Variety support and exercise adherence behavior: experimental and mediating effects

Abstract

The purpose of this study was to examine the extent to which the provision of variety (i.e., variety support) is related to exercise behavior among physically inactive adults and the extent to which the ‘experience of variety’ mediates those effects. One hundred and twenty one inactive university students were randomly assigned to follow a high or low variety support exercise program for 6 weeks. Assessments were conducted at baseline, 3- and 6-weeks. Participants in the high variety support condition displayed higher levels of adherence to the exercise program than those in the low variety support condition [F(1, 116) = 5.55, p = .02, η
p2 = .05] and the relationship between variety support and adherence was mediated by perceived variety (β = .16, p < .01). Exercise-related variety support holds potential to be an efficacious method for facilitating greater exercise adherence behaviors of previously inactive people by fostering perceptions of variety.

Full text

Variety support and exercise adherence behavior: experimental
and mediating effects
Benjamin D. Sylvester
1
•Martyn Standage
2
•Desmond McEwan
1
•Svenja A. Wolf
1
•
David R. Lubans
3
•Narelle Eather
3
•Megan Kaulius
1
•Geralyn R. Ruissen
1
•
Peter R. E. Crocker
1
•Bruno D. Zumbo
4
•Mark R. Beauchamp
1
Received: July 23, 2015 / Accepted: October 7, 2015
ÓSpringer Science+Business Media New York 2015
Abstract The purpose of this study was to examine the
extent to which the provision of variety (i.e., variety sup-
port) is related to exercise behavior among physically
inactive adults and the extent to which the ‘experience of
variety’ mediates those effects. One hundred and twenty
one inactive university students were randomly assigned to
follow a high or low variety support exercise program for
6 weeks. Assessments were conducted at baseline, 3- and
6-weeks. Participants in the high variety support condition
displayed higher levels of adherence to the exercise pro-
gram than those in the low variety support condition [F(1,
116) =5.55, p=.02, g
p
2
=.05] and the relationship
between variety support and adherence was mediated by
perceived variety (b=.16, p\.01). Exercise-related
variety support holds potential to be an efficacious method
for facilitating greater exercise adherence behaviors of
previously inactive people by fostering perceptions of
variety.
Keywords Diverse Physical activity Resistance
training Mediation Perceived variety
Introduction
Researchers and public health agencies have consistently
identified that the vast majority of North American adults
are physically inactive (i.e., Centers for Disease Control
and Prevention 2014; Colley et al. 2011) and that physical
inactivity is linked to an increased risk for numerous causes
of morbidity (and mortality) such as cardiovascular disease
and some types of cancer [World Health Organization
(WHO) 2009]. To address the public health concern
ensuing from this global physical inactivity pandemic
(Hallal et al. 2012), there have been calls to develop effi-
cacious exercise intervention strategies (Ma
ˆsse et al. 2011;
WHO 2007).
One intervention strategy that holds potential for influ-
encing individuals’ exercise behavior relates to the provi-
sion of variety (e.g., Juvancic-Heltzel et al. 2013). Variety
refers to the experience of an assortment or alternation of
(novel and familiar) tasks, actions, and opportunities (cf.
Juvancic-Heltzel et al. 2013; Sheldon and Lyubomirsky
2012; Sylvester et al. 2014b). Variety has been examined
as both a feature of an activity or environment (i.e., variety
support; e.g., Lyubomirsky and Layous 2013), and as an
experience (i.e., one’s felt experience; e.g., Sylvester et al.
2014b). Variety support refers to the manner in which
activities, behaviors, and opportunities are structured to
facilitate (or thwart) the experience of variety, whereas the
experience of variety refers to the extent to which a person
feels as though they experience an assortment of tasks,
actions, and opportunities. In the present investigation, we
focus on both variety support and the experience of variety
in the context of exercise.
In previous work, Glaros and Janelle (2001) found that
participants who varied their use of aerobic exercise
equipment every fortnight for 8 weeks adhered to their
&Benjamin D. Sylvester
bsylvest@alumni.ubc.ca
1
Psychology of Exercise Health and Physical Activity Lab,
School of Kinesiology, The University of British Columbia,
122-6081 University Boulevard, Vancouver V6T 1Z1,
Canada
2
Department for Health, University of Bath, Bath,
England, UK
3
School of Education, The University of Newcastle,
Newcastle, Australia
4
Department of Educational and Conselling Psychology, and
Special Education, The University of British Columbia,
Vancouver, Canada
123
J Behav Med
DOI 10.1007/s10865-015-9688-4

exercise program more so than participants who did the
same aerobic exercise each session. In addition, Juvancic-
Heltzel et al. (2013) found that participants who encoun-
tered greater variety support in a single bout of exercise
(i.e., the opportunity to use ten vs two pieces of equipment)
spent more time exercising and performed a greater num-
ber of repetitions. In these studies the authors structured
exercise-related variety support by prescribing variation
both between sessions (i.e., changing the mode of exercise
from one session to another; Glaros and Janelle 2001) and
within a session (i.e., offering a greater number of exercises
in a single bout; Juvancic-Heltzel et al. 2013). Furthermore,
Dimmock et al. (2013) provided variety support within a
single exercise by instructing participants that the second
half of a cycling task would require different resources and
would be experienced differently than in the first half.
Although these studies provide insight in terms of how to
structure exercise-related variety support (e.g., Dimmock
et al. 2013), and the subsequent effect on exercise behavior
(i.e., Glaros and Janelle 2001; Juvancic-Heltzel et al.
2013), these studies examined exercise-related variety
support using atheoretical approaches which limits
researchers understanding of the process through which
(i.e., why/how) interventionists can change exercise
behavior (Rothman 2004).
