Threshold assessment of attitude, subjective norm, and perceived behavioral control for predicting exercise intention and behavior

Psychology of Sport and Exercise 6 (2005) 349–361 www.elsevier.com/locate/psychsport

Threshold assessment of attitude, subjective norm, and perceived behavioral control for predicting exercise intention and behavior Ryan E. Rhodesa,*, Kerry S. Courneyab a

University of Victoria, School of Physical Education, P.O. Box 3015, STN CSC, Victoria, BC, Canada V8W 3P1 University of Alberta, E-424 Van Vliet Center, Faculty of Physical Education, Edmonton, AB, TG6 2H9, Canada

b

Received 13 October 2003; received in revised form 25 February 2004; accepted 23 April 2004 Available online 17 July 2004

Abstract According to the theory of planned behavior (TPB), attitude, subjective norm, and perceived behavioral control (PBC) are hypothesized to have linear associations with intention and behavior. However, no previous research has examined this hypothesized linearity across scale responses. Further, no study using social cognitive measures has detailed the incremental increases in the proportion of people meeting the American College of Sports Medicine’s (ACSM) exercise guidelines across scale responses. The purpose of this study was to examine mean scores of intention and behavior and detail the proportion of participants meeting ACSM’s exercise guidelines for each TPB construct response category (i.e. 1 – 7). Participants were university undergraduates ðN ¼ 585Þ who completed measures of the TPB and a 2-week follow-up of exercise behavior. Results were evaluated using effect sizes ðd; wÞ and p-levels, and provided general support for the linear effects of affective and instrumental attitude, but PBC and subjective norm were identified as having specific thresholds. Further, thresholds of positive, but not negative, exercise social cognition were identified for meeting ACSM’s criteria with the exception of affective attitude. Threshold analysis was discussed as a novel way of analyzing TPB data by providing additional information about the expected success of intervention efforts focused on TPB constructs. q 2004 Elsevier Ltd. All rights reserved. Keywords: Theory of planned behavior; Exercise prescription

There is substantial evidence that physical inactivity is associated with the development of several chronic diseases and premature mortality (Booth, Gordon, Carlson, & Hamilton, 2000; Katzmarzyk, Gledhill, & Shephard, 2000; Blair & Brodney, 1999). There is also extensive literature indicating that physical activity is an effective preventive strategy against cardiovascular disease, obesity, stroke, * Corresponding author. Tel.: þ1-250-721-8384; fax: þ1-250-721-6601. E-mail address: [email protected] (R.E. Rhodes). 1469-0292/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.psychsport.2004.04.002

350

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

hypertension, type 2 diabetes, colon cancer, breast cancer, osteoporosis, and several psychological disorders (Bouchard & Shephard, 1994; Blair & Brodney, 1999; US Department of Health and Human Services, 1996). Despite this information, the majority of adults do not meet the minimal requirements for physical activity wherein health benefits are thought to occur (Canadian Fitness and Lifestyle Research Institute, 2001; US Department of Health and Human Services, 1996). Thus, there is a need to understand the determinants of physical activity behavior in order to develop appropriate intervention strategies. One leading theoretical model explaining informational and motivational influences on behavior is the theory of planned behavior (TPB; Ajzen, 1985, 1991). Further, empirical reviews of the TPB have supported a relationship for the prediction of many disparate health behaviors including exercise and physical activity (Godin & Kok, 1996; Hagger, Chatzisarantis, & Biddle, 2002; Hausenblas, Carron, & Mack, 1997). The TPB suggests that the proximal determinant of volitional behavior is one’s intention to engage in that behavior. Attitudes and subjective norm are hypothesized to influence behavior through intentions. Attitude is the affective and instrumental evaluations of performing the behavior by the individual. Subjective norm is the social pressure on the individual to perform or not to perform a particular behavior. The TPB attempts to also predict behaviors that are not completely volitional by incorporating perceptions of control over performance of the behavior as an additional predictor of intention and behavior (Ajzen, 1991). Perceived behavioral control (PBC) is the perceived ease or difficulty of performing the behavior, and takes into account the individual’s perception of skills, resources, and opportunities needed to perform a behavior (Ajzen, 1991). The TPB is hypothesized to act as a linear model (Ajzen, 1991). Accordingly, the greater one’s attitude, subjective norm, or PBC, the greater one’s resulting intention and behavior should be. Certainly evidence of this linear effect has been repeatedly demonstrated with the use of bivariate correlations, ordinary least squares regression, and structural equation models (Armitage & Conner, 2001; Hagger et al., 2002). However, no research has actually examined this hypothesized linearity across scale responses. Thus, although general linearity has been supported through correlation analyses, it is possible that critical thresholds are responsible for linear effects previously reported (Fishbein, Von Haeften, & Appleyard, 2001). Thresholds between particular scale metrics may be of import, as they indicate a potential dose-response relationship between social cognition and exercise behavior. For example it may be that a difference between ‘slight’ PBC and ‘moderate’ PBC results in a strong effect on subsequent exercise behavior but no differences exist between moderate PBC and “strong” PBC. Although this still partially supports a linear model because differences between slight and moderate PBC are meaningful, the effect of PBC on behavior would be at a ceiling threshold once moderate PBC is obtained rather than linear across the PBC scale to strong levels. Testing these theoretical assumptions of linearity in the TPB is warranted for a better understanding of exercise motivation, but the practical value of threshold analysis may be of even more interest. First, linearity or critical thresholds for the effects of TPB constructs has direct application in intervention efforts. Improving physical activity interventions and understanding the social cognitive mediators of change is an utmost concern (Baranowski, Anderson, & Carmack, 1998; Sallis, 2001). If effects of the TPB on intention and behavior are indeed linear across a scale, then intervention efforts should continue to focus on improving TPB constructs to a maximum value. However, if threshold ceilings are identified, it suggests that interventions need only improve TPB constructs to a certain point. For example, if a meaningful subjective norm threshold for improvements in intention is identified between slight and moderate but not between moderate and strong, then targeting subjective norm to moderate represents

