Appendix C
Data extraction, Studies of Affective Expectations and Physical Activity
Correlational Studies
Study- Hoyt, Rhodes, Hausenblas & Giacobbi (2009)
Sample- 507 University students
Design- Cross-sectional
Setting/recruitment- University setting
IV- Semantic differential, affective attitude, 2 items
DV- GLTEQ
Theory- Theory of Planned Behaviour
r Correlation- .38 (.51 corrected)
Outcomes- The structural model explained 46% of the variance in exercise behaviour. TPB variables, including affective attitude (standardized effect = .26) had indirect effects on exercise through intention. Intention and PBC were significant predictors of exercise behaviour. The integrated TPB model explained 70% of the variance in intention, with affective attitude (standardized effect = .62) contributing to the direct explanation of intention.
Study- Karvinen, Courneya, Plotnikoff, Spence, Venner & North (2009)
Sample- 397 Adults with bladder cancer
Design- Prospective, 3 months
Setting/recruitment – Participants approached through the Alberta Cancer Registry by mail
IV- Semantic differential, affective attitude, 3 items
DV- GLTEQ
Theory- Theory of Planned Behaviour
r Correlation- .34, p < .001 (.46 corrected)
Outcomes- Exercise behaviour correlated the strongest with intention (r=.41, p < .001), and intention had the strongest correlations with PBC (r=.56, p < .001), instrumental attitude (r=.52, p < .001), and affective attitude (r=.51, p<.001). First hierarchical regression analysis: intention (β=.25, p <.001), PBC (β=.18, p<.001), and planning (β=.12, p <.018) explained 20.9% of the variance in exercise behaviour. Second regression analysis: instrumental attitude (β=.15, p <.025), affective attitude (β=.18, p<.002), descriptive norm (β=.10, p <.032) and PBC (β=.32, p<.001) explained 39.1% of the variability in exercise intention.
Study – Lawton, Conner & McEachen (2009)
Sample – 390 adults
Design – Prospective, 1 month
Setting/recruitment – rural setting; local newspaper, newsletters, internet website, poster, word of mouth
IV – Semantic differential, affective attitude, 1 item
DV –Frequency of exercise, number of days in past 4 weeks
Theory –Social Cognitive Theory, Theory of Planned Behaviour, Health Belief Model
r Correlation - .36, p<.001 (corrected .51)
Outcomes – Affective attitude was a significant independent predictor of exercise intention and behaviour.
Study – McArthur & Raedeke (2009)
Sample – 636 Undergraduate students
Design – Cross-sectional
Setting/recruitment – University fitness class
IV –Semantic differential,intrinsic motivation (test-retest reliability=.76), PACES; 8 items (test-retest reliability= .91)
DV – GLTEQ
Theory: None stated
r Correlation- .29, p< .01 (corrected .42)
Outcomes – Correlations between exercise motives and self-reported physical activity were significant, but weak in magnitude; exercise enjoyment and self-management skills were most strongly related to activity level, with correlations in the moderate range; intrinsic motives reported greater enjoyment of exercise than did students who placed less importance on these motives (correlations ranges from .3 to .54); regularly active students placed more importance on health/fitness and intrinsic reasons for exercise than did the other 2 groups with effect sizes being the small to moderate range (d= .33 to .63); regularly active students reported higher exercise enjoyment (active vs inactive, d= 1.2; regularly active vs somewhat active, d= .53; somewhat active vs inactive, d=.61); no significant race or sex differences emerged on intrinsic motives for exercise; males reported enjoying exercise more than females; no race differences were found on physical activity enjoyment.
Study – McIntyre & Rhodes (2009)
Sample – 139 Women with at least one child between the ages of 0-4
Design – Cross-sectional
Setting/recruitment–Five parenting programs within five municipalities in Greater Victoria area were randomly selected from a master list.
IV –Semantic differential, affective attitude, 2 items (aggregated correlations = .37)
DV –GLTEQ
Theory–Theory of Planned Behaviour
r Correlation - .26 (corrected .39)
Outcomes – Regression results: in the first model, TPB constructs explained 22% of the variance in PA (intention, beta = .33; PBC, beta = .20; no significant contribution from other TPB constructs). In the second model, TPB constructs explained 47% of the variance in intention (PBC, beta = .40; affective attitude, beta = .39; subjective norm, beta = .25).