One theory that provides insight in terms of the extent to
which (and manner through which) various contextual
factors lead to exercise behavior (i.e., through psycholog-
ical experiences) is self-determination theory (SDT; Deci
and Ryan 1985; Ryan and Deci 2002). Embedded within
SDT, Ryan and Deci (2002) posit that people have uni-
versal and innate basic psychological needs for compe-
tence, relatedness, and autonomy, and the extent to which
these needs are supported in one’s social environment leads
to subsequent behavior (through the intermediary role of
autonomous motivation). Competence refers to feeling
capable and effective in one’s environment (Ryan and Deci
2002; White 1959), relatedness refers to feeling connected
to others (Baumeister and Leary 1995; Ryan and Deci
2002), and autonomy refers to feelings of self-governance,
and volition in one’s choices and behaviors (deCharms
1968; Ryan and Deci 2002).
In their conception of SDT, Ryan and Deci (2002)
theorized that in any given context, the way in which that
context is structured will influence downstream psycho-
logical variables and subsequent behavior. While there is
mounting empirical evidence supporting the notion that
fostering satisfaction of the needs for competence, relat-
edness, and autonomy in exercise (through social support)
leads to exercise behavior (e.g., Teixeira et al. 2012),
Sheldon (2011) noted that a lack of research examining
alternative/additional psychological experiences that may
support adaptive behavior (in addition to satisfaction of the
needs for competence, relatedness and autonomy advanced
within SDT) is a limitation in the extant SDT literature, and
one that should be empirically examined.
The experience of variety may operate as a salient and
unique psychological experience worth investigating from
an SDT perspective. In the context of exercise, previous
work has found that perceived variety is empirically dis-
tinct from perceptions of competence, relatedness, and
autonomy (Sylvester et al. 2014b). Moreover, perceptions
of variety (in addition to satisfaction of basic psychological
needs for competence, autonomy, and relatedness) predict
variance in indices of exercise-related well-being (e.g.,
Sylvester et al. 2014b), motivation and exercise behavior
(Sylvester et al. 2014a). One of the notable limitations of
the studies by Sylvester et al. (2014a,b), however, is that
they used observational (i.e., non-experimental) designs,
which substantively limits inferences of causality. Drawing
from the work of Sylvester et al. (2014a,b), theorizing
from the perspective of SDT (Ryan and Deci 2002), as well
as observations by Sheldon (2011), the diversity (or
invariance) of exercises that one engages in (i.e., exercise-
related variety support), may act to facilitate the subse-
quent experience of variety in exercise, which in turn could
have substantive implications for exercise behavior.
Thus, in the present study, we sought to examine the
effects of experimentally manipulated variety support in a
resistance exercise program in relation to exercise adher-
ence behavior, while first examining the extent to which
variety support is differentially related to the experience of
variety, when considered in comparison to perceptions of
competence, relatedness, and autonomy. This initial step
was designed to provide evidence of discriminant validity,
whereby we hypothesized that the provision of variety
support in the context of exercise would result in changes
in perceived variety, but not in perceived competence,
relatedness, and autonomy. Beyond this manipulation
check, the main purpose of the study was to examine the
effects of variety support in relation to exercise behavior
and whether the experience of variety in exercise mediates
those effects. This line of enquiry was designed to shed
light on whether perceived variety acts as a psychological
experience (cf. Sheldon, 2011) that influences (and
explains the relationship between variety support and)
exercise behavior. Based on Ryan and Deci’s (2002) con-
ceptual framework and previous research (e.g., Glaros and
Janelle 2001; Juvancic-Heltzel et al. 2013; Sylvester et al.
2014a,b), we hypothesized that exercise-related variety
support would foster perceptions of variety (but not satis-
faction of the needs for competence, relatedness, and
autonomy) in exercise, as well as exercise adherence
behavior. Furthermore, we hypothesized that the relation-
ship between variety support and exercise adherence
behavior would be mediated through perceived variety.
J Behav Med
123

Methods
Participants
Following ethical approval from the first author’s institu-
tional research ethics board, a sample of university students
(n =144) between the ages of 17 and 38 years old were
recruited to participate in the study. To be eligible, par-
ticipants had to (a) be currently enrolled as a university
student, (b) be between the ages of 17 and 40 years old,
(c) be able to read and converse in English, (d) report no
health risks that would interfere with exercise (as identified
by responses to the Physical Activity Readiness Ques-
tionnaire for Everyone; PARQ+, Warburton et al. 2011),
and (e) be classified as physically inactive (i.e., report two
or fewer bouts, of at least 20 min, of moderate to vigorous
exercise in a typical week; cf. Wilcox et al. 1999).
The final sample (N=121) was comprised of 87
females (M
age
=20.87 years; SD
age
=3.09 years) and 34
males (M
age
=21.88 years; SD
age
=3.57 years). The
sample was ethnically diverse, as most participants self-
identified as Chinese (n=43; 35.5 %), White (n=32;
26.4 %), multi-racial (n=17; 14.1 %), or Korean (n=9;
7.4 %). Most participants lived on their own off-campus
(n=43; 35.5 %), in an on-campus residence (n=38;
31.4 %), or with family (n=35; 28.9 %) and reported
being in their third (n=36; 29.8 %), first (n=32;
26.4 %), second (n=28; 23.1 %), or fourth (n=20;
16.5 %) year of university.
Procedure
This study was conducted at a university fitness centre in
British Columbia, Canada. Participants attended an intro-
ductory session where they were briefed on the study pro-
tocol (e.g., that they could drop-in to complete the exercise
program at their convenience) and asked to provide written
informed consent. They subsequently provided baseline data
and were then randomly assigned (through a random number
generator) to either a high variety support (HVS) or low
variety support (LVS) exercise program (i.e., condition).
Participants were blinded to the program conditions. Trained
research assistants and employees (i.e., Certified Personal
Trainers) at the fitness centre supervised the exercise ses-
sions and monitored participants for safety and technique.
All participants were given the same exercise protocol
instructions with regard to exercise frequency, duration, and
intensity (e.g., three 1-h training sessions per week) and both
exercise programs were designed to target upper and lower
body muscle groups (e.g., chest, legs). To control for volume
and intensity, following a warm-up consisting of aerobic
exercise, dynamic stretching, and a light set for each exercise,
participants were instructed to perform sets of ten repetitions
(of each prescribed exercise) at a selected weight such that a
consecutive repetition (i.e., [10) would not be possible
without compromising proper technique. As such, to maintain
the same relative intensity between participants, the absolute
resistance for each given exercise was individually tailored.