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

351

the minimum value for the optimal behavioral outcome. This resembles the current focus on dose-response effects of physical activity upon health outcomes (e.g. Bouchard, 2001; Dunn, Madhukar, & O’Neal, 2001; Spirduso & Cronin, 2001) wherein minimal doses for optimal outcomes are sought. Thus, identifying the extent of linearity in TPB scales helps understand how to optimally focus interventions. Second, detailing TPB effects on the American College of Sports Medicine’s (American College of Sports Medicine, 2000) recommended exercise guidelines examines each scale’s relative value for achieving this guideline. This moves away from the more abstract analysis of theoretical linearity and instead provides a threshold analysis of the TPB using a clinically important exercise behavior outcome (American College of Sports Medicine, 2000). The direct application for exercise prescription is apparent. For example, it may be that participants who respond with perceiving affective evaluations towards exercise as slight result in no significant difference in meeting ACSM’s exercise behavior guidelines from those perceiving affective evaluations as exercise as moderate or extreme. This would suggest that raising a client’s affective attitude rating of exercise to slight will provide the recommended dose of regular exercise (Health Canada, 1998; American College of Sports Medicine, 2000). In contrast, we may find values are much higher, suggesting that higher ratings of various social cognitive constructs are more likely to produce recommended exercise levels. Finally, item distribution may require a more detailed analysis. Ajzen (2002) argues for the importance of measurement variability in the TPB and Fishbein et al. (2001) argue similarly for the theory of reasoned action but scant research into item distribution has been published for either model. Previous research reports basic descriptives (mean and standard deviations) and social desirability (Armitage & Conner, 1999), but no research has completely detailed the proportions of participants responding at each scale metric. Thus, it is possible that more sensitive scales are required to measure TPB constructs. Currently, no study using the TPB in any behavioral domain has detailed linearity across scale responses. Further, no study in the exercise domain has examined potential thresholds in social cognitive constructs using American College of Sports Medicine (2000) recommended exercise guidelines. Therefore, the purpose of this study was twofold. First, we wished to detail the mean scores of intention and behavior for each TPB construct response category to discriminate potential linear thresholds in exercise intention and behavior that are not identified using a standard correlation analysis. Second, we wished to detail the proportion of participants meeting ACSM’s exercise guidelines for each TPB construct response category to examine whether any thresholds occur across scales. Our hypotheses followed the assumption of linearity in the TPB, but all results were considered somewhat exploratory given the novelty of this research question.

Method Participants and procedure Participants were undergraduate students taking part in the study for extra credit in their introductory psychology course. The sample consisted of 585 (426 female and 159 male) participants who attended large group sessions during January and February 2000, and January and February 2001 completing selfreport measures of the TPB at their own pace. These participants also completed a 2-week follow-up measure of exercise behavior. The mean age of participants was 20.07 (SD ¼ 4.00 yr) and the mean year