Study- Nigg (2009)
Sample- 3,533 Adults
Design- Cross-sectional
Setting/recruitment- Random digit dial survey of Hawaii’s non-institutionalized adult population (baseline for a longitudinal study)
IV- Semantic differential, affective attitude, 2 items (α= .93)
DV- International Physical Activity Questionnaire (IPAQ)
Theory- Theory of Planned Behaviour
r Correlation- NA
Outcomes- Intercorrelations between attitude, subjective norm and behavioural control were constrained to be equal across age groups; the explained variance of intention was substantial (about 50%) and comparable in all age groups. There were few differences of significance between ethnic groups; the Fillipino sample had a significantly lower path coefficient compared to all other sub-samples on their relationship of attitude and intention.
Study- Stevinson, Capstick, Schepansky, Ladha, Vallance, Faught, Steed & Courneya (2009)
Sample- 359 Ovarian cancer survivors
Design- Cross-sectional
Setting/recruitment- Postal survey of ovarian cancer survivors identified through the Alberta Cancer Registry
IV- Semantic differential, affective attitude, 3 items
DV- GLTEQ
Theory- Theory of Planned Behaviour
r Correlation- .36, p < .001 (.50 corrected)
Outcomes- The strongest association was between PA intention and behaviour (r= .60, p<.001); PA intention had the strongest correlations with affective (r= .51, p <.001) and instrumental (r=.54, p< .001) attitudes and PBC (r= .51, p<.001). Hierarchical multiple-regression analysis: 36% of the variance in PA behaviour was explained with intention being the sole independent correlate (β=.56, p<.001).
Study – Bellows-Riecken, Rhodes & Hoffert (2008)
Sample – 150 Undergraduate Students
Design – Cross-sectional
Setting/recruitment–General education courses
IV –Semantic differential, affective attitude, 2 items (α = .77)
DV –GLTEQ
Theory–Theory of Planned Behaviour
r Correlation - .3, p < .01 (corrected .43)
Outcomes – Affective attitude (lifestyle PA < exercise) and PBC (lifestyle PA > exercise) amounted to a small effect size q difference for the TPB-behaviour correlations. Affective attitude and intention (lifestyle PA < exercise) correlations also showed a small effect size difference. For frequency of strenuous activities, correlation differences were present for affective attitude, intention, and the multivariate TPB model; all differences favoured larger correlations for exercise than lifestyle PA and effect sizes were small to medium.
Study – Blanchard, Fisher, Sparling, Nehl, Rhodes, Courneya & Baker (2008)
Sample – 238 African American & 197 Caucasian Students
Design – Cross-sectional
Setting/recruitment–Undergraduate health and fitness classes
IV –Semantic differential, affective attitude, 6 items (α=.74)
DV –LSI-GLTEQ
Theory–Theory of Planned Behaviour
r Correlation – African American, .19, p < .05 (corrected .28); Caucasian, .32, p < .01 (corrected .47)
Outcomes – The model accounted for 65% of the variance in intention for the Caucasian students, with affective (Beta = .23) & 49% of variance for the African American students, with affective attitude (Beta= .18); the model accounted for 22% of variance in PA for the Caucasian students, with intention (Beta= .33) being the sole predictor of PA; the intention-PA relationship was significantly stronger for Caucasian students (Beta= .14, p < .05); sex did not interact with any of the TPB variables when predicting intention or PA, or did ethnicity.