Participants were provided with an exercise booklet (available
from the first author upon request) that had a printed copy of
their exercise program as well as information about the study
protocol and exercise techniques. The booklet remained at the
exercise facility throughout the study for participants to follow
their assigned program and record their attendance. Partici-
pants were asked to abstain from other strength-training
exercise programs over the course of the study (to avoid
compromising internal validity). Finally, participants in both
conditions completed measures of exercise-related perceived
variety and the psychological needs at two time points [i.e., at
baseline prior to commencing the exerciseprogram, and at the
end of week three (Time 2)] as part of the experimental
pretest-midtest-posttest control group design. Those who
received LVS served as the control group in this study.
Intervention
The exercise programs were designed to be as identical as
possible with the exception of the level of variety support
that was provided. Participants in each condition performed
the same number of exercise sets and repetitions, at the
same relative intensity. The volume of exercise (i.e., 160
total repetitions) was equal for both conditions and was
consistent with procedures developed by Sparkes and
Behm (2010) who outlined the provision of resistance
exercise programs for previously inactive adults within a
university setting. The rest intervals (i.e., one-minute)
between exercises and sets were also identical for each
group. In the HVS condition, however, participants com-
pleted varied resistance-based exercises (using machine
weights, free weights, and one’s own body weight) during
each session, while participants in the LVS condition
completed the same exercises each session. To foster the
experience of variety in exercise, participants in the HVS
group engaged in an exercise program designed to (a) al-
ternate exercises between sessions (cf. Glaros and Janelle
2001), (b) include more diverse exercises within each
session (while holding the total number of sets and repe-
titions in each session equal with the LVS group; cf.
Juvancic-Heltzel et al. 2013), and (c) vary within individ-
ual exercises by incorporating modifications (cf. Dimmock
et al. 2013). Those in the HVS condition had unique
combinations of eight exercises to perform each session
(two sets of each exercise), which was expected to con-
sistently support the experience of variety throughout the
study, while those in the LVS condition repeated the same
four exercises each session (four sets of each exercise).
J Behav Med
123

Sample size determination
G*Power 3 (Faul et al. 2007) software was used to conduct
an a priori power analysis to determine the total sample
size necessary for this study. The sample size was selected
based on our primary research question regarding the
extent to which experimentally manipulated variety sup-
port in the context of a resistance exercise program leads to
exercise adherence behavior in a sample of physically
inactive adults. We used G*power (Faul et al. 2007)to
determine that an ANCOVA with a=.05, a moderate
effect size (g
2
=.06) based on Glaros and Janelle (2001)
and Juvancic-Heltzel et al. (2013), and a conservative
power estimate (b=.80) requires a sample of N =128.
To answer our second research question regarding the
extent to which the experience of variety mediates that
relationship, we used a single mediation model with a
latent variable of the mediator (with the independent
variable operationalized as an observed variable reflecting
the two experimental conditions and the dependent variable
as an observed measure of attendance). For structural
equation modeling (SEM), several researchers suggest at
least 5 or 10 observations per estimated parameter (Bentler
and Chou 1987; Bollen 1989); 10 parameters were esti-
mated in our mediation model (i.e., using a conservative
approach based on these recommendations, a sample of
100 was required). Others have provided more omnibus
recommendations for sample size estimates with SEM,
such as suggesting samples of at least 200 (Kline 2005).
However, sample size depends on many factors such as the
size of the model (e.g., number of parameters) and the
estimated size of effects, with researchers also recently
advocating that sample size estimates for SEM models can
be smaller in instances with less measurement error (e.g.,
Wolf et al. 2013). We reduced the risk of Type 1 error by
creating a latent variable (to reduce measurement error)
and used bootstrapping procedures to estimate indirect
effects (Preacher and Hayes 2008). Bootstrapping analysis
is recommended to test for mediation with small sample
sizes (e.g., Fritz and MacKinnon 2007; Shrout and Bolger
2002). When taken together, our a priori sample of
n=144 was deemed appropriate to address both our pri-
mary (effects of variety support on physical activity) and
secondary (mediation) research questions, while account-
ing for modest attrition (final sample n =121).
Measures
Perceived variety in exercise
Perceived variety in exercise was assessed using the five-
item Perceived Variety in Exercise (PVE) questionnaire
(Sylvester et al. 2014b). Items on the PVE questionnaire
are anchored on a six-point Likert-type rating scale with
responses ranging from 1 (False)to6(True). Higher scores
reflect greater levels of perceived variety in exercise. In
their original instrument development work, Sylvester et al.
(2014b) reported ordinal composite reliability (Zumbo
et al. 2007) of PVE scores to be .97. In the current study,
ordinal composite reliability of the PVE scores was .91 at
Time 1 and .94 at Time 2.
Basic psychological needs satisfaction
The Psychological Needs Satisfaction in Exercise (PNSE)
questionnaire (Wilson et al. 2006) was used to measure the
satisfaction of the needs for competence, relatedness, and
autonomy in the context of exercise. The PNSE is an 18-item
instrument with each of the three psychological needs mea-
sured using six items. Responses to eachitem are anchored on
a scale that ranges from 1 (False)to6(True). Higher scores
reflect greater satisfaction of the needs for (perceived) com-
petence, relatedness, and autonomy in exercise. Structural and
criterion validity of scores derived from an adult population
regarding each subscale of the PNSE was initially reported by
Wilson et al. (2006). In the current study, ordinal composite
reliability was found to be C.87 for each of the psychological
needs at both Time 1 and Time 2 (see Table 1).
Exercise behavior
Exercise behavior was operationalized as the percentage of
recorded adherence to the exercise program over the
6-week period. For each exercise session the participants
attended (up to 18 sessions over 6-weeks), they recorded
whether they completed the prescribed exercises in their
exercise booklets. Adherence was calculated as a percent-
age of sessions completed (i.e., total number of sessions
completed, divided by the maximum number of sessions
(i.e., 18), and multiplied by 100). This variable was used as
the dependent measure of exercise behavior.