352

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

in university for the sample was 1.66 (SD ¼ 0.96). This sample consists of the aggregation of two previously published data sets used to examine personality and the theory of planned behavior (Rhodes, Courneya, & Hayduk, 2002; Rhodes, Courneya, & Jones, 2004). Instruments Regular exercise was defined for all participants as activities performed at a vigorous intensity three or more times per week for at least 30 min each time (American College of Sports Medicine, 2000). Participants were asked to use this definition when answering all exercise-related questions. Exercise attitude was measured using 7-point bipolar adjective scales as suggested by Ajzen (2002). Three items were utilized to tap the instrumental attitude component (useful – useless, wise– foolish, and beneficial – harmful) and three items were used to tap the affective attitude component (enjoyable– unenjoyable, interesting– boring, relaxing–stressful). Our previous research (e.g. Rhodes & Courneya, 2003a,b,c; Rhodes, Courneya, & Jones, 2003) and the research of others (e.g. Crites, Fabrigar, & Petty, 1994; Edwards, 1990; Edwards & Von Hippel, 1995; Lowe, Eves, & Carroll, 2002) has consistently identified the construct distinction of these attitudinal components. Thus, affective and instrumental attitude were modeled as distinct constructs. The statement that preceded the adjectives was ‘For me, exercising regularly over the next 2 weeks would be…’ Internal consistency was a ¼ 0:80 for instrumental attitude and a ¼ 0:74 for affective attitude. Subjective norm was measured by 7-point scale items that ranged from 1 (strongly disagree) to 7 (strongly agree). The two items that measured subjective norm were: (1) ‘Most people in my social network want me to exercise regularly in the next 2 weeks’, and (2) ‘Most people in my social network would approve if I exercised regularly in the next 2 weeks’. Internal consistency between the items was a ¼ 0:70: These items are based on previous TPB research with undergraduate students (e.g. Courneya, Bobick, & Schinke, 1999) and Ajzen (2002). Perceived behavioral control was measured by the following three items similar to those suggested by Ajzen (2002): (1) ‘How confident are you over the next 2 weeks that you could exercise regularly if you wanted to do so’; on a 7-point scale ranging from 1 (very unconfident) to 7 (very confident), (2) ‘How much personal control do you feel you have over exercising regularly in the next 2 weeks’; on a 7-point scale from 1 (very little control) to 7 (complete control), and (3) How much I exercise regularly over the next 2 weeks is completely up to me; on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). Internal consistency for the aggregated scale was a ¼ 0:78: Exercise intention was assessed by the following item: ‘Over the next 2 weeks, I intend to exercise-times per week’ rated on an open scale (Courneya, 1994) This item has indicated the best predictive validity with exercise behavior in comparison to other intention items (Courneya, 1994). Further, using this item in the present study allowed for meaningful interpretation (i.e. each unit represents an additional bout of intended exercise). Exercise behavior was measured by the Godin Leisure Time Exercise Questionnaire (Godin, Jobin, & Bouillon, 1986; Godin & Shephard, 1985). The instrument contains three open-ended questions covering the frequency of mild (e.g. easy walking), moderate (e.g. fast walking), and strenuous (e.g. jogging) exercise completed during free time. Duration of these intensities was set for at least 30 min. The GLTEQ has demonstrated a one month test – retest reliability of 0.62 and concurrent validity coefficients of 0.32 with an objective activity indicator (CALTRAC accelerometer), 0.56 with VO2max (as measured by expired gases), and 2 0.43 with % body fat (as measured by hydrostatic weighing).

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

353

Table 1 Correlations for the theory of planned behavior and a two-week follow-up of exercise behavior ðN ¼ 585)

Affective Attitude Instrumental attitude Subjective norm Perceived behavioral control Intention Exercise behavior

2

3

4

5

6

M

SD

0.50**

0.16** 0.35**

0.19** 0.17** 0.19**

0.41** 0.33** 0.23** 0.23**

0.34** 0.24** 0.19** 0.24** 0.63**

4.87 5.98 5.51 5.71 3.14 2.19

1.19 0.93 1.23 1.28 1.65 1.85

**P , 0:01:

These levels of reliability and validity compared very favorably to nine other self-report measures of exercise that were examined (Jacobs, Ainsworth, Hartman, & Leon, 1993). Mild and moderate exercises were not included as indicators of exercise behavior due to their incongruence with our definition of regular exercise for young adults.

Results Means, standard deviations, and bivariate correlations are provided in Table 1. Mean scores varied from 4.78 (i.e. affective attitude) to 5.98 (i.e. instrumental attitude) for TPB constructs. All correlations were statistically significant ðP , 01Þ and varied from 0.16 (i.e. affective attitude and subjective norm) to 0.63 (i.e. intention and behavior). For the threshold assessment, scales of TPB constructs were categorized (1– 7) by rounding to the nearest response category. Frequencies for all constructs are presented in Table 2. These data were used to evaluate meaningful groups across response categories. A minimum n of 30 was considered as a meaningful response category. This was derived from a power calculation based on Cohen (1988, 1992) and minimum sample size recommended when testing proportions (Glass & Hopkins, 1984). Using an expected maximum of seven potentially meaningful categories (i.e. groups), a medium effect size for our dependent variables of interest ðf ¼ 0:25Þ; a power of 0.80, and an alpha of 0.05, a minimum n per group of 30 is recommended. Thus, at least 5% of the sample was needed to respond in a particular scale value in order for it to be considered meaningful for analysis. Items reflecting affective attitude had the overall largest variability of meaningful groups per response category ranging from 3 to 7 on the 7-point scale. All other TPB constructs ranged from 4 to 7 on Table 2 Proportions of scores for TPB constructs Scores

1

Affective attitude Instrumental attitude Subjective norm PBC

3 0 5 2

2 (0%) (0%) (1%) (0%)

16 2 8 11

(3%) (0%) (1%) (2%)

3

4

5

6

66 (11%) 11 (2%) 21 (4%) 22 (4%)

128 (22%) 37 (6%) 48 (8%) 70 (12%)

172 (29%) 93 (16%) 132 (23%) 106 (18%)

162 232 191 169

7 (28%) (40%) (33%) (29%)

38 210 180 205

(7%) (36%) (31%) (35%)

