ann. behav. med.

(2014) 48:125129
DOI 10.1007/s12160-013-9578-3

BRIEF REPORT

Psychosocial Mediators of Two Community-Based Physical

Activity Programs
Katie Becofsky, M.S. & Meghan Baruth, Ph.D. &
Sara Wilcox, Ph.D.

Published online: 18 December 2013

# The Society of Behavioral Medicine 2013

Abstract
Background Previous findings are inconclusive regarding the
mediators of physical activity behavior change.
Purpose To test self-efficacy and social support as mediators
of Active Choices, a telephone-delivered physical activity
intervention, and Active Living Every Day, a group-based
physical activity intervention, implemented with midlife and
older adults in community settings.
Methods MacKinnon's product of coefficients was used to
examine social support and self-efficacy as mediators of
change in physical activity. The proportion of the total effect
mediated was calculated. Each model controlled for age,
gender, race (white vs. non-white), body mass index (BMI),
and education (high school graduate or less vs. at least some
college).
Results Increases in self-efficacy mediated increases in physical activity among Active Choices (n =709) and Active Living Every Day (n =849) participants. For Active Living Every
Day, increases in social support also mediated increases in
physical activity in single mediator models.
Conclusions Increasing self-efficacy and social support may
help increase physical activity levels in older adults.
K. Becofsky : S. Wilcox (*)
Department of Exercise Science, Arnold School of Public Health,
University of South Carolina, 921 Assembly Street, Columbia,
SC 29208, USA
e-mail: wilcoxs@mailbox.sc.edu
K. Becofsky
e-mail: becofsky@email.sc.edu
M. Baruth
College of Health and Human Services, Saginaw Valley State
University, 7400 Bay Rd University Center,
Columbia, MI 48710, USA
S. Wilcox
Prevention Research Center, University of South Carolina,
Columbia, SC, USA

Keywords Physical activity . Intervention . Self-efficacy .

Social support . Older adults . Behavior change

Introduction
Mediation analyses can help explain how physical activity
interventions achieve their outcomes [1]. Acquiring this information may provide support for current behavior change
theories, or may prompt revision or new theory development.
The ultimate goal of mediation analyses is to understand the
critical components of successful interventions (or the missing
components of unsuccessful ones) to maximize the efficiency
and effectiveness of future programs.
Physical activity self-efficacy and social support have
long been considered key factors for physical activity adoption and maintenance. Physical activity and self-efficacy refers to an individual's confidence in his or her ability to engage
in physical activity, even when encountering barriers, whereas
social support encompasses various types of encouragement
and assistance received for engaging in physical activity.
These constructs are consistent with social cognitive theory
[2] and the transtheoretical model [3], both of which are
commonly used as the theoretical bases for physical activity
interventions.
Although physical activity self-efficacy and social support
are staples of the physical activity behavior change literature,
evidence of their ability to mediate intervention effects is
conflicting [4, 5]. Even less information is known about their
mediating role in interventions targeting older adults. Two
review papers have found limited evidence for the role of
these behavioral constructs in mediating physical activity
interventions, although one excluded studies with adults 65+
[5], and the other included only 1 study with older adults [4].
Cerin et al. suggested that the inconclusive findings regarding
the mediators of physical activity behavior change may be

126

due, in part, to the use of statistical methods that are incapable

of detecting mediation effects in relatively small samples [6].
The current study sought to test whether self-efficacy for
physical activity and social support mediated the effects of
two physical activity programs implemented with older adults
in community settings. Both programs were theory based [2,
3], and specifically targeted these constructs. Importantly, our
study will address two of the limitations Cerin et al. [6] cite as
common to physical activity mediation studies: our study has
a large sample size and utilizes a powerful analytic approach
(MacKinnon [7]).