Exercise behavior at baseline (i.e., Time 1) was mea-
sured using the Godin Leisure Time Exercise Question-
naire (GLTEQ; Godin and Shephard 1985). The GLTEQ is
comprised of 3-items that assess the frequency of mild,
moderate, and strenuous leisure-time exercise behavior
enduring at least 15 min per session in a typical week. A
score was calculated using the formula [(Mild 93) +
(Moderate 95) + (Strenuous 99)] to produce weekly
estimates of leisure-time exercise, with higher scores
reflecting higher levels of energy expenditure (Godin
2011). Godin and Shephard (1985) reported support for the
validity evidence of adult’s GLTEQ scores in the form of
positive correlations with estimates of cardiorespiratory
fitness (i.e., VO
2
max) and negative correlations with body
fat scores. Score stability has been examined through test–
J Behav Med
123

retest reliability coefficients, which have been found to
range from .24 to .96 (Godin and Shephard 1985; Jacobs
et al. 1993).
Data analysis
In line with our study objective to examine the efficacy of
exercise-related variety support on exercise program
adherence, participants who attended at least one exercise
session (i.e., received the variety support) were included in
the analysis. Prior to the main analyses, descriptive data
were obtained and Little’s Chi-square test (Little 1988) was
conducted to examine any potential patterns of missing
data.
Next, to examine whether exercise-related variety sup-
port differentially leads to the experience of variety, and/or
satisfaction of the needs for competence, relatedness, and
autonomy in exercise, we examined a latent variable
multivariate analysis of covariance (LVMANCOVA) using
Mplus 6.11 software. The latent model was utilized to
(a) treat the PVE and PNSE data as ordinal, (b) reduce
potential bias from measurement error, (c) estimate the
model simultaneously and therefore reduce the risk of Type
1 error, and (d) provide sufficient degrees of freedom in the
model. Weighted least squares means and variance-ad-
justed (WLSMV) method of estimation was used to
account for the ordered categorical nature of the Likert-
type response scale scores (Finney and DiStefano 2006).
To model ordinal data, a polychoric correlation matrix is
considered to be the best option when there are less than
seven response options (cf. Beauducel and Herzberg 2006).
Missing data were estimated using all of the available data
via the WLSMV algorithm within Mplus 6.11.
Based on recommendations by Brown (2006), Hu and
Bentler (1999), and Marsh et al. (2004), goodness of fit for
the model was assessed using the v
2
goodness of fit index,
the comparative fit index (CFI), Tucker-Lewis index (TLI),
and the root mean square error of approximation
(RMSEA). CFI and TLI values [.90, and RMSEA values
\.08 were considered to indicate good model-data fit,
whereas CFI and TLI values [.95, and RMSEA values
\.06 were considered to indicate excellent fit (cf. Hu and
Bentler 1998,1999). In addition to fit indices, we examined
the reliability of the scores through composite reliability
(CR) where scores from each item are individually
weighted in the composite load (see Bollen 1989; Fornell
and Larcker 1981). Ordinal composite reliability is based
on the polychoric correlation matrix and was assessed to
account for the Likert-type response format used in the
PVE and PNSE measures (Zumbo et al. 2007).
The LVMANCOVA was used to examine whether
exercise-related variety support influenced perceptions of
variety, competence, relatedness, and/or autonomy in
exercise at Time 2, controlling for within-person (base-
line) scores of perceived variety, competence, relatedness,
and autonomy in exercise at Time 1. The experimental
condition was the independent variable, while latent
variables were constructed using multiple categorical
items regarding perceived variety (five items), satisfaction
of the needs for competence (six items), relatedness (six
items), and autonomy (six items). Time 2 scores of per-
ceived variety, competence, relatedness, and autonomy in
exercise were the dependent variables, and baseline
scores of those variables at Time 1 were specified as
covariates.
On the basis of the finding that the intervention resulted
in changes in perceived variety, but not the three psycho-
logical needs (see Results section), we subsequently con-
ducted an analysis of covariance (ANCOVA) to assess
whether there were differences in adherence to the exercise
Table 1 Correlations and reliability estimates of study variables
Variable CR 1 2 34567891011
1. Variety support –
2. Variety-T1 .91 -.02 –
3. Competence-T1 .94 -.08 .67* –
4. Relatedness-T1 .92 -.11 .49* .54* –
5. Autonomy-T1 .87 -.09 .52* .59* .47* –
6. Variety-T2 .94 .42* .56* .37* .27* .29* –
7. Competence-T2 .94 .05 .46* .68* .37* .40* .60* –
8. Relatedness-T2 .95 -.03 .20 .21 .40* .19 .23 .40* –
9. Autonomy-T2 .94 .03 .36* .41* .33* .69* .58* .64* .17 –
10. Exercise behavior-T1 -.12 .11 .09 -.07 .17 -.08 .06 .07 -.04 –
11. Exercise adherence .20* .02 .01 .02 .13 .39* .19* .25* .01 -.05 –
CR composite reliability, T1 Time 1, T2 Time 2
*p\.05
J Behav Med
123

program based on the provision of (high or low) exercise-
related variety support. To examine this research question,
the experimental condition was specified as the indepen-
dent variable, scores of exercise program adherence over
the 6-week intervention was specified as the dependent
variable, and gender and baseline scores of exercise
behavior at Time 1 were specified as covariates. Exercise
adherence behavior was operationalized as an observed
variable.
Finally, through a structural equation model we exam-
ined whether receiving exercise-related variety support
explains variance in adherence to the 6-week exercise
program, through the mediating role of perceived variety in
exercise (measured at Time 2). To examine the full range
of adherence scores, we included participants who either
completed or dropped out of the intervention at any time
(i.e., before or after Time 2 data collection). For those
participants who had dropped out of the study before we
measured their perceived variety in exercise at Time 2, we
imputed the last value obtained from those participants as a
conservative estimate (i.e., no manipulation) of the par-
ticipant’s perceived variety in exercise (cf. intention-to-
treat analysis recommendations; Unnebrink and Windeler
2001). Specifically, we used scores of perceived variety in
exercise available from Time 2 (n=88) as the mediator,
but if the participant had dropped out of the study by this
point, we carried forward their score from Time 1 (n=33;
n
HVS
=13; n
LVS
=20) to retain their adherence data in the
model.