354

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

the 7-point scales. In order to retain as much data for the main analyses as possible, we aggregated lower response categories. Low scale responses (1– 3) for affective attitude, subjective norm and PBC achieved an n of at least 30 when aggregated and thus were included in the subsequent analyses. Examining means and standard deviations and testing for intention and behavior thresholds among the TPB constructs was achieved by using a series of one-way analysis of variance (ANOVA) tests. Exercise intention and behavior were dependent variables and response category (i.e. meaningful groups of responses) was the independent variable. Due to the number of tests, a more stringent alpha of P , 0:01 was employed to help account for potential experiment-wise error. Finally, significant ANOVAs were followed by post hoc tests to discern which groups were significantly different from one another. Given that differences in group n can dramatically influence p-level (Glass & Hopkins, 1984), Cohen (1988, 1992) effect size d was reported in addition to a traditional Tukey post hoc criterion. Cohen (1988, 1992) suggests d , 0:20 is trivial. Results of this analysis for predicting exercise intention behavior are presented in Table 3. Results of effect size d; point-biserial correlations, and p-level by increment are subsequently reported in Table 4 for intention and for behavior. All TPB constructs displayed significant ðP , 0:01Þ F-tests on intention and behavior. However, post hoc analysis suggested threshold differences among TPB constructs. Post hoc analysis of affective attitude identified that mean intention and behavior differences were present ðd . 0:19Þ starting at the scale floor (i.e. low affective attitude) to the scale ceiling (extremely positive affective attitude). For instrumental attitude, post hoc analysis identified that mean differences were present ðd . 0:19Þ for 5 – 6, and 6 – 7 on intention and from the scale midpoint (4) to the scale ceiling (7) on behavior. The analysis of subjective norm suggested only one mean difference present ðd . 0:19Þ on intention and behavior. Specifically, only differences between slight levels of subjective norm and moderate levels of subjective norm resulted in mean differences in intention and behavior. Post hoc analysis for PBC found Table 3 Mean differences of TPB scores on intention and behavior Mean (SD) Scores

1– 3

F 4

Intention Affective 2.10 (1.40) 2.73 (1.48) attitude Instrumental 2.22 (1.81) attitude Subjective norm 2.65 (1.84) 2.54 (1.75) PBC 2.43 (1.84) 2.57 (1.64) Behavior Affective 1.24 (1.44) 1.77 (1.48) attitude Instrumental 1.12 (1.39) attitude Subjective norm 1.59 (1.58) 1.71 (2.09) PBC 1.17 (1.42) 1.51 (1.62)

5

6

h2

Post hoc

7

3.14 (1.65) 3.69 (1.42) 4.63 (1.64) 26.52* 0.16 (1– 3) , 4 , 5 , 6 , 7 2.42 (1.47) 3.04 (1.51) 3.80 (1.61) 23.54* 0.11 4,5 , 6 , 7 2.71 (1.64) 3.27 (1.55) 3.57 (1.57) 2.79 (1.61) 3.34 (1.46) 3.47 (1.74)

8.31* 0.05 (1– 3),4,5 , 6,7 7.89* 0.05 (1-3) , 5 , 6,7; 4,5 , 6,7

2.18 (1.65) 2.69 (1.42) 3.78 (1.64) 19.06* 0.12 (1– 3) , 4 , 5 , 6 , 7 1.67 (1.67) 2.14 (1.90) 2.68 (1.82) 12.20* 0.06 4 , 5 , 6 , 7 1.88 (1.62) 2.31 (1.75) 2.54 (2.00) 1.87 (2.04) 2.37 (1.74) 2.61 (1.82)

4.58* 0.03 (1– 3),4,5 , 6,7 9.38* 0.05 (1– 3) , 4,5 , 6,7

Note: *P , 0:01; post hoc tests are based on Cohen (1988, 1992) effect size dð. 0:20Þ:

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

355

Table 4 Between increment analysis of TPB constructs on intention and behavior Scores

Intention Affective attitude Instrumental attitude Subjective norm PBC Behavior Affective attitude Instrumental attitude Subjective norm PBC

(1 – 3)– 4

4–5

5–6

6–7

d

r

p-level

d

r

p-level

d

r

p-level

d

r

p-level

0.44

0.19

0.01

2 0.06 0.08

2 0.03 0.04

0.79 0.69

0.26 0.12 0.10 0.14

0.13 0.06 0.05 0.07

0.03 0.51 0.55 0.36

0.36 0.42 0.35 0.37

0.18 0.19 0.17 0.18

0.00 0.00 0.00 0.00

0.61 0.49 0.19 0.08

0.25 0.24 0.10 0.04

0.00 0.00 0.06 0.46

0.36

0.14

0.06

0.07 0.22

0.03 0.10

0.78 0.29

0.26 0.36 0.09 0.19

0.11 0.16 0.04 0.09

0.07 0.08 0.57 0.22

0.33 0.26 0.25 0.26

0.15 0.12 0.12 0.13

0.01 0.04 0.03 0.03

0.71 0.29 0.12 0.13

0.26 0.15 0.06 0.07

0.00 0.00 0.23 0.20

Note: d refers to Cohen (1988, 1992) effect size d: r refers to a point-biserial correlation.