Study – Blanchard, Kupperman, Sparling, Nehl, Rhodes, Courneya, Baker & Rupp (2008)
Sample – 2 week, 237 African American & 176 Caucasian college students; 2 month, 280 African American & 273 Caucasian students
Design – Longitudinal
Setting/recruitment– Undergraduate health and fitness classes
IV –Semantic differential, affective attitude, 2 items (α= .70-.89)
DV –LSI-GLTEQ
Theory–Theory of Planned Behaviour
r Correlation – 2 week: African American .35, p < .001 (corrected .54),Caucasian .40, p < .001 (corrected .47); 2 month: African American .06, not significant (corrected .09), Caucasian .32, p < .001 (corrected .46)
Outcomes – Affective attitudes and PBC were significant predictors of intention in both samples; intention was a significant predictor of exercise in both ethnic groups for both samples, explained variation in exercise was small for the African American students at 2 months; Caucasian students had significantly higher scores on affective attitude and PBC compared to the African American students, both of which led to increased intentions and significantly more exercise at 2 weeks and 2 months (TPB had a small mediating effect on the ethnicity/exercise relationship)
Study- Calitri, Lowe, Eves & Bennett (2008)
Sample- 125 Undergraduate students
Design- Cross-sectional
Setting/recruitment- University setting
IV- Semantic Differential, affective attitude, 4 items (α= .88)
DV- 7-day Physical Activity Recall (Sallis et al., 1985)
Theory- None stated
r Correlation- .27, p <.01 (.34 corrected)
Outcomes- No significant effect of task order on any of the study variables; task order did not moderate the association between implicit attitude and PA. Positive correlations were observed between implicit attitude and previous PA; explicit attitude (affective and instrumental) and intention were all positively associated with eachother; affective attitude and intention were positively associated with previous PA. Implicit attitude did not moderate the association between attention and PA, however, both affective and instrumental attitude did. For affective attitude, the model accounted for 18% of the variance in PA, and instrumental attitude accounted for 9% of the variance in PA. Those with a low (more negative) attitude, there was no significant association between attention and PA, however, those with a high (more positive) attitude there was a significant positive association between attention and PA. Instrumental and affective attitude did not interact with implicit attitude in their association with previous PA.
Study – Cerin, Leslie, Vandelanotte & Merom (2008)
Sample – Adults from 32 urban communities (Two samples: n = 1959 moderate to vigorous LTPA, n = 2134 walking for recreation)
Design – Cross-sectional
Setting/recruitment–Random sampling from census collection districts
IV –PA Enjoyment, 3 items (α= .92)
DV – International Physical Activity Questionnaire (IPAQ), LTPA moderate & vigorous PA, MET score
Theory–Ecological
r Correlation - .29 average for moderate (.21) and vigorous (.36) PA (corrected .34); walking .15 (corrected .17)
Outcomes – Participants with missing data tended to enjoy PA less; enjoyment of and self-efficacy for vigorous LTPA moderated the relationship between presence of outdoor individual sport/fitness facilities and MET-minutes of LTPA, the presence of these facilities had a small positive effect on respondents with a below-average enjoyment and self-efficacy for vigorous LTPA only
Study – Craike (2008)
Sample – 250 Adults
Design – Cross-sectional
Setting/recruitment–Sequential selection of 729 potential participants from the State Electoral Division role of an Australian city.
IV –Intrinsic motivation towards LTPA (ILM-modified), 4-5 items; Enjoyment of LTPA, 5 items
DV –Regular participation of PA (at least 3 times/week) over past 4 weeks
Theory–Self Determination Theory
r Correlation - NA
Outcomes – Enjoyment did not predict regular participation in LTPA either directly or indirectly.
Study – Milne, Wallman, Guilfoyle, Gordon & Courneya (2008)
Sample – 558 Breast Cancer Survivors
Design – Cross-sectional
Setting/recruitment–Western Australian Cancer Registry (WACR) at the Department of Health Western Australia
IV –BREQ-2, Intrinsic Motivation, 4 items
DV –GLTEQ
Theory–Self Determination Theory
r Correlation – .38 (d= .83, p < .001; corrected r= .58)
Outcomes – Group meeting the public health PA guidelines reported significantly higher identified regulation and intrinsic motivation; multivariate regression analysis (block 3): intrinsic motivation was significant predictor of PA (Beta = .18, p < .01)
Study – Peddle, Plotnikoff, Wild, Au & Courneya (2008)
Sample – 414 Colorectal Cancer Survivors
Design – Cross-sectional
Setting/recruitment–Random sample of 1,498 CRC-S residing in Alberta, Canada (diagnosed between January 1980 and May 2003) through provincial cancer registry
IV –BREQ-2, Intrinsic regulation, 19 items (test-retest reliability .94)
DV –LSI-GLTEQ
Theory–Self Determination Theory
r Correlation - .42, p < .001 (corrected .55)
Outcomes – There was a moderate positive correlation between exercise behaviour and intrinsic regulation.