In line with Rucker et al. (2011) recommendations for
testing mediation, the main outcome of interest was the
indirect effect of exercise-related variety support on exer-
cise program adherence through perceived variety in
exercise. The indirect effect was estimated using Preacher
and Hayes’ (2008) bootstrapping procedure (k=5000
samples) to construct bias-corrected 95 % confidence
intervals (CIs). Bootstrapping is a non-parametric resam-
pling procedure recommended for estimating indirect
effects and CIs, and to optimize statistical power (Preacher
and Hayes 2008).
Results
Of the 144 people who attended the baseline appointment,
121 participants received the exercise-related variety sup-
port manipulation (i.e., n
HVS
=58; n
LVS
=63) by attend-
ing at least one exercise session, and were subsequently
included in the analyses. Examination of Little’s (1988)
test indicated that missing data were Missing Completely at
Random (MCAR), v
2
(502) =519.20, p=.289.
Descriptive statistics for exercise adherence were as
follows: M=56.80 %; SD =30.71; skewness =-.144
(SE =.220); kurtosis =-1.304 (SE =.437). Results
from the LVMANCOVA showed that overall, the model
had good fit, v
2
(1015) =1292.24, p\.00, CFI =.96,
TLI =.96, RMSEA =.06. Correlations and CR values for
the study variables are presented in Table 1. In the
LVMANCOVA (see Table 2), after statistically controlling
for baseline scores of exercise-related perceived variety,
competence, relatedness, and autonomy, there was a sta-
tistically significant intervention effect on perceived vari-
ety (b=.47, p\.001), but not perceived competence
(b=.05, p[.05), relatedness (b=-.04, p[.05), or
autonomy (b=.03, p[.05) at Time 2.
An ANCOVA was then conducted to examine potential
differences between exercise-related variety support con-
ditions with regard to exercise program adherence, after
statistically controlling for gender and baseline scores of
exercise behavior as covariates. There was a significant
intervention effect on adherence to the program F(1,
116) =5.55, p=.02, g
p
2
=.05, after statistically con-
trolling for gender F(1, 116) =0.01, p[.05, g
p
2
=.00,
and exercise behavior at Time 1 F(1, 116) =0.05, p[.05,
g
p
2
=.00. Participants who received high variety support
had greater exercise adherence than those who received
low variety support, M
HVS
=64.22 %, SD =30.99;
M
LVS
=50.89 %, SD =28.80 (see Fig. 1). That is, on
average participants in the high variety support group
completed 11.56 exercise sessions, whereas participants in
the low variety support group completed only 9.16 exercise
sessions, on average (out of 18).
Table 2 Intervention effects on perceived variety, competence, relatedness, and autonomy
Variables Standardized
estimates
Unstandardized
estimates
SE Bootstrapped 95 %
confidence interval
Effects of variety support on
Perceived variety (T2) .474 .897 .177 [.668,1.338]
Competence (T2) .053 .085 .179 [-.228, .458]
Relatedness (T2) -.036 -.056 .161 [-.341, .321]
Autonomy (T2) .028 .034 .128 [-.233, .282]
Boldface confidence intervals do not contain 0. T2 Time 2
J Behav Med
123

Finally, we examined whether perceived variety in
exercise mediated the relationship between exercise-related
variety support and adherence to the exercise program,
after statistically controlling for gender and baseline scores
of exercise behavior. Overall, the model had excellent fit,
v
2
(23) =21.56, p=.55, CFI =1.00, TLI =1.00,
RMSEA =.00. In the structural model, exercise-related
variety support positively predicted variance in perceived
variety in exercise at Time 2 (b=.43, p\.001), which
subsequently predicted variance in adherence to the 6-week
exercise program (b=.37, p=.001; see Fig. 2; Table 3).
Neither exercise behavior at Time 1 (b=-.02, p[.05), or
gender (b=.01, p[.05), was found to be a statistically
significant covariate of exercise adherence behavior over
the course of the 6-week program. The indirect effect was
found to be significant for the relationship between exercise-
related variety support and adherence to the program,
through perceived variety in exercise (b=.16, p\.01).
After statistically controlling for the effects of perceived
variety (i.e., the mediator), the direct effect of variety sup-
port in relation to exercise adherence was non-significant
(b=.07, p[.05), which provided evidence of mediation.
Discussion
The purpose of this study was to examine the extent to
which experimentally manipulated variety support in the
context of a resistance exercise program leads to the
experience of variety (when examined alongside satisfac-
tion of the three psychological needs embedded within
SDT; Ryan and Deci 2002), and exercise adherence
behavior in a sample of physically inactive adults. We also
sought to examine whether the relationship between variety
support and exercise adherence behavior is mediated by
perceived variety. The results showed that receiving high
(compared to low) exercise-related variety support led to
higher perceived variety in exercise, but not satisfaction of
the needs for competence, relatedness, or autonomy, three
weeks later. Furthermore, higher variety support led to an
increase in exercise adherence behavior over the course of
six weeks and the relationship between variety support and
exercise adherence was mediated by perceptions of variety.