only a difference between slight PBC and moderate PBC on intention but differences between low (1-3) PBC and neutral (4) PBC, and slight (5) PBC and (6) moderate PBC on behavior. Threshold analysis of TPB constructs using American College of Sports Medicine (2000) guidelines as a cut-point was achieved by re-scoring exercise frequency as a dichotomous variable. Thus, those participants who exercised at least three times per week over the 2-week duration were coded as having met American College of Sports Medicine (2000) recommended guidelines. Those individuals who exercised less than three times per week were coded as not meeting ACSM’s guidelines. Differences in the proportion of individuals meeting ACSM’s recommended guidelines by construct response category were subsequently examined using x 2 tests of proportions. Similar to the ANOVA tests in the preceding analysis, an effect size w was reported as a post hoc criterion. Cohen (1988, 1992) suggests w , 0:10 is trivial. Results of this analysis are presented in Table 5. All TPB constructs identified significant ðP , 0:01Þ differences in the proportion of participants meeting ACSM’s criteria across responses. However, post hoc analysis identified different thresholds for various TPB constructs. Specifically, affective attitude had differences ðw . 0:09Þ in the proportion of participants meeting American College of Sports Medicine (2000) guidelines from the scale floor (1 –3) Table 5 Number and percentage meeting ACSM criteria by scale response category Construct

x2

Frequency (%) (1 – 3)

4

5

6

Post hoc

7

Affective attitude 11 (13%) 35 (27%) 66 (38%) 89 (55%) 28 (74%) 68.02* 3,4,5,6,7 Instrumental attitude 7 21 (19%) 84 (23%) 113 (36%) 36.79* (54%) 4,5 , 6 , 7 Subjective norm 9 (26%) 13 (27%) 41 (31%) 85 (45%) 81 (45%) 13.74* (1 – 3),4,5 , 6,7 Perceived behavioral control 7 (20%) 17 (24%) 28 (26%) 70 (41%) 107 (52%) 34.11* (1 – 3),4,5 , 6 , 7 Note: *P , 0:01: post hoc tests are based on Cohen (1988, 1992) effect size wð. 0:10Þ:

356

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

to the scale ceiling (7). Instrumental attitude had differences ðw . 0:09Þ in ACSM criteria proportions for responses of neutral/slightly positive levels (4,5) to moderately positive levels (6), and for moderately positive levels (6) to strong levels (7). Subjective norm had a difference ðw . 0:09Þ in ACSM criteria proportions only for responses of low/neutral/slightly positive levels (1– 3,4,5) to moderately/ strongly positive levels (6,7). Finally, PBC had differences ðw . 0:09Þ in the proportion of participants meeting American College of Sports Medicine (2000) guidelines from responses of low/neutral/ slightly positive levels (1– 3,4,5) to moderately positive levels (6), and from moderately positive levels (6) to strongly positive levels (7).

Discussion Our results suggested that exercise related TPB constructs have different thresholds on intention and behavior as well as American College of Sports Medicine (2000) behavioral guidelines. These findings may have theoretical import for dose-response relations in the TPB and application for the expected success of intervention efforts. Specifically, it is assumed that the inter-individual relationship of a TPB construct and its outcome variable can act as a proxy for understanding potential intra-individual behavior change. This logic mirrors cross-sectional studies of the transtheoretical model and exercise behavior (see Marshall & Biddle, 2001 for a review), where it is assumed that individuals at higher stages of change have moved through the lower stages of behavior change. In threshold analysis, it is assumed that individuals at higher discernable thresholds of intention and behavior based on TPB cognitions have achieved these thresholds through changes (i.e. increases) in TPB cognitions at an earlier juncture. Thus, individuals at higher discernable thresholds of TPB cognitions represent the potential change outcomes in intention and behavior for individuals at lower thresholds of TPB cognitions. Similar to previous research using correlation coefficients, a general trend of linearity was supported across scale responses. Using Cohen’s (1988, 1992) effect size r estimates of small (0.10) medium (0.30) and large (0.50) to help interpret the data, a medium correlation was found for affective attitude and instrumental attitude on exercise intention, and affective attitude on behavior. Small to medium effects were found for instrumental attitude on intention, and subjective norm and PBC on intention and behavior. Overall, this suggests that interventions targeting attitude change may result in more effective behavioral changes than interventions targeting subjective norm or PBC in undergraduate populations. Threshold analysis, however, yielded more specific information on the linear effect present within each construct. Results will be interpreted using Cohen’s (1988, 1992) effect size d suggesting d ¼ 0:20 is a small effect, d ¼ 0:50 is a medium effect, and d ¼ 0:80 is a large effect. Affective attitude followed a linear effect on both intention and behavior. Further, a jump from small effects to a medium effect for the increment between moderately positive affective attitude and extremely positive affective attitude on both intention ðd ¼ 0:61Þ and behavior ðd ¼ 0:71Þ were the strongest effects of all increment transitions for any TPB construct. This suggests that interventions successful in improving affective attitude may continue to improve behavioral outcomes and a particularly powerful behavioral change threshold may be from moderately to extremely positive levels of affective attitude. Instrumental attitude displayed nontrivial ðd . 0:19Þ differences in intention from neutral/slightly positive instrumental attitude to moderately positive instrumental attitude, and from moderately positive