Study – Rhodes & Blanchard (2008)
Sample – 174 Undergraduate Students, 206 Adults (community sample)
Design – Prospective, 2 weeks (undergraduates), Cross-sectional (community)
Setting/recruitment–Community: random sample of 1000 addresses obtained from a local telephone company in the Greater Victoria Capital Region District; Undergraduate: volunteers from primary and secondary Education courses.
IV –Affective attitude, 2 items (internal consistency: community= .73; undergraduate= .87)
DV –LSI-GLTEQ
Theory–Theory of Planned Behaviour
r Correlation - .25, p < .01 (undergraduate, corrected .34); .27, p < .01 (community, corrected .4)
Outcomes – Overall, all TPB constructs correlated with PA intention, but only affective attitude, PBC, and intention correlated with PA behaviour. Hierarchical regressions indicated that PA affective attitude and PA PBC relations with PA behaviour were mediated by PA intention. For the prediction of PA intention, PA affective attitude (beta = .51, p < .01), PA PBC (beta = .31, p < .01) and TV affective attitude (beta = -.14,p < .01) explained 50% of the variance in the community sample. In the undergraduate sample, PA affective attitude (beta = .18, p < .01), PA PBC (beta = .39, p < .01) and TV PBC (beta = -.19, p < .01) explained 25% of the variance in intention.
Study – Rhodes, Blanchard & Blacklock (2008), same sample Rhodes & Blanchard (2007)
Sample – 6,739 Adults, stratified by age/gender
Design – Cross-sectional
Setting/recruitment–National U.S. mail-out survey via hired vendor (MarketFacts; mail panel database of approximately 500,000 people)
IV –Behavioural Beliefs, 11 items (fun, internal consistency of aggregate behavioural beliefs = .9)
DV –BRFSS
Theory–Theory of Planned Behaviour
r Correlation - .15 (corrected .19)
Outcomes – Overall, no meaningful differences across groups were found for any of the normative or behavioural beliefs. Correaltion differences among age and gender of behavioural belief (fun) and PA: 18-24: M=.15, F= .15 (ns); 24-34: M=.12, F=.18 (p < .01); 35-44: M=.24, F=.12 (p < .01); 45-54: M=.20, F=.16 (p < .01); 55-64: M=.19, F=.13 (p < .01); 65-74: M=.08 (ns), F=.23(p < .01); 75+: M=-.01, F=.11(ns)
Study- Rhodes, Hunt Matheson, Blanchard & Blacklock (2008)
Sample- 409 Undergraduate students
Design- Cross-sectional
Setting/recruitment- Introductory Psychology Classes
IV- Semantic differential, affective attitude, 3 items (α = .8)
DV- GLTEQ, Past PA
Theory- Theory of Planned Behaviour
r Correlation- .3 (.43 corrected)
Outcomes- Time frame did not affect mean values in more than 60% of the TPB constructs; there was no clear difference between affective and instrumental beliefs. PA expectancies and intention measured by TPB were robust to timeframe. Overall, the behavioural expectancies about feeling good, relieving stress, PA taking too much time and the behavioural value of increasing fitness appear to be important targets (medium effect size correlates of PA) for PA intention promotion.
Study- Rhodes, Plotnikoff & Courneya (2008)
Sample- 611 Adults
Design- Prospective, 6 months
Setting/recruitment- Participants were employees from three large organizations in Alberta, Canada.
IV- Semantic differential, affective attitude, 3 items (α = .79)
DV- Past and future PA; staging algorithm, Yes/No
Theory- Theory of Planned Behaviour, Transtheoretical Model, Protection Motivation Theory
r Correlation- .2 (.27 corrected)
Outcomes- Five profiles were examined: nonintenders, unsuccessful adopters, successful adopters, unsuccessful maintainers, and successful maintainers. Affective attitude, PBC, behavioural processes, and self-efficacy were the key correlates across intention-behaviour profiles and correctly identified 50% of cases. For TPB variables, follow-up tests showed that affective attitude discriminated several of the profiles but did not identify differences between unsuccessful and successful adoption or maintenance. Instrumental attitude discriminated nonintenders from all profiles of intenders.
Study- Rogers, Courneya, Robbins et al (2008)
Sample- 59 Adults with head and neck cancer
Design- Cross-sectional
Setting/recruitment- Convenience sample in an academic Head and Neck Cancer outpatient clinic
IV- Physical Activity Enjoyment, single item
DV- Modified GLTEQ
Theory- Social Cognitive Theory
r Correlation- .41, p < .01 (.62 corrected)
Outcomes- The strongest correlates of PA included enjoyment, symptom index, alcohol use, task self-efficacy, perceived barriers and comorbidity score; stepwise regression demonstrated that two variables had independent associations with PA: enjoyment and symptom index.
Study – Rogers, McAuley, Courneya & Verhulst (2008)
Sample – 198 Breast Cancer Survivors
Design – Cross-sectional
Setting/recruitment–State cancer registry identified Breast Cancer survivors who had been diagnosed within a specific time frame (January 1997 to December 2002).
IV –Enjoyment of walking, 1 item
DV – GLTEQ
Theory–Social Cognitive Theory
r Correlation - .29, ns (corrected .44)
Outcomes – Enjoyment was significantly associated (p < .05) with barrier self-efficacy (beta = .12) and task self-efficacy (beta = .25). Barrier self-efficacy (beta = .12) and task self-efficacy (beta = .15) were significantly associated (p < .05) with physical activity. Breast Cancer survivors with more social support, reporting less fatigue, and who were more active prior to diagnosis and enjoyed physical activity wer more efficacious with respect to being physically active.
Study- Standage, Sebire & Loney (2008)
Sample- 55 University students
Design- Prospective, 1 week
Setting/recruitment – University setting
IV- BREQ, 15 items, intrinsic motivation
DV- AHR device for 8 days (heart rate and accelerometry data)
Theory- Self-determination Theory
r Correlation- .34, p < .05 (.42 corrected)
Outcomes- Intrinsic motivation and identified regulation shared moderate and positive association with the objective assessments of exercise behaviour. Sequential regression analysis: the addition of motivation variables to the model accounted for an additional 13% of the variance in moderate-intensity exercise behaivor.
Study – Ball, Timperio, Salmon, Giles-Corti, Roberts & Crawford (2007)
Sample – 1,282 Women
Design – Cross-sectional
Setting/recruitment–Stratified random sampling from 45 Melbourne neighbourhoods of different levels of socioeconomic disadvantage.
IV – Enjoyment of Walking, 8 items (α = .96)
DV –International Physical Activity Questionnaire (IPAQ); walking for leisure-time and transport
Theory–Multilevel Framework (intrapersonal, social, and perceived & objectively assessed physical environmental factors to explain socioeconomic variations in women’s walking)
r Correlation - .65 (corrected .74), Odds Ratio: 3.05
Outcomes – Enjoyment was associated with leisure-time walking.Multilevel logistic regression analyses: enjoyment remained a significant predictor of leisure-time walking; in general, the personal variables showed the strongest associations in the full regression model.
Study – Blanchard, Kupperman, Sparling, Nehl, Rhodes, Courneya, Baker & Hunt (2007)
Sample – 170 African American students, 180 Caucasian students
Design – Prospective, 2 weeks
Setting/recruitment–2 Universities in a major city on the southern United States, undergraduate health & fitness classes
IV –Semantic differential, affective attitude, 6 items (internal reliability: African American = .72, Caucasian= .79)
DV –LSI-GLTEQ
Theory–Theory of Planned Behaviour
r Correlation – African American= .17, p < .05 (.25 corrected); Caucasian= .13, p < .05 (.19 corrected)
Outcomes – Caucasian students engaged in significantly more PA. Regression analyses, TPB constructs with intention: Caucasian, 46% of variance in intention, with affective attitude (beta=.18), subjective norm (beta=.14) & PBC (beta=.54) making significant contributions to intention; African American, 49% of variance in intention, with affective attitude (beta=.32), subjective norm (beta=.21) & PBC (beta=.35) making significant contributions to intention. Association of TPB constructs with PA: Caucasian, 23% of variance in PA, with intention being the sole predictor of PA; African American, 18% of variance in PA, with intention as the only significant predictor; none of the TPB relationships were moderated by ethnicity.