Fig. 1 Dependent variable scores after controlling for gender and
exercise behavior at Time 1. Estimated marginal means are reported
(i.e., the mean values at post-test include controlling for pre-test
measures). Bars denote standard errors. *p\.01
Perceived
Variety (T2)
Exercise
Adherence
.43*
Variety
Support
R2= .16
R2= .18
.37*
Gender
Exercise
Behavior
(T1)
Fig. 2 Path diagram of the relationships between exercise-related
variety support (i.e., condition) and exercise adherence via perceived
variety in exercise at Time 2 (T2) after controlling for gender and
exercise behavior at Time 1 (T1) as covariates. Solid lines represent
standardized significant path coefficients and dashed lines represent
non-significant path coefficients. *p\.01
J Behav Med
123

The positive effects of variety support on exercise
adherence behavior are consistent with findings reported by
Glaros and Janelle (2001) who found evidence that pro-
viding variety support (i.e., switching the mode of exercise
in an aerobic exercise program every two weeks) leads to
improved exercise adherence when compared to thwarting
variety support (i.e., by prescribing only one mode of
aerobic exercise for eight weeks). Adherence rates in the
current study were similar to those reported by Glaros and
Janelle (2001), as participants in the high and low variety
support conditions in the current study attended 64 and
51 % of exercise sessions respectively, compared to 63 and
54 % in Glaros and Janelle’s (2001) study. The results of
our study substantively extend this work by explicating a
mechanism through which variety support fosters
improvements in exercise behavior. Specifically, from an
SDT perspective, Ryan and Deci (2002) posit that satis-
faction of the needs for competence, relatedness, and
autonomy represent the most salient psychological expe-
riences (i.e., needs) through which well-being, motivation
and achievement behavior are supported. The results of the
current study provide experimental evidence in support of
the contention that perceived variety may act as an addi-
tional psychological experience (cf. Sheldon 2011; Sylve-
ster et al. 2014a,b) that might bring about improved
exercise behavior. Indeed, these findings may have theo-
retical implications, as the experience of variety may be an
additional type of positive experience (beyond satisfaction
of the three basic psychological needs within SDT) that is
involved in supporting adherence behavior.
Although the mediation analysis provides valuable
insight regarding how and why the intervention had an
effect on exercise adherence, plausible additional expla-
nations exist and should be noted. For example, researchers
have found empirical evidence that perceptions of variety
explain the prospective relationship between perceived
variety in exercise and exercise behavior through the
mediating role of autonomous motivation (i.e., Sylvester
et al. 2014a). Although we did not test an extended mul-
tiple mediation model that also included autonomous
motivation (e.g., interest/enjoyment inherent within an
activity; Ryan and Deci 2002), future research is warranted
that examines the following sequence: exercise-related
variety support ?perceived variety in exercise ?au-
tonomous motivation toward exercise ?exercise behav-
ior. It is also entirely possible that the absence of variety (in
the LVS condition) may have resulted in reduced adher-
ence via other (physical/physiological) mechanisms.
Because our sample was comprised of non-exercisers, we
recognized the importance of buffering against factors such
as delayed onset muscle soreness (DOMS; Smith 1992),
and we did so by following a 3-days-per-week resistance
exercise program protocol that was previously conducted
with inactive adults within a university setting (Sparkes
and Behm 2010). However, it is entirely possible that
participants in this study may have experienced different
levels of muscle soreness based on the condition to which
they were assigned (due to the specificity of exercises), that
in turn may have differentially contributed to their moti-
vation and attendance of subsequent sessions. When taken
together, although perceptions of variety were found to
mediate the effects of variety support in relation to exercise
adherence, other mechanisms warrant greater scrutiny in
future research.
Nevertheless, from an applied perspective, the results of
this study suggest that the provision of a varied exercise
program may be an efficacious intervention strategy for
those concerned with health promotion that can influence
exercise adherence behavior. In this study, we successfully
manipulated the experience of variety through three
Table 3 Direct and indirect effects of variety support, covariates, and perceived variety on exercise adherence
Variables Standardized
estimates
Unstandardized
estimates
SE Bootstrapped
95 %
confidence
interval
Direct effects on exercise adherence
Variety support .065 3.918 5.408 [-6.568, 14.332]
Perceived variety (T2) .365 11.460 3.250 [5.481,18.251]
Exercise behavior (T1) -.019 -.029 .204 [-.247, .465]
Gender .010 .673 6.118 [-11.285, 12.775]
Direct effect on perceived variety
Variety support .425 .821 .181 [.461, 1.166]
Indirect effect of variety support on exercise adherence
Perceived variety (T2) .155 9.405 3.120 [4.421, 17.246]
Boldface confidence intervals do not contain 0. T1 Time 1, T2 Time 2
J Behav Med
123

modalities, namely, varying the exercise activities between
sessions (i.e., from bout to bout), within sessions (i.e.,
prescribing eight vs four exercise activities), as well as
prescribing variation within the exercises (e.g., progres-
sions). Within the current research it was not our aim to
elucidate the relative importance of each discrete (i.e.,
micro) method of providing variety support, and instead
took a more macro approach to maximizing variety in the
HVS condition by operationalizing variety support between
sessions, within sessions, and within exercises. Neverthe-
less, we certainly recognize that in future work it would be
particularly informative for researchers to disentangle the
unique effects of each method of variety support in their
own right.
While the experimental design and the mediation anal-
ysis are notable strengths of the study, limitations should
also be acknowledged. Specifically, the mediation results
should be interpreted with some caution due to the fact that
some participants dropped out of the study during the first
3-weeks (n=33). As a conservative approach, we used
intent-to-treat analytic procedures (cf. Unnebrink and
Windeler 2001), by carrying forward the Time 1 scores of
the mediator for those participants. Although such an
approach is preferable to listwise deletion (Unnebrink and
Windeler 2001), we recognize that dropout from any
intervention study represents a challenge to internal
validity of a study’s findings (Shadish et al. 2002). A
second limitation corresponds to the relative short-term
nature of the exercise program. Although we used a
prospective experimental design, and found significant
effects with regard to the efficacy of providing variety
support in facilitating exercise behavior over time, it
should be noted that the program was limited to 6 weeks.
Given that sustained participation is required to achieve
sufficient health outcomes (Physical Activity Guidelines
Advisory Committee 2008), future research is required to
examine the efficacy of such (varied) exercise programs
over a much longer period of time (i.e., 6 months or more).
Additionally, caution should be exercised in generaliz-
ing the findings beyond physically inactive adults.
Although inactive adults represent an important population
for intervention (Centers for Disease Control and Preven-
tion 2014), in future, researchers are encouraged to
examine the external validity of variety support as a means
of intervention in relation to physical activity adherence
behaviors with other populations and in different contexts
(e.g., physical education in schools, community exercise
programs, adherence to rehabilitation protocols). Finally,
while the present study sought to examine a psychological
mediator (perceived variety) of the relations between
variety support and exercise adherence, future work should
also examine potential moderators (i.e., boundary condi-
tions) that might interact with variety support in relation to
supporting physical achievement outcomes. Such modera-
tors might include variables such as age (children vs adults)
and dose of variety. For example, an obvious question is
how much variety is optimal to support exercise adherence?