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

357

instrumental attitude to extremely positive instrumental attitude. For the effect of instrumental attitude on behavior, a complete linear effect was identified. Given that behavioral change is the principle outcome (Sutton, 1998, 2002), these results suggest that instrumental attitude supports a linearity hypothesis. This implies that interventions targeting increases in instrumental attitude across the evaluation continuum may result in subsequent increases in behavior and that such interventions should not be abandoned until instrumental attitudes are ‘extremely’ positive. Threshold analysis of the subjective norm construct on intention and behavior suggested only one critical threshold of transition. Specifically, moving from lower levels of subjective norm to moderately positive levels of subjective norm was the only identified nontrivial ðd , 0:19Þ effect on intention and behavior. This result does not support the overall linearity assumption of the TPB, suggesting instead that raising subjective norm to a moderate level is the critical target for the optimal behavioral outcome in the exercise domain. From an intervention perspective, moving from lower levels of subjective norm to moderate levels of subjective norm would appear the key target objective in this potential dose-response relationship and that interventions attempting to induce extremely positive subjective norms are not warranted. The analysis of PBC yielded two thresholds for behavior. Although moving from lower levels of PBC to moderately positive levels of PBC was the only identified nontrivial ðd , 0:19Þ effect for intention, effects from low PBC to neutral levels of PBC, and from neutral/slight levels of PBC to moderately positive levels of PBC were identified on behavior. Like subjective norm, PBC does not appear to support a general linearity assumption. Instead, it appears that moving from low PBC to neutral PBC and from neutral or slightly positive levels of PBC to moderately positive levels of PBC represent the two key cognitive transition thresholds. Further, from an intervention perspective this suggests that reaching moderate levels of PBC will result in the optimal behavioral outcome and that there is very little to be gained by inducing extremely high perceptions of control. Overall, this threshold analysis is also interesting when compared to the more traditional correlational analyses used for the TPB. Correlational analyses test for general linearity across the variability of constructs. However, the threshold analysis suggests that constructs of the TPB have particular thresholds of linearity. The most obvious example of the additional information provided by threshold analysis is apparent when comparing instrumental attitude and PBC relations with exercise behavior. In the bivariate correlations (Table 1), both constructs had effects of 0.24 with behavior. However, the threshold analyses showed that PBC contained two specific linear thresholds on behavior from low PBC to neutral levels of PBC, and from neutral/slight levels of PBC to moderately positive levels of PBC. In contrast, a complete linear effect was identified for instrumental attitude. For the second purpose of this study, we sought to move away from the more abstract analysis of theoretical linearity and instead provide a threshold analysis of the TPB using a commonly accepted exercise behavior criterion (American College of Sports Medicine, 2000). Threshold analysis using American College of Sports Medicine’s (2000) criteria of vigorous exercise behavior at least three times per week for 30 min yielded practical information about the levels of social cognition required to increase the proportion of people meeting this guideline. Affective attitude was the only TPB construct to discriminate proportions of people meeting American College of Sports Medicine’s (2000) guidelines in a complete linear formulation ðw . 0:09Þ: Instrumental attitude and PBC start discriminating ACSM exercise behavior criteria with a threshold of 5 (slightly positive), and follow linear formulation to 7 (extremely/strongly positive). Subjective norm follows an identical pattern to its previous behavioral and intention results with a single key

358

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

discriminating transition at moderate levels of subjective norm. Thus, in general, interventions targeting a change of TPB cognitions of instrumental attitude, subjective norm, and PBC to at least moderately positive levels may result in a greater proportion of individuals meeting American College of Sports Medicine’s (2000) criteria than targeting any lower level. Further, interventions changing TPB cognitions to extreme/strong positive levels should result in the optimal proportion of individuals meeting ACSM’s exercise behavior criteria with the exception of subjective norm. Of interest, scores of 7 on affective attitude had the highest ðP , 0:01Þ proportion of individuals meeting American College of Sports Medicine’s (2000) guidelines in comparison to all other constructs. When considering threshold effects based on ACSM behavioral guidelines, no prior research has examined which TPB construct provides the largest proportion of people meeting these guidelines when scale responses are at a maximum (i.e. 7). Both of these analyses allude to the importance of TPB scale precision at the higher end of the 7-point scales. These analyses also suggest that 5-point scales ranging from ‘strongly/extremely disagree’ to ‘strongly/extremely agree’ may not be suitable for measuring TPB constructs in the exercise domain. The loss of precision may reduce the meaningful groupings of respondents for all TPB constructs. Therefore, we recommend the use of at least 7-point scales when measuring TPB constructs and possibly even 9-point scales to provide additional discrimination at the high end. Despite the interesting theoretical and potential practical findings of this study, some key limitations warrant mention. First, the present study uses a convenience sample of university undergraduates and selfreported exercise behavior. Future research needs to examine threshold findings with diverse samples of multiple age groups and by using more objective measures of exercise (e.g. attendance to a fitness facility) for cross-validation. This may be particularly important for threshold analyses, since dose-response analysis typically includes objective measures of the dependent variable. Second, TPB scale thresholds may vary dramatically by health behavior and we encourage theorists using the TPB to understand other health behaviors to replicate this analysis. Third, the categorization of response categories (1 –7) for each TPB construct results in a loss of actual precision. Still, we feel adherence to multiple item aggregation of TPB constructs (i.e. classical test score theory) is beneficial at this early stage of study for controlling random item response error. Fourth, this is a relatively short prospective observational design. Though we use the results of this study to speculate about the application of these findings to TPB intervention research, actual experimental research is most certainly required to validate these speculations. Fifth, we were forced to aggregate the low end of the TPB scales in order to create an adequate sample size for our analyses. As a result, we cannot comment on precise thresholds for a small number of people who respond at the lower end of the TPB scales. Sixth, the single item intention measure may not be as reliable as a multi-item measure. Finally, the study may have been limited by scale correspondence. Attitude, subjective norm, and PBC were measured with a fixed exercise frequency (i.e. three or more times per week), while intention and behavior were measured using open frequencies. Research by Courneya (1994) has demonstrated that achieving scale correspondence results in optimal predictive results. Still, this violation of scale correspondence occurs in almost all TPB studies in health behavior due to the repeated nature of health behavior and the fixed scaling of TPB constructs (Courneya, 1994). Further, this scale correspondence issue is not relevant for the ACSM criteria analysis, as this criteria was identical to the fixed TPB constructs. In summary, TPB constructs are hypothesized to have linear associations with intention and behavior. However, no previous research had actually examined linearity across scale responses. Further, no study using social cognitive measures has detailed incremental increases in the proportion of people meeting

