fiber determine population body fat levels: the seven country studies. International Journal of Obesity and Related Metabolic Disorders, 25, 301e306. Lally, P., Van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. How are habits formed: modelling habit formation in the real world. European Journal of Social Psychology, in press. Lippke, S., Ziegelmann, J. P., & Schwarzer, R. (2004). Behavioral intentions and action plans promote physical exercise: a longitudinal study with orthopedic rehabilitation patients. Journal of Sport and Exercise Psychology, 26, 470e483. Maddison, R., Ni Mhurchu, C., Jiang, Y., Vander Hoorn, S., Rodgers, A., Lawes, C., et al. (2007). International physical activity questionnaire (IPAQ) and New Zealand physical activity questionnaire (NZPAQ): a doubly labelled water validation. International Journal of Behavioral Nutrition and Physical Activity, 4(1), 62. Norman, P., & Conner, M. (2005). The theory of planned behavior and exercise: evidence for the mediating and moderating roles of planning on intentionebehavior relationships. Journal of Sport and Exercise Psychology, 27, 488e504. Rhodes, R. E., De Bruijn, G. J., & Matheson, D. H. (2010). Habit in the physical activity domain: integration with intention temporal stability and action control. Journal of Sport and Exercise Psychology, 32, 84e98. Rhodes, R. E., & Courneya, K. S. (2005). Threshold assessment of attitude, subjective norm, and perceived behavioural control for predicting exercise intention and behaviour. Psychology of Sport and Exercise, 6, 349e361.

Rhodes, R. E., Courneya, K. S., & Jones, L. W. (2003). Translating exercise intentions into behavior: personality and social cognitive correlates. Journal of Health Psychology, 8(4), 447e458. Rhodes, R. E., Fiala, B., & Conner, M. (2009). A review and meta-analysis of affective judgments and physical activity in adult populations. Annals of Behavioral Medicine, 38(3), 180e204. Rhodes, R. E., & Pfaeffli, L. (2010). Mediators of physical activity behaviour change among adult non-clinical populations: a review update. International Journal of Behavioral Nutrition and Physical Activity, 7(1), 37. Rhodes, R. E., & Plotnikoff, R. C. (2006). Understanding action control: predicting physical activity intentionebehavior profiles across six months in a Canadian sample. Health Psychology, 25(3), 292e299. Rhodes, R. E., Plotnikoff, R. C., & Courneya, K. S. (2008). Predicting the physical activity intentionebehaviour profiles of adopters and maintainers using three social cognition models. Annals of Behavioral Medicine, 36(3), 244e252. Rise, J., Thompson, M., & Verplanken, B. (2003). Measuring implementation intentions in the context of the theory of planned behavior. Scandinavian Journal of Psychology, 44, 87e95. Sheeran, P. (2002). Intentionebehavior relations: a conceptual and empirical review. European Review of Social Psychology, 12, 1e36. Sniehotta, F. F. (2009). Towards a theory of intentional behaviour change: plans, planning, and self-regulation. British Journal of Health Psychology, 14, 261e273. Sniehotta, F. F., Nagy, G., Scholz, U., & Schwarzer, R. (2006). The role of action control in implementing intentions during the first week of behaviour change. British Journal of Social Psychology, 45, 87e106. Sniehotta, F. F., Scholz, U., & Schwarzer, R. (2005). Bridging the intentionebehaviour gap: planning, self-efficacy, and action control in the adoption and maintenance of physical exercise. Psychology & Health, 20(2), 143e160. Sniehotta, F. F., Scholz, U., & Schwarzer, R. (2006). Action plans and coping plans for physical exercise: a longitudinal intervention study in cardiac rehabilitation. British Journal of Social Psychology, 11, 23e37. Sniehotta, F. F., Schwarzer, R., Scholz, U., & Schuz, B. (2005). Action planning and coping planning for long-term lifestyle change: theory and assessment. European Journal of Social Psychology, 35, 565e576. Triandis, H. (1977). Interpersonal behavior. Montery, CA: Brooks/Cole. Van den Putte, B., & Dhondt, G. (2005). Developing successful communication strategies: a test of an integrated framework for effective communication. Journal of Applied Social Psychology, 35(11), 2399e2420. Verplanken, B., & Aarts, H. (1999). Habit, attitude and planned behaviour: is habit an empty construct or an interesting case of goal-directed automaticity? European Review of Social Psychology, 10, 101e134. Verplanken, B., Hofstee, G., & Janssen, H. J. W. (1998). Accessibility of affective versus cognitive components of attitudes. European Journal of Social Psychology, 28, 23e35. Verplanken, B., & Orbell, S. (2003). Reflections of past behavior: a self-report index of habit strength. Journal of Applied Social Psychology, 33(6), 1313e1330. Verplanken, B., & Wood, W. (2006). Interventions to break and create consumer habits. Journal of Public Policy and Marketing, 25(1), 90e103. Wiedemann, A. U., Schuz, B., Sniehotta, F. F., Scholz, U., & Schwarzer, R. (2009). Disentangling the relationship between intentions, planning, and behaviour: a moderated mediation analysis. Psychology and Health, 24(1), 67e79. Zajonc, R. (1980). Feeling and thinking: preferences need no inferences. American Psychologist, 35, 151e175.