Methods
Active For Life was a 4-year translational initiative that successfully implemented two physical activity programs, Active
Choices (a 6-month telephone-based program) and Active
Living Every Day (a 20-week group-based program), into
community settings with diverse midlife and older adult populations [8]. Both Active Choices [9, 10] and Active Living
Every Day [11, 12] were previously tested in rigorous randomized controlled trials. Both led to significant improvements in physical activity when translated to community
settings (see Wilcox et al. 2008 [8] for methodology details
and other major findings). In the last year of the Active for
Life initiative, the Active Living Every Day program was
shortened from 20 to 12 weeks at the request of the participating organizations; steps were taken to ensure essential
elements were preserved [8]. The current study used a pre
post design, with data collected from 2003 to 2007.
Participants
All Active for Life participants were 50 years of age, not
meeting physical activity recommendations (2 days/week
and <120 min/week), and free of serious medical conditions
or disabilities that required higher levels of supervision [8].
Measures
Self-reported moderate-to-vigorous physical activity was
measured via the 41-item Community Health Activities Model Program for Seniors questionnaire; total hours per week
spent in moderate-to-vigorous physical activity (3 METs)
was determined based on the frequency and duration of all
reported activities [13]. Physical activity self-efficacy was
measured with a 5-item scale which asked participants to rate
their confidence in overcoming common barriers to physical
activity [14]. Social support from friends and family was
measured with a 5-item scale developed for the US Women's
Determinants Study [15] and derived from an original scale
developed by Sallis and colleagues [16]. Participants

K. Becofsky et al.

completed surveys at baseline and post-intervention (6 months

for Active Choices; 20 weeks for Active Living Every Day).
Analyses
The independent variable was intervention dose, measured as
the number of phone calls received (Active Choices) or the
percentage of classes attended (Active Living Every Day).
The dependent variable was self-reported moderate-tovigorous physical activity. We tested the mediating effects of
self-efficacy and social support in both interventions, separately, using MacKinnon's product of coefficients [7]. This
study includes data across 3 years of recruitment for Active
Living Every Day and 3 years of recruitment for Active
Choices. Year 2 data were not collected for either program
(planning year), and year 1 individual participant attendance
data was not collected for Active Living Every Day.
Square root transformations corrected skewness in baseline
and post-program physical activity scores. Residualized
change scores (pretest to posttest) for physical activity, selfefficacy, and social support were computed and used in subsequent regression models. SAS PROC MIXED controlled
for site clustering. We ran two successive models for each
mediator (self-efficacy and social support) for both Active
Choices and Active Living Every Day (single mediator
models). Each model controlled for age, gender, race (white
vs. non-white), BMI, and education (high school graduate or
less vs. at least some college). The first model regressed
change in self-efficacy/social support on intervention dose
(-coefficient). The second model regressed change in physical activity on intervention dose and change in self-efficacy/
social support (-coefficient). Next, a multiple mediator model was conducted where self-efficacy and social support were
examined simultaneously (see Fig. 1 for a visual display of the
single and multiple mediator models). Asymmetric confidence limits based on the distribution of the product were
constructed using the PRODCLIN program [17, 18]; if
the confidence interval did not include zero, the mediating effect was considered significant [7, 19]. Finally, the
proportion of the total effect mediated was calculated as
/c, where c was the direct effect of the intervention
dose on change in physical activity.

Results
Table 1 shows the baseline characteristics of all participants
included in analyses. For Active Choices (n =709) and Active
Living Every Day (n =849), respectively, participants were
66.7 (8.9) and 70.4 (9.0)years old, 80 and 84 % female, and
50.6 and 63.8 % white. Table 2 shows the - and coefficients and the asymmetric confidence limits of all single
and multiple mediator regression analyses. For Active

Psychosocial Mediators of Two Physical Activity Programs

Fig. 1 Single and multiple
mediator models for Active
Choices (AC) and Active Living
Every Day (ALED). a Single
mediator model, b multiple
mediator model

127

a. Single Mediator Model

Psychosocial Mediator
(Social Support/Self Efficacy

pathway

pathway

Dose (# of calls
completed for AC or %
of classes attended for
ALED)

Hrs/Week of moderate to
vigorous physical activity

b. Multiple Mediator Model

Mediator 1:
Social Support

1 pathway

1 pathway

Dose (# of calls
completed for AC or %
of classes attended for
ALED

Hrs/Week of moderate to
vigorous physical activity

2 pathway

Mediator 2:

2 pathway

Self-Efficacy

of moderate-to-vigorous physical activity in both the single

and multiple mediator models (-pathways). Tests for mediation indicated that increased self-efficacy was a significant
mediator in both the single and multiple mediator analyses.

Choices, there was a significant positive relationship between

the percentage of phone calls received and change in selfefficacy (-pathway). Increased social support and selfefficacy were significantly related to increased hours/week
Table 1 Baseline characteristics
of participants taking part in the
Active Choices (n =709) and Active Living Every Day (n =849)
programsa

Only participants with complete

data required for mediation analyses were included in this study,
thus sample will differ from that
reported in previous Active for
Life papers

b
Percentage of phone calls (Active Choices) and group-based
sessions (Active Living Every
Day) completed

Age, years
BMI, kg/m2
Gender
Male
Female
Education
HS grad or less
At least some college
Race
Non-white
White
Social support
Self-efficacy
Attendance, %b
MVPA, hours/week

Active Choices

Active Living Every Day

Mean (SD) or %

Mean (SD) or %

709
709

66.7 (9.8)
30.5 (7.0)

849
849

70.4 (9.0)
30.0 (6.9)

142
567

20.0
80.0

134
715

15.8
84.2

186
523

26.2
73.8

335
514

39.5
60.5

350
359
709
709

49.4
50.6
13.4 (3.1)
20.9 (6.9)

307
542
849
849

36.2
63.8
13.4 (3.0)
20.0 (7.8)

709
709

78.7 (24.8)
2.9 (3.6)

849
849

75.7 (21.5)
2.7 (3.7)

128

K. Becofsky et al.

Table 2 Mediation effects for physical activity in Active Choices (n =709) and Active Living Every Day (n =849)a
Alpha estimate (SE b)

Active Choicessingle mediator models

Social support
0.293 (0.415)
0.480
Self-efficacy
3.645 (1.046)
0.001
Active Choicesmultiple mediator models
Social support
0.293 (0.415)
0.480
Self-efficacy
3.645 (1.046)
0.001
Active Living Every Daysingle mediator models
Social support
0.908 (0.434)
0.037
Self-efficacy
2.342 (1.088)
0.032
Active Living Every Daymultiple mediator models
Social support
0.908 (0.434)
0.037
Self-efficacy
2.342 (1.088)
0.032

Beta estimate (SE)

Asymmetric confidence limit

Proportion mediated

0.051 (0.015)
0.037 (0.006)

0.001
<0.001

0.027, 0.063
0.054, 0.228

4%
37 %

0.040 (0.015)
0.034 (0.006)

0.007
<0.001

0.022, 0.052
0.054, 0.228

3%
35 %

0.036 (0.014)
0.0342 (0.005)

0.009
<0.001

0.001, 0.081
0.007, 0.163

7%
17 %

0.0267 (0.014)
0.033 (0.005)

0.049
<0.001

0.002, 0.066
0.007, 0.158

5%
16 %

Each model controlled for age, gender, race (white vs. non-white), body mass index, and education (high school graduate or less vs. at least some
college)

SE standard error

Increased self-efficacy explained 37 (single mediator model)

and 35 % (multiple mediator model) of the effect of program
attendance on change in physical activity. For Active Living
Every Day, there was a significant positive relationship between the percentage of classes attended and change in both
social support and self-efficacy (-pathways). Increased social support and self-efficacy were significantly related to
increased hours/week of moderate-to-vigorous physical activity in both single and multiple mediator models (-pathways).
Tests for mediation showed that change in self-efficacy and
social support were significant mediators in single mediator
models (explaining 17 and 7 % of the effect of program
attendance on change in physical activity). Self-efficacy was
also a significant mediator in the multiple mediator model
(explaining 16 % of the program effect).

Discussion
In this study, increases in self-efficacy mediated the relationship between intervention dose and increases in moderate-tovigorous physical activity in older adults in both Active
Choices and Active Living Every Day. Social support also
mediated this relationship in Active Living Every Day, but
only when considered separately from self-efficacy. Significance in all -pathways indicates that changes in self-efficacy
and social support were indeed related to physical activity
adoption in Active for Life. Significance in the self-efficacy
-pathway for Active Choices and both the self-efficacy and
social support -pathways for Active Living Every Day
suggest that the mode of delivery may have determined
each program's psychosocial effects; the one-on-one
feedback provided by the phone-based Active Choices

intervention built self-efficacy, whereas Active Living

Every Day was a group-based intervention that was able
to foster both self-efficacy and social support. Our findings speak to the indispensable role of self-efficacy in
physical activity adoption, as this construct mediated
program effects regardless of the delivery mode.
With one exception, the Active for Life programs were
successful in increasing social support and self-efficacy in
older adults (evidenced by significance in all but one pathway). Although both programs ultimately increased physical activity, the telephone-based intervention (Active
Choices) did not increase social support. While a face-toface, group-based delivery mode seems naturally better suited
for fostering a sense of social support in this population, it is
also possible that an intermediary assessment might have
demonstrated different results. For example, perhaps the
greatest changes in social support were found earlier and thus
we failed to detect them in Active Choices simply due to the
timing of our measurements. Alternatively, other factors not
investigated in this study may have mediated changes in
Active Choices physical activity outcomes; for example, multiple studies have shown behavioral processes of change (e.g.,
reminding oneself, rewarding oneself) to mediate physical
activity adoption [20, 21]. Unfortunately, due to the nature
of community-based work where participant burden is recognized, we were unable to measure additional mediators.
A major strength of this study is the use of formal analyses
to investigate the psychosocial mediators of physical activity
behavior change in Active for Life. Furthermore, the translational nature of Active for Life allowed us to study mediation
effects in real-world settings. Additionally, the sample was
large and diverse (both racially and in terms of socioeconomic
status). A limitation of this study is the lack of a control group,

Psychosocial Mediators of Two Physical Activity Programs

which may threaten internal validity. Another limitation is the

reliance on a self-report measure of physical activity, although
the Community Health Activities Model Program for Seniors
questionnaire is considered reliable and valid for use with
older adults [13, 22].
The current study found that increases in self-efficacy
mediated increases in moderate-to-vigorous physical activity
among midlife and older adults who participated in Active
Choices, a telephone-delivered intervention, and Active Living Every Day, a group-based intervention. Changes in social
support also mediated increases in physical activity in Active
Living Every Day participants, but not when considered simultaneously with self-efficacy. Importantly, the Active for
Life programs were implemented in community settings, and
thus these findings reflect real-world factors that contribute
to physical activity adoption in older adulthood. A recent
study by Carlson et al. found that environmental factors may
interact with psychosocial factors to predict physical activity
levels in older adults; among their findings, neighborhood
walkability interacted with social support to predict
moderate-to-vigorous physical activity [23]. This finding suggests that multi-level interventions may be most effective for
getting older adults more active, and thus future studies based
on socialecological models (which consider multiple levels
of influence, from individual to environmental/policy [24])
might improve our understanding of the causal pathway between program participation and physical activity adoption.
Acknowledgments The Active for Life initiative was funded by the
Robert Wood Johnson Foundation. The findings and conclusions in this
report are those of the authors and do not necessarily represent the views
of the Robert Wood Johnson Foundation. We gratefully acknowledge the
many participants who took part in the initiative.
Conflict of Interest The authors have no conflict of interest to disclose.

References
1. MacKinnon DP, Fairchild AJ, Fritz MS. Mediation analysis. Annu
Rev Psychol. 2007; 58: 593-614.
2. Bandura A. Social foundations of thought and action: a social
cognitive theory. Englewood Cliffs, NJ: Prentice-Hall; 1986.
3. Prochaska JO, DiClemente CC, Norcross JC. In search of how people
change. Applications to addictive behaviors. Am Psychol. 1992; 47:
1102-1114.
4. Lewis BA, Marcus BH, Pate RR, Dunn AL. Psychosocial mediators
of physical activity behavior among adults and children. Am J Prev
Med. 2002; 23: 26-35.
5. Rhodes RE, Pfaeffli LA. Mediators of physical activity behavior
change among adult non-clinical populations: a review update. Int J
Behav Nutr Phys Act. 2010; 7: 37.

129
6. Cerin E, Taylor LM, Leslie E, Owen N. Small-scale randomized
controlled trials need more powerful methods of mediational analysis
than the BaronKenny method. J Clin Epidemiol. 2006; 59: 457-464.
7. MacKinnon DP, Lockwood CM, Hoffman JM, West SG, Sheets V. A
comparison of methods to test mediation and other intervening
variable effects. Psychol Methods. 2002; 7: 83-104.
8. Wilcox S, Dowda M, Leviton LC, et al. Active for life: final results
from the translation of two physical activity programs. Am J Prev
Med. 2008; 35: 340-351.
9. King AC, Haskell WL, Young DR, Oka RK, Stefanick ML. Longterm effects of varying intensities and formats of physical activity on
participation rates, fitness, and lipoproteins in men and women aged
50 to 65 years. Circulation. 1995; 91: 2596-2604.
10. King AC, Baumann K, OSullivan P, Wilcox S, Castro C. Effects of
moderate-intensity exercise on physiological, behavioral, and emotional responses to family caregiving: a randomized controlled trial.
J Gerontol A Biol Sci Med Sci. 2002; 57: M26-M36.
11. Dunn AL, Marcus BH, Kampert JB, Garcia ME, Kohl HW 3rd, Blair
SN. Reduction in cardiovascular disease risk factors: 6-month results
from Project Active. Prev Med. 1997; 26: 883-892.
12. Dunn AL, Marcus BH, Kampert JB, Garcia ME, Kohl HW 3rd, Blair
SN. Comparison of lifestyle and structured interventions to increase
physical activity and cardiorespiratory fitness: a randomized trial.
JAMA. 1999; 281: 327-334.
13. Stewart AL, Mills KM, King AC, Haskell WL, Gillis D, Ritter PL.
CHAMPS physical activity questionnaire for older adults: outcomes
for interventions. Med Sci Sports Exerc. 2001; 33: 1126-1141.
14. Marcus BH, Selby VC, Niaura RS, Rossi JS. Self-efficacy and the
stages of exercise behavior change. Res Q Exerc Sport. 1992; 63:
60-66.
15. Eyler AA, Brownson RC, Donatelle RJ, King AC, Brown D, Sallis
JF. Physical activity social support and middle- and older-aged
minority women: results from a US survey. Soc Sci Med. 1999; 49:
781-789.
16. Sallis JF, Grossman RM, Pinski RB, Patterson TL, Nader PR. The
development of scales to measure social support for diet and exercise
behaviors. Prev Med. 1987; 16: 825-836.
17. MacKinnon DP, Fritz MS, Williams J, Lockwood CM. Distribution
of the product confidence limits for the indirect effect: program
PRODCLIN. Behav Res Methods. 2007; 39: 384-389.
18. Tofighi D, MacKinnon DP. RMediation: an R package for
mediation analysis confidence intervals. Behav Res Methods .
2011; 43: 692-700.
19. MacKinnon DP, Lockwood CM, Williams J. Confidence limits for
the indirect effect: Distribution of the product and resampling
methods. Multivar Behav Res. 2004; 39: 99-128.
20. Baruth M, Wilcox S, Dunn AL, et al. Psychosocial mediators of
physical activity and fitness changes in the activity counseling trial.
Ann Behav Med. 2010; 39: 274-289.
21. Napolitano MA, Papandonatos GD, Lewis BA, et al. Mediators of
physical activity behavior change: a multivariate approach. Health
Psychol. 2008; 27: 409-418.
22. Harada ND, Chiu V, King AC, Stewart AL. An evaluation of three
self-report physical activity instruments for older adults. Med Sci
Sports Exerc. 2001; 33: 962-970.
23. Carlson JA, Sallis JF, Conway TL, et al. Interactions between psychosocial and built environment factors in explaining older adults'
physical activity. Prev Med. 2012; 54: 68-73.
24. Cohen DA, Scribner RA, Farley TA. A structural model of health
behavior: a pragmatic approach to explain and influence health
behaviors at the population level. Prev Med. 2000; 30: 146-154.