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

359

the American College of Sports American College of Sports Medicine’s (2000) exercise guidelines. Results of this study demonstrated that threshold analysis can provide additional information about the linear relationships between TPB constructs and exercise behavior. TPB constructs were identified as having specific thresholds. Further, only affective attitude discriminated the proportion of people meeting American College of Sports Medicine’s (2000) guidelines across all of its response categories. Thresholds of positive response categories, but not negative response categories, were identified for meeting ACSM’s criteria for the TPB constructs of instrumental attitude, subjective norm, and PBC.

Acknowledgements Ryan E. Rhodes is supported by a scholar award from the Michael Smith Foundation for Health Research and with funds from the Canadian Foundation for Innovation, the British Columbia Knowledge and Development Fund, and internal grants from the University of Victoria. Kerry S. Courneya is supported by an Investigator Award from the Canadian Institutes of Health Research (CIHR) and a Research Team Grant from the National Cancer Institute of Canada (NCIC) with funds from the Canadian Cancer Society (CCS) and the CCS/NCIC Sociobehavioral Cancer Research Network.

References Ajzen, I. (1985). From intention to actions: a theory of planned behavior. In J. Kuhl, & J. Beckmann (Eds.), Action-control: From cognition to behavior (pp. 11 – 39). Heidelberg: Springer. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179–211. Ajzen, I (2002). Construction of a standard questionnaire for the theory of planned behavior. Retrieved February 15, 2003, from http://Www-Unix.Oit.Umass.Edu/,Aizen/. American College of Sports Medicine (2000). Guidelines for exercise testing and prescription. Baltimore, MD: Lippincott, Williams & Wilkins. Armitage, C. J., & Conner, M. (1999). Predictive validity of the theory of planned behavior: the role of questionnaire format and social desirability. Journal of Community and Applied Social Psychology, 9, 261– 272. Armitage, C. J., & Conner, M. (2001). Efficacy of the theory of planned behavior: a meta-analytic review. British Journal of Social Psychology, 40, 471– 499. Baranowski, T., Anderson, C., & Carmack, C. (1998). Mediating variable framework in physical activity interventions: how are we doing? How might we do better? American Journal of Preventive Medicine, 15, 266– 297. Blair, S. N., & Brodney, S. (1999). Effects of physical inactivity and obesity on morbidity and mortality: current evidence and research issues. Medicine and Science in Sports and Exercise, 31, S646– S662. Booth, F. W., Gordon, S. E., Carlson, C. J., & Hamilton, M. T. (2000). Waging war on modern chronic diseases: primary prevention through exercise biology. Journal of Applied Physiology, 88, 774– 787. Bouchard, C. (2001). Physical activity and health: introduction to the dose-response symposium. Medicine and Science in Sports and Exercise, 33, S347 – S350. Bouchard, C., & Shephard, R. J. (1994). Physical activity fitness and health: the model and key concepts. In C. Bouchard, R. J. Shephard, & T. Stephens (Eds.), Physical activity fitness and health: International Proceedings and Consensus Statement (pp. 77 – 88). Champaign, IL: Human Kinetics. Canadian Fitness and Lifestyle Research Institute (2001). Physical activity monitor. Ottawa: Canadian Fitness and Lifestyle Research Institute. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed). Hillsdale, NJ: Lawrence Erlbaum. Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155– 159.