Summary
In conclusion, the provision of exercise-related variety
support was found to result in improved levels of exercise
adherence among a sample of inactive university students,
when compared to those who took part in a low variety
support exercise program. Furthermore, participants’ per-
ceptions of ‘felt’ variety acted as the psychological
mechanism that drove this effect. When taken together the
results point to the potential utility of variety support as an
efficacious strategy for supporting exercise behavior, with
further research now required to examine the long-term
(C6 months) effects of this intervention strategy in sup-
porting health-enhancing physical activity, potential mod-
erators, as well as the external validity of this approach
with other populations and in different contexts.
Acknowledgments This research was supported by a graduate
scholarship awarded to Ben Sylvester by the Social Sciences and
Humanities Research Council of Canada, as well as a career inves-
tigator award from the Michael Smith Foundation for Health
Research awarded to Mark Beauchamp.
Compliance with ethical standards
Conflict of interest Benjamin D. Sylvester, Martyn Standage,
Desmond McEwan, Svenja A. Wolf, David R. Lubans, Narelle
Eather, Megan Kaulius, Geralyn R. Ruissen, Peter R. E. Crocker,
Bruno D. Zumbo, Mark R. Beauchamp declare that they have no
conflict of interest.
Human and animal rights and Informed consent All procedures
followed were in accordance with ethical standards of the responsible
committee on human experimentation (institutional and national) and
with the Helsinki Declaration of 1975, as revised in 2000. Informed
consent was obtained from all participants for being included in the
study.
References
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire
for interpersonal attachments as a fundamental human motiva-
tion. Psychological Bulletin, 117, 497–529. doi:10.1037/0033-
2909.117.3.497
Beauducel, A., & Herzberg, P. Y. (2006). On the performance of
maximum likelihood versus means and variance adjusted
weighted least squares estimation in CFA. Structural Equa-
tion Modeling, 13, 186–203. doi:10.1207/s15328007sem1302_2
Bentler, P. M., & Chou, C. H. (1987). Practical issues in structural
modeling. Sociological Methods & Research, 16, 78–117.
Bollen, K. A. (1989). Structural equations with latent variables. New
York: Wiley.
J Behav Med
123

Brown, A. (2006). Confirmatory factor analysis for applied research.
New York, NY: Guilford Press.
Centers for Disease Control and Prevention (2014). Health, United
States, 2014. Available from: http://www.cdc.gov/nchs/data/hus/
hus14.pdf#063
Colley, R. C., Garriguet, D., Janssen, I., Craig, C. L., Clarke, J., &
Tremblay, M. S. (2011). Physical activity of Canadian adults:
Accelerometer results from the 2007 to 2009 Canadian health
measures survey. Ottawa, ON: Statistics Canada.
deCharms, R. (1968). Personal causation: The internal affective
determinants of behavior. New York, NY: Academic Press.
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-
determination in human behavior. New York: Plenum.
Dimmock, J., Jackson, B., Podlog, L., & Magaraggia, C. (2013). The
effect of variety expectations on interest, enjoyment, and locus
of causality in exercise. Motivation & Emotion, 37, 146–153.
doi:10.1007/s11031-012-9294-5
Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*Power
3: A flexible statistical power analysis program for the social,
behavioral, and biomedical sciences. Behavior Research Meth-
ods, 39, 175–191.
Finney, S. J., & DiStefano, C. (2006). Non-normal and categorical
data in structural equation modeling. In G. R. Hancock & R.
O. Mueller (Eds.), Structural equation modeling: A second
course (pp. 269–314). Charlotte, NC: Information Age Publish-
ing Inc.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation
models with unobservable variables and measurement error:
Algebra and statistics. Journal of Marketing Research, 18,
39–50. doi:10.2307/3151312
Fritz, M. S., & MacKinnon, D. P. (2007). Required sample size to
detect the mediated effect. Psychological Science, 18, 233–239.
Glaros, N., & Janelle, M. (2001). Varying the mode of cardiovascular
exercise to increase adherence. Journal of Sport Behavior, 24,
42–62.
Godin, G. (2011). The Godin-Shephard leisure-time physical activity
questionnaire. The Health & Fitness Journal of Canada, 4,
18–22.
Godin, G., & Shephard, R. J. (1985). A simple method to assess
exercise behavior in the community. Canadian Journal of
Applied Sport Sciences, 10, 141–146.
Hallal, P. C., Andersen, L. B., Bull, F. C., Guthold, R., Haskell, W.,
Ekelund, U., & Lancet Physical Activity Series Working Group.
(2012). Global physical activity levels: surveillance progress,
pitfalls, and prospects. Lancet, 380, 247–257.
Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure
analysis: Sensitivity to underparameterized model misspecifica-
tion. Psychological Methods, 3, 424–453. doi:10.1037/1082-
989X.3.4.424
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in
covariance structure analysis: Conventional criteria versus new
alternatives. Structural Equation Modeling: A Multidisciplinary
Journal, 6, 1–55. doi:10.1080/10705519909540118
Jacobs, D. R., Ainsworth, B. E., Hartman, T. J., & Leon, A. S. (1993).
A simultaneous evaluation of 10 commonly used physical
activity questionnaires. Medicine and Science in Sports and
Exercise, 25, 81–91. doi:10.1249/00005768-199301000-00012
Juvancic-Heltzel, J. A., Glickman, E. L., & Barkley, J. E. (2013). The
effect of variety on physical activity: A cross-sectional study.
The Journal of Strength & Conditioning Research, 27, 244–251.
doi:10.1519/JSC.0b013e3182518010
Kline, R. B. (2005). Principles and practice of structural equation
modelling (2nd ed.). New York: Guilford Press.
Little, R. J. A. (1988). A test of missing completely at random for
multivariate data with missing values. Journal of the American
Statistical Association, 83, 1198–1202. doi:10.1080/01621459.
1988.10478722
Lyubomirsky, S., & Layous, K. (2013). How do simple positive
activities increase well-being? Current Directions in Psycholog-
ical Science, 22, 57–62. doi:10.1177/0963721412469809
Marsh, H. W., Hau, K., & Wen, Z. (2004). In search of golden rules:
Comments on hypothesis-testing approaches to seeing cut-off
values for fit indexes and dangers in overgeneralizing Hu and
Bentler’s (1999) findings. Structural Equation Modeling, 11,
320–341. doi:10.1207/s15328007sem1103_2
Ma
ˆsse, L. C., Nigg, C. R., Basen-Engquist, K., & Atienza, A. A.
(2011). Understanding the mechanism of physical activity
behavior change: Challenges and a call for action. Psychology
of Sport and Exercise, 12, 1–6. doi:10.1016/j.psychsport.2010.
07.011
Physical Activity Guidelines Advisory Committee. (2008). Physical
activity guidelines advisory committee report. Washington DC:
U.S. Department of Health and Human Services.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling
strategies for assessing and comparing indirect effects in
multiple mediator models. Behavior Research Methods, 40,
879–891. doi:10.3758/BRM.40.3.879
Rothman, A. J. (2004). ‘‘Is there nothing more practical than a good
theory?: Why innovations and advances in health behavior
change will arise is interventions are used to test and refine
theory. International Journal of Behavioral Nutrition and
Physical Activity, 1, 1–7. doi:10.1186/1479-5868-1-11
Rucker, D. D., Preacher, K. J., Tormala, Z. L., & Petty, R. E. (2011).
Mediation analysis in social psychology: Current practices and
new recommendations. Social and Personality Psychology
Compass, 5, 359–371. doi:10.1111/j.1751-9004.2011.00355.x
Ryan, R. M., & Deci, E. L. (2002). An overview of self-determination
theory. In E. L. Deci & R. M. Ryan (Eds.), Handbook of self-
determination research (pp. 3–33). Rochester, NY: University of
Rochester Press.
Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental
and quasi-experimental designs for generalized causal inference.
Boston, MA: Houghton Mifflin.
Sheldon, K. M. (2011). Integrating behavioral-motive and experien-
tial-requirement perspectives on psychological needs: A two
process model. Psychological Review,118, 552–569. doi:10.
1037/a0024758
Sheldon, K. M., & Lyubomirsky, S. (2012). The challenge of staying
happier: Testing the Hedonic Adaptation Prevention Model.
Personality & Social Psychological Bulletin, 38, 670–680.
doi:10.1177/0146167212436400
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and
nonexperimental studies: New procedures and recommendations.
Psychological Methods, 7, 422–445.
Smith, L. L. (1992). Causes of delayed onset muscle soreness and the
impact on athletic performance: A review. Journal of Applied
Sport Science Research, 6, 135–141.
Sparkes, R., & Behm, D. G. (2010). Training adaptations associated
with an 8-week instability resistance training program with
recreationally active individuals. Journal of Strength & Condi-
tioning Research, 24, 1931–1941.
Sylvester, B. D., Standage, M., Ark, T., Sweet, S. N., Crocker, P.
R. E., Zumbo, B. D., & Beauchamp, M. R. (2014a). Is variety a
spice of (an active) life?: Perceived variety, exercise behavior,
and the mediating role of autonomous motivation. Journal of
Sport & Exercise Psychology, 36, 516–527. doi:10.1123/jsep.
2014-0102
Sylvester, B. D., Standage, M., Dowd, A. J., Martin, L. J., Sweet, S.
N., & Beauchamp, M. R. (2014b). Perceived variety, psycho-
logical needs satisfaction, and exercise-related well-being.
J Behav Med
123

Psychology & Health, 29, 1044–1061. doi:10.1080/08870446.
2014.907900
Teixeira, P. J., Carrac¸a, E. V., Markland, D., Silva, M. N., & Ryan, R.
M. (2012). Exercise, physical activity, and self-determination
theory: A systematic review. International Journal of Behavioral
Nutrition & Physical Activity, 9, 78. doi:10.1186/1479-5868-9-
78
Unnebrink, K., & Windeler, J. (2001). Intention-to-treat: Methods for
dealing with missing values in clinical trials of progressively
deteriorating diseases. Statistics in Medicine, 20, 3931–3946.
doi:10.1002/sim.1149
Warburton, D. E. R., Jamnik, V. K., Bredin, S. S. D., & Gledhill, N.
(2011). The physical activity readiness questionnaire for every-
one (PAR-Q+) and electronic physical activity readiness medical
examination (ePARmed- X+). Health & Fitness Journal of
Canada, 4, 3–17.
White, R. W. (1959). Motivation reconsidered: The concept of
competence. Psychological Review, 66, 297–333. doi:10.1037/
h0040934
Wilcox, S., King, A. C., Brassignton, G. S., & Ahn, D. K. (1999).
Physical activity preferences of middle-aged and older adults: A
community analysis. Journal of Aging & Physical Activity, 7,
386–399.
Wilson, P. M., Rogers, W. T., Rodgers, W. M., & Wild, T. C. (2006).
The psychological need satisfaction in exercise scale. Journal of
Sport & Exercise Psychology, 28, 231–251.
Wolf, E. J., Harrington, K. M., Clark, S. L., & Miller, M. W. (2013).
Sample size requirements for structural equation models an
evaluation of power, bias, and solution propriety. Educational
and Psychological Measurement, 73, 913–934. doi:10.1177/
0013164413495237
World Health Organization. (2007). A guide for population-based
approaches to increasing levels of physical activity: Implemen-
tation of the WHO global strategy on diet, physical activity and
health. Geneva: World Health Organization.
World Health Organization. (2009). Global health risks: Mortality and
burden of disease attributable to selected major risks. Retrieved
from: http://www.who.int/healthinfo/global_burden_disease/
GlobalHealthRisks_report_full.pdf
Zumbo, B. D., Gadermann, A. M., & Zeisser, C. (2007). Ordinal
versions of coefficients alpha and theta for likert rating scales.
Journal of Modern Applied Statistical Methods, 6, 21–29.
J Behav Med
123