360

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

Courneya, K. S. (1994). Predicting repeated behavior from intention: the issue of scale correspondence. Journal of Applied Social Psychology, 24, 580– 594. Courneya, K. S., Bobick, T. M., & Schinke, R. J. (1999). Does the theory of planned behavior mediate the relation between personality and exercise behavior? Basic and Applied Social Psychology, 21, 317– 324. Crites, S. L., Fabrigar, L. R., & Petty, R. E. (1994). Measuring the affective and cognitive properties of attitudes: conceptual and methodological issues. Personality and Social Psychology Bulletin, 20, 619–634. Dunn, A. L., Madhukar, H. T., & O’Neal, H. A. (2001). Physical activity dose-response effects on outcomes of depression and anxiety. Medicine and Science in Sports and Exercise, 33, S587– S597. Edwards, K. (1990). The interplay of affect and cognition in attitude formation and change. Journal of Personality and Social Psychology, 59, 202– 216. Edwards, K., & Von Hippel, W. (1995). Hearts and minds: the priority of affective versus cognitive factors in person perception. Personality and Social Psychology Bulletin, 21, 996– 1011. Fishbein, M., Von Haeften, I., & Appleyard, J. (2001). The role of theory in developing effective interventions: implications from project safer. Psychology, Health, and Medicine, 6, 223– 238. Glass, G. V., & Hopkins, K. D. (1984). Statistical methods in education and psychology (2nd ed). Massachusetts: Allyn and Bacon: Needham Heights. Godin, G., & Kok, G. (1996). The theory of planned behavior: a review of its applications to health-related behaviors. American Journal of Health Promotion, 11, 87 – 98. Godin, G., & Shephard, R. J. (1985). A simple method to assess exercise behavior in the community. Canadian Journal of Applied Sport Science, 10, 141–146. Godin, G., Jobin, J., & Bouillon, J. (1986). Assessment of leisure time exercise behavior by self-report: a concurrent validity study. Canadian Journal of Public Health, 77, 359– 361. 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 –12. Hausenblas, H. A., Carron, A. V., & Mack, D. E. (1997). Application of the theories of reasoned action and planned behavior to exercise behavior: a meta-analysis. Journal of Sport and Exercise Psychology, 19, 36 – 51. Health Canada (1998). Canada’s physical activity guide to healthy active living. Ottawa: Health Canada. Jacobs, D. R., Ainsworth, B. E., Hartman, T. J., & Leon, A. S. (1993). A simultaneous evaluation of ten commonly used physical activity questionnaires. Medicine and Science in Sports and Exercise, 25, 81 – 91. Katzmarzyk, P. T., Gledhill, N., & Shephard, R. J. (2000). The economic burden of physical inactivity in Canada. Canadian Medical Association Journal, 163, 1435– 1440. Lowe, R., Eves, F., & Carroll, D. (2002). The influence of affective and instrumental beliefs on exercise intentions and behavior: a longitudinal analysis. Journal of Applied Social Psychology, 32, 1241– 1252. Marshall, S. J., & Biddle, S. J. H. (2001). The transtheoretical model of behavior change: a meta-analysis of applications to physical activity and exercise. Annals of Behavioral Medicine, 23, 229– 246. Rhodes, R. E., & Courneya, K. S. (2003a). Modelling the theory of planned behaviour and past behaviour. Psychology, Health, and Medicine, 8, 57 – 69. Rhodes, R. E., & Courneya, K. S. (2003b). Investigating multiple components of attitude, subjective norm, and perceived control: an examination of the theory of planned behavior in the exercise domain. British Journal of Social Psychology, 42, 129– 146. Rhodes, R. E., & Courneya, K. S. (2003c). Relationships between personality, an extended theory of planned behaviour model, and exercise behaviour. British Journal of Health Psychology, 8, 19 – 36. Rhodes, R. E., Courneya, K. S., & Hayduk, L. A. (2002). Does personality moderate the theory of planned behavior in the exercise domain? Journal of Sport and Exercise Psychology, 24, 120–132. Rhodes, R. E., Courneya, K. S., & Jones, L. W. (2003). Translating exercise intentions into behaviour: personality and social cognitive correlates. Journal of Health Psychology, 8, 449– 460. Rhodes, R. E., Courneya, K. S., & Jones, L. W. (2004). Personality and social cognitive influences on exercise behavior: adding the activity trait to the theory of planned behavior. Psychology of Sport and Exercise, 5, 243– 254. Sallis, J. F. (2001). Progress in behavioral research on physical activity. Annals of Behavioral Medicine, 23, 77 – 78. Spirduso, W. W., & Cronin, D. L. (2001). Exercise dose-response effects on quality of life and independent living in older adults. Medicine and Science in Sports and Exercise, 33, S598– S608.

R.E. Rhodes, K.S. Courneya / Psychology of Sport and Exercise 6 (2005) 349–361

361

Sutton, S. (2002). Using social cognition models to develop health behaviour interventions: problems and assumptions. In D. Rutter, & L. Quine (Eds.), Intervention research with social cognition models (pp. 193– 208). Buckingham, England: Open University Press. Sutton, S. (1998). Predicting and explaining intentions and behavior: how are we doing? Journal of Applied Social Psychology, 28, 1317– 1338. US Department of Health and Human Services (1996). Physical activity and health: a report of the surgeon general. Atlanta, GA: Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion.