ArticlePDF AvailableLiterature Review

The effectiveness of multi-component goal setting interventions for changing physical activity behaviour: A systematic review and meta-analysis

Authors:

Abstract

Drawing from goal setting theory (Latham & Locke, 1991; Locke & Latham, 2002; Locke et al., 1981), the purpose of this study was to conduct a systematic review and meta-analysis of multi-component goal setting interventions for changing physical activity (PA) behaviour. A literature search returned 41,038 potential articles. Included studies consisted of controlled experimental trials wherein participants in the intervention conditions set PA goals and their PA behaviour was compared to participants in a control group who did not set goals. A meta-analysis was ultimately carried out across 45 articles (comprising 52 interventions, 126 effect sizes, n = 5912) that met eligibility criteria using a random-effects model. Overall, a medium, positive effect (Cohen's d(SE) = .552(.06), 95% CI = .43-.67, Z = 9.03, p<.001) of goal setting interventions in relation to PA behaviour was found. Moderator analyses across 20 variables revealed several noteworthy results with regard to features of the study, sample characteristics, PA goal content, and additional goal-related behaviour change techniques. In conclusion, multi-component goal setting interventions represent an effective method of fostering PA across a diverse range of populations and settings. Implications for effective goal setting interventions are discussed. Supplemental Material Supplemental Material.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=rhpr20
Download by: [The University of British Columbia] Date: 13 October 2015, At: 08:13
Health Psychology Review
ISSN: 1743-7199 (Print) 1743-7202 (Online) Journal homepage: http://www.tandfonline.com/loi/rhpr20
The effectiveness of multi-component goal setting
interventions for changing physical activity
behaviour: A systematic review and meta-analysis
Desmond McEwan, Samantha M. Harden, Bruno D. Zumbo, Benjamin D.
Sylvester, Megan Kaulius, Geralyn R. Ruissen, A. Justine Dowd & Mark R.
Beauchamp
To cite this article: Desmond McEwan, Samantha M. Harden, Bruno D. Zumbo, Benjamin
D. Sylvester, Megan Kaulius, Geralyn R. Ruissen, A. Justine Dowd & Mark R. Beauchamp
(2015): The effectiveness of multi-component goal setting interventions for changing physical
activity behaviour: A systematic review and meta-analysis, Health Psychology Review, DOI:
10.1080/17437199.2015.1104258
To link to this article: http://dx.doi.org/10.1080/17437199.2015.1104258
View supplementary material Accepted online: 07 Oct 2015.
Submit your article to this journal Article views: 8
View related articles View Crossmark data
Accepted Manuscript
RUNNING HEAD: Goal Setting Interventions
The effectiveness of multi-component goal setting interventions for changing physical
activity behaviour: A systematic review and meta-analysis
Desmond McEwan1, Samantha M. Harden1,2, Bruno D. Zumbo3, Benjamin D. Sylvester1, Megan
Kaulius1, Geralyn R. Ruissen1, A. Justine Dowd4, and Mark R. Beauchamp1
1School of Kinesiology, University of British Columbia, Vancouver, Canada.
2Department of Human Nutrition, Foods, & Exercise, Virginia Tech, Blacksburg, USA.
3Faculty of Education, University of British Columbia, Vancouver, Canada.
4School of Exercise & Health Sciences, University of British Columbia Okanagan, Kelowna,
Canada.
Address correspondence to: Mark Beauchamp, School of Kinesiology, War Memorial Gym,
University of British Columbia, 122 – 6081 University Blvd, Vancouver, British Columbia,
Canada, V6T 1Z1. Tel: (604) 822 4864; Fax: (604) 822 6842; E-mail: mark.beauchamp@ubc.ca
Funding: This study was supported by a Canadian graduate scholarship for the first author from
the Social Sciences and Humanities Research Council, as well as a scholar award for the last
author from the Michael Smith Foundation for Health Research.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 2
Abstract
Drawing from goal setting theory (Latham & Locke, 1991; Locke & Latham, 2002; Locke et al.,
1981), the purpose of this study was to conduct a systematic review and meta-analysis of multi-
component goal setting interventions for changing physical activity (PA) behaviour. A literature
search returned 41,038 potential articles. Included studies consisted of controlled experimental
trials wherein participants in the intervention conditions set PA goals and their PA behaviour was
compared to participants in a control group who did not set goals. A meta-analysis was
ultimately carried out across 45 articles (comprising 52 interventions, 126 effect sizes, n = 5912)
that met eligibility criteria using a random-effects model. Overall, a medium, positive effect
(Cohen’s d(SE) = .552(.06), 95% CI = .43-.67, Z = 9.03, p<.001) of goal setting interventions in
relation to PA behaviour was found. Moderator analyses across 20 variables revealed several
noteworthy results with regard to features of the study, sample characteristics, PA goal content,
and additional goal-related behaviour change techniques. In conclusion, multi-component goal
setting interventions represent an effective method of fostering PA across a diverse range of
populations and settings. Implications for effective goal setting interventions are discussed.
Keywords: goal setting, physical activity, exercise, health, intervention, systematic
review, meta-analysis
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 3
The Effectiveness of Multi-Component Goal Setting Interventions for Changing Physical
Activity Behaviour: A Systematic Review and Meta-Analysis
It is well established that engaging in regular physical activity (PA) is associated with
numerous health benefits including reduced risk of cardiovascular disease, diabetes, stroke,
obesity, multiple cancers, as well as improved psychological functioning and quality of life
(World Health Organization, 2009). In spite of these physical and mental health benefits, the
majority of people across a range of populations currently do not engage in sufficient levels of
PA (World Health Organization, 2010). In response to this ongoing public health concern, there
have been growing calls to develop efficacious cost-effective PA interventions (World Health
Organization, 2007).
A prominent intervention strategy that has been the focus of much research interest is
goal setting. A goal has been described as the object or aim of an action (Latham & Locke, 1991;
Locke & Latham, 2002; Locke, Shaw, Saari, & Latham, 1981). Formulated upon Ryan’s (1970)
contention that conscious goals influence subsequent action, goal setting theory was originally
developed within the context of industrial and organizational psychology (Locke et al., 1981;
Latham & Locke, 1991; Locke & Latham, 2002; 2006), and over the past four decades has been
used to explain and change behaviour across several domains of human functioning (e.g., Kyllo
& Landers, 1995; Tubbs, 1986; Zetik & Stuhlmacher, 2002). According to Locke and Latham
(2002; 2006), the process of setting goals facilitates behaviour change by guiding individuals’
attention and efforts, and increasing persistence towards obtaining a specified level of
proficiency.
In their seminal work on goal setting theory, Locke and Latham (1991; 2002; 2006)
emphasized the importance of several variables that have the potential to moderate the effects of
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 4
goal setting in achieving a desired set of outcomes. First, they theorized that a series of
individual difference variables (on the part of the person setting the goal) might have a
substantive effect on goal attainment. Specifically, they contended that greater goal commitment,
ability and self-efficacy to perform the tasks related to the goal, perceived importance of the goal,
and anticipated satisfaction if the goal is attained (i.e., goal valence) act as personal variables by
which the effects of goal setting can be maximized (Latham & Locke, 1991). Other individual
difference factors such as ethnicity, age, and sex (i.e., demographic variables) were theorized to
not moderate the effects of goal setting (Latham & Locke, 1991).
Beyond these individual difference variables, Latham and Locke (1991) also point to
various “goal attributes” (p. 213) that may act as moderators. These specifically relate to facets
of the goal being set that might either facilitate or debilitate the effects of this intervention. For
instance, in terms of the content of goals, they contended that making goals specific,
difficult/challenging, public to other individuals, and positively-framed is advantageous. Other
goal content considerations include the person who prescribes the goal (i.e., self-set versus set by
others), proximity of the goal timeframe (i.e., short- vs long-term goals), and task complexity.
With regard to the former attribute, Latham and Locke (1991) contend that goals are effective
regardless of whether they are self-set by an individual, assigned by another person (e.g., an
expert or authority figure), or set collaboratively between the two. With regard to timeframe (vis
a vis proximity), shorter-term goals can be beneficial (such as by increasing self-efficacy in
performing a task), as can longer-term goals (such as by improving behaviour in long term
behaviour change programs; Latham & Locke, 1991). Finally, goals can be beneficial with
simple tasks but also with complex tasks; however, in these latter situations, it is important that
individuals incorporate suitable strategies for obtaining the goal (Latham & Locke, 1991).
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 5
Within the family of ‘goal attributes’ specified by Latham and Locke (1991) are several
variables that can be considered distinct behaviour change techniques (BCTs; cf. Abraham &
Michie, 2008; Kok, Gottlieb, Peters, Mullen, Parcel, Ruiter, … & Bartholomew, in press) in their
own right. For instance, according to Latham and Locke (1991) the attainment of goals can be
enhanced by incorporating feedback (i.e., data about recorded behaviour or evaluating
performance in relation to a set standard), suitable task strategies (i.e., how to perform/attain
behaviour), and rewards (i.e., contingent incentives that are explicitly linked to the achievement
of specified behaviour) (cf. Abraham & Michie, 2008). In fact, Latham and Locke (1991)
highlight a process known as the “high-performance cycle” (p. 233) wherein they hypothesize
that combining feedback, task strategies, and rewards with regard to one’s goals can maximize
individuals’ success as opposed to merely setting a goal alone. Moreover, self-control training
and subconscious priming may also increase the likelihood of reaching one’s goal (Latham &
Locke, 1991; Locke & Latham, 2006).
Finally, in addition to the abovementioned personal variables and goal attributes that
might act as moderator variables, Latham and Locke (1991) also highlight the potential for two
environmental variables, situational constraints and norms, to moderate the effects of goal
setting. First, when situational constraints are high, individuals are theorized to be less likely to
achieve their goals compared to when these constraints are low (Latham & Locke, 1991).
Second, being made aware of the average/normal performance of others (i.e., norms) can impact
the goal one chooses to set (e.g., individuals may make their previously-set goal more difficult if
they feel they are too easy compared to others’ goals and/or behaviour; Latham & Locke, 1991).
A number of studies within the field of behavioural medicine have pointed to the benefits
of goal setting in fostering PA behaviour. Positive relationships derived from goal setting
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 6
interventions have been found within multiple settings, such as in primary care centers (e.g.,
Trinh, Wilson, Williams, Sum, & Naylor, 2012), grade school classrooms (e.g., Wang, 2004),
workplaces (e.g., Dishman, Vandenberg, Motl, Wilson, & DeJoy, 2010), and faith-based
organizations (e.g., Penn, 2010). Goal setting interventions have also been beneficial in relation
to PA with a variety of populations, including males (e.g., Moy, Weston, Wilson, Hess, &
Richardson, 2012) and females (e.g., Sidman, Corbin, & Le Masurier, 2013), across a range of
ages from children (e.g., Horne, Hardman, Lowe, & Rowlands, 2009) to older adults (e.g.,
Strath, Swartz, Parker, Miller, Grimm, & Cashin, 2011), as well as in persons with chronic
conditions such as chronic obstructive pulmonary disease (e.g., Altenburg, ten Hacken,
Bossenbroek, Kerstjens, de Greef, & Wempe, 2015), osteoarthritis (e.g., Talbot, Gaines, Huynh,
& Metter, 2003), intellectual impairments (e.g., Tilley, 2010), and cancer survivors (e.g.,
Matthews, Wilcox, Hanby, Der Ananian, Heiney, Gebretsadik, & Shintani, 2007). Furthermore,
positive relationships with PA have been shown in goal setting interventions of several durations
from one week (e.g., Gardiner, Eakin, Healy, & Owen, 2011) to over one year (e.g., Narayan &
Mazzola, 2014), as well as in interventions guided by an array of theories including social
cognitive theory (e.g., Croteau, 2004), goal setting theory (e.g., Booth, Nowson, & Matters,
2008), the transtheoretical model (e.g., Fitzsimons, Baker, Gray, Nimmo, & Mutrie, 2012), self-
regulation theory (e.g., Moy, Janney, Nguyen, Matthess, Cohen, Garshick, & Richardson, 2010),
and the theory of planned behaviour (e.g., Heron, Tully, McKinley, & Cupples, 2013).
Despite the collection of research pointing to the benefits of goal setting interventions on
PA behaviour, a systematic review and quantitative synthesis of these effects has yet to be
conducted. With this in mind, the first purpose of this study was to conduct a meta-analysis to
assess the overall averaged effect of goal setting interventions in relation to individual PA
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 7
behaviour. We hypothesized that positive, significant effects of goal setting interventions in
relation to PA behaviour would emerge from this analysis. We also sought to examine whether
variability in the size of this effect can be explained by moderator variables related to
characteristics of the study, sample characteristics, and goal attributes (see Methods section for
an overview of these moderators). When taken together, such a meta-analytic synthesis is not
only a particularly timely endeavor, but also has the potential to substantively inform PA health-
promotion interventions in a variety of settings and involving diverse populations.
Methods
Literature Search and Eligibility Criteria
Searches for possible research articles were conducted in the following databases:
PubMed, PsycInfo, PsycArticles, Medline, Cochrane Central Register of Controlled Trials, ISI
Web of Science, SportDiscus, and ProQuest Dissertations and Theses. These searches were
conducted in January 2015. In each database search, we used the following combinations of
search terms: (1) goal*, with (2) physical activity, physically active, exercis*, walking. Potential
articles were then reviewed for eligibility by the first author and a second co-author. Each article
was first subjected to title elimination, then abstract elimination, and finally full-text elimination.
We also searched the reference sections of the articles that met our inclusion criteria in order to
determine if any additional articles could be retrieved (see Figure 1).
To be included in the meta-analysis, an article needed to meet the following criteria: (1)
the study must be a between-subjects controlled experimental study (i.e., include a control group
who did not set goals); (2) the study must be an intervention wherein the primary focus involved
setting PA goals and assessing PA behaviour. Studies could be included if additional BCTs (e.g.,
goal-related feedback, rewards) were incorporated into the goal setting intervention (i.e., a multi-
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 8
component intervention) for the purpose of fostering goal attainment. However, if an
intervention’s primary focus and/or additional components were unrelated to the PA goals (e.g.,
general stress management, cognitive behaviour therapy), these studies were excluded; (3) the
PA goal had to be individually-focused and measured (group goal setting interventions were
included only if they also involved individuals setting personal PA goals); and (4) the article had
to provide appropriate statistics to compute effect sizes. If the requisite statistical information
was missing from a given manuscript, we contacted the corresponding authors for this
information. Studies that focused on improving task performance (e.g., improving race times of
elite runners, increasing athletes’ performance in their sport) were not included.
Data Analysis
Articles that met our eligibility criteria were extracted and subsequently reviewed
independently by the first author and another co-author with respect to 20 moderator variables
and risk of bias (described below). When discrepancies in coding occurred, the authors met to
resolve these differences by referring back to the article in question. Data were then analyzed as
a random-effects model using the software Comprehensive Meta-Analysis, Version 2
(Borenstein, Hedges, Higgins, & Rothstein, 2005). A random-effects model assumes
heterogeneity in the effect sizes from a population of studies, and is the appropriate model to use
in social science research (compared to a fixed-effects model which assumes that the average
effect size does not vary from study to study; Borenstein, Hedges, Higgins, & Rothstein, 2009;
Field & Gillett, 2010).
Where possible, effect sizes for each study were derived from means, standard
deviations, and sample sizes at baseline and post-intervention of experimental and control
conditions (Borenstein et al., 2009; DeCoster & Claypool, 2004). If such statistics were missing,
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 9
we used F-statistics, t scores, and p-values. Each study was given a relative weight based on its
precision, which is determined by the study’s variance, standard error, and confidence interval
(i.e., more precise data is given a larger relative weight compared to less precise data; Borenstein
et al., 2009). In instances where a study provided more than one effect size (such as when
multiple PA outcomes were measured), these effect sizes were combined into one overall effect
size statistic for that study, so as to not give greater relative weight to these studies and
potentially skew the overall results (Borenstein et al., 2009). The exception to this was when
articles reported the effects of multiple interventions (i.e., multiple subgroups), each of which
was subject to a unique goal setting protocol. In these cases, an effect size from each intervention
was computed; thus, this article contributed multiple effect sizes to the total number of
comparisons within the meta-analysis. If studies reporting results from multiple interventions
compared PA outcomes for each of the experimental conditions to the same control condition,
we corrected for potential unit-of-analysis errors by dividing the sample size of the control
condition by the number of within-study comparisons. For example, if three experimental
conditions were compared to one control condition (e.g., which had a sample size of 60
participants), we divided n of the control condition by 3 (i.e., 60/3 = 20; Higgins & Green, 2011).
Cohen’s d was calculated as the effect size metric to represent the standardized effect (i.e., the
average magnitude of effectiveness) of goal setting on PA behaviour (Cohen, 1992). Standard
errors and 95% confidence intervals were computed to test for the accuracy of the standardized
effects obtained.
Tests of heterogeneity within the meta-analysis were also performed by assessing the
variability in the observed effect sizes across studies (Q value), as well as the ratio of the true
heterogeneity to the total observed variation (I2). To assess potential publication bias, we
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 10
calculated the fail-safe N statistic, which estimates the number of unpublished studies with null
findings that would have to exist to reduce the effect size to zero (Rosenthal, 1979). If this
number is sufficiently large—Rosenberg (2005) advises a critical value of 5N+10—one can be
confident that the chance of such a number of studies existing is low. We also obtained funnel
plots to provide a visual depiction of potential publication bias. In addition, we examined risk of
bias within the included studies using the Cochrane Collaboration’s tool (Higgins, Altman,
Gøtzsche, Jüni, Moher, Oxman,... & Sterne, 2011) and by following recommendations from
Higgins and Green (2008) and de Bruin et al. (2015). Finally, sensitivity analyses were
conducted by obtaining an effect size when a study is removed from the analysis, thereby
assessing the impact of each individual intervention on the overall effect size.
Moderator Analyses
In total, we examined 20 potential moderators related to study characteristics, sample
characteristics, as well as goal attributes, which included the content of the goal and goal-related
BCTs (see Appendix A for a detailed description and coding of moderators). For each moderator
variable, we computed an effect size, standard error, 95% confidence interval, Z-value, and p-
value to test for the effects of each category on PA, as well as a Q statistic and corresponding p-
value to assess heterogeneity across these effects (Borenstein et al., 2009). Study characteristics
that were examined included publication source, whether a theoretical framework was used to
guide the intervention, intervention setting, mode of intervention delivery, type of physical
activity targeted, PA intensity targeted, and PA measure used to assess PA behaviour; a meta-
regression was also conducted to address whether the continuous variable of intervention
duration (total, in weeks) moderated the effects of goal setting on PA behaviour. Sample
characteristics included age, sex, baseline weight status (as sample mean BMI), baseline PA
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 11
levels, and population targeted (general or a specific population). Although we sought to also
examine additional personal variables (cf. Latham & Locke, 1991; Locke & Latham, 2002;
2006) including goal commitment, goal-related self-efficacy, goal-related ability, goal valence,
perceived importance of the goal, none of the studies in our meta-analysis assessed these
variables. We were also unable to examine ethnicity as a potential moderator because all study
samples either had a mix of different ethnicities or did not present data on this variable. Goal
content moderators—based on goal setting theory (Latham & Locke, 1991; Locke & Latham,
2002; 2006)—that were examined included goal specificity, source of goal prescription, goal
timeframe, and frequency of goal setting/modifications. Although we sought to also examine
goal difficulty, goal framing, task complexity, and whether goals were made public, none of the
studies in our meta-analysis reported data on these goal content variables; therefore, we were
unable to include these within our moderator analyses. We also sought to examine the inclusion
of the five goal-related BCTs highlighted by Locke and Latham (1991; 2002)—feedback,
strategy planning, rewards, subconscious priming, and self-control/management training—as
moderators within the goal setting interventions. None of the included studies operationalized
subconscious priming or self-control/management training; therefore, only feedback,
strategy/planning, and rewards were coded as candidate BCT moderators. Finally, none of the
studies operationalized situational constraints and norms, and so we were unable to examine
these environmental variables as potential moderators.
With regard to the goal-related BCTs highlighted within Locke and Latham’s (1991)
framework, we were also interested in whether various combinations of feedback, strategy
planning, and/or rewards (i.e., multi-component interventions) improve PA behaviour beyond
just setting a goal. We, therefore, conducted a method co-occurrence analysis as per recent
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 12
recommendations by Peters, de Bruijn, and Crutzen (2015) to determine “what works and under
what parameters” (p. 7).1 Hence, in the context of our study, the question is whether
interventions that combine multiple goal-related BCTs are more effective than those
interventions that do not. Specifically, using BCT absence as a reference category (i.e., when
goals were set but feedback, strategy planning, and rewards were not included within the
intervention), we dummy coded categories based on the inclusion of one or more of these three
BCTs. This provides a means of examining the extent to which goal setting is more effective (in
supporting PA behaviour) when it is augmented with each of these attributes (e.g., goal setting
alone versus goal setting plus feedback versus goal setting plus feedback and rewards, and so
on). This provides an important opportunity to test theoretical assertions of goal setting theory;
namely, that the effects of goal setting will be maximized if these BCTs are included as part of
the intervention for achieving the set PA goals. Aligning with the propositions of Locke and
Latham’s ‘high performance cycle’ of goal setting (1991), we hypothesized that goal setting
interventions that incorporated a more comprehensive set of goal-related BCTs would be more
effective in supporting PA behaviour when compared to those that used fewer (or no) BCTs.
Results
Literature Search
The literature searches from the eight databases returned 40,139 potentially relevant
articles. An additional 899 articles were retrieved from the bibliographic sections of manuscripts
that originally met our inclusion criteria, which resulted in an initial pool of 41,038 total articles.
After removing duplicates, 22,199 articles were subject to title and abstract review. Based on
these reviews, 21,698 articles were eliminated, while 501 were full-text reviewed. Ultimately, 45
articles met our eligibility criteria—see Figure 1 for the PRISMA (Moher, Liberati, Tetzlaff, &
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 13
Altman, 2009) flow diagram and Appendix B for the studies included. Of these studies, nine
included multiple experimental conditions—each of which was subject to a separate goal setting
intervention—which resulted in 52 total comparisons (k), a total sample size (n) of 5912, and 126
individual effect sizes. Inter-coder agreement of moderators and risk of bias assessment was over
90%, kappa (SE) = 0.80 (0.01). Overviews of each study with regard to experimental design,
sample characteristics, PA measures utilized, and effect sizes calculated are provided in Table 1.
Summary Statistics
Results of the overall goal setting effect along with summary statistics, sensitivity
analyses, and forest plots for the included studies are presented in Table 2. Overall, a medium
positive effect was found (d=0.552) based on Cohen’s (1992) criteria, with study effect sizes
ranging from -0.31 (Bycura, 2009) to 1.91 (Duncan & Pozehl, 2003). Sensitivity analyses
showed the smallest total effect size was 0.530 when the study by Annesi (2002) was removed,
and the largest effect size was 0.569 when the study by Reijonsaari et al. (2012) was removed.
Tests of heterogeneity revealed significant variability in the observed effect sizes across
interventions, Q(df) = 192.62(51), p<0.001. The I2 value was 73.52, indicating that a high
proportion (i.e., over 73%) of the observed between-study variance reflects true differences in
the effect sizes (Higgins, Thompson, Deeks, & Altman, 2003). Regarding potential publication
bias, the fail-safe N was 3354, which is sufficiently large (cf. Rosenberg, 2005). A funnel plot is
provided in Appendix C. With regard to potential risk of bias (Higgins et al., 2011), there were
no significant differences (Q(df) = 0.701(2), p = 0.704) in effect sizes between interventions
coded as being at a low (k=9, d=0.471, p = 0.002), high (k=9, d=0.497, p = 0.001), or unclear
(k=34, d=0.592, p < 0.001) risk (see Appendix D for the Table describing these results).
Moderator Analyses
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 14
The results of the moderator analyses are provided in Table 3.
Study Characteristics. Significant positive effects of goal setting on PA behaviour were
shown for both peer-reviewed journal articles and other publications (i.e., conference abstracts or
dissertations; ds0.54), as well as for atheoretical interventions and those guided by theory
(ds0.48). Significant effects also emerged across all intervention settings (ds0.43) except for
workplace settings whose effect size (d=0.21) approached significance. Significant effects were
also shown regardless of the mode of delivery (ds0.42). In terms of type of PA, we found
significant effects when aerobic activity was targeted or when participants could perform any
type of activity they desired (ds0.60); however, null effects emerged when the type of activity
targeted was not specified (d=0.14). Significant effects were shown when the goal targeted
moderate intensity PA or when participants were free to be active at any intensity they desired as
well as when this information was not specified (ds0.49). Non-significant effects were found
when the targeted activity was specified to be of moderate-vigorous intensity (d=0.32). With
regard to the PA measures used, significant effects were evident regardless of whether objective
(i.e., technology) or subjective (i.e., self-report) methods were used (ds0.44). The exception to
this was for those two studies that used accelerometers only (d=0.33).
Finally, the results of the meta-regression indicated that the length of the intervention did
not moderate the relationship between the goal setting intervention and PA (a visual depiction of
this meta-regression is provided in Appendix E). Specifically, the intercept of this regression was
statistically significant, Cohen’s d(SE)=0.58(0.08), 95% confidence interval (0.41, 0.74), Z=6.98,
p<.00001, while the slope of this regression was not, Cohen’s d(SE)=-0.002(0.004), 95%
confidence interval (-0.01, 0.005), Z=-0.51, p=0.61. In other words, goal setting interventions
had a medium positive effect on PA regardless of the length of the intervention.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 15
Sample Characteristics. Significant positive effects were evident across all age ranges
(ds0.41), in both female only and female plus male samples (ds0.54), for all weight statuses
(ds0.48), with both insufficiently and sufficiently active (at baseline) samples (ds0.49), as well
as for samples consisting of the general population and a specific population (ds0.51).
Goal Attributes. Significant effects were evident regardless of the specificity of the goal
(ds0.42) or the individual who prescribed the goal (ds0.47). In terms of goal timeframe,
significant effect sizes were shown in interventions targeting daily PA as well as a combination
of daily and weekly PA (ds0.56), but not weekly PA alone. With regard to the frequency of
goal setting/modification, significant positive effects were shown when goals were set at baseline
only, modified on a weekly or bi-weekly basis, and when modifications could be made at any
time as per the participant’s discretion (ds0.49).
With regard to the three goal-related BCTs emphasized within goal setting theory, we
first analyzed each individually by comparing interventions in which the BCT was present versus
those in which the BCT was absent (summaries for each study with regard to the inclusion of the
three goal-related BCTs are provided in Appendix F). Significant positive effects were found in
interventions that included the respective BCT as well as in those that did not (ds0.46).
Specifically, no significant differences emerged between studies that did or did not include (a)
feedback related to the goal (Q(df) = 0.16(1), p = 0.693), (b) strategy planning for attaining the
goal (Q(df) = 0.50(1), p = 0.480), and (c) rewards for goal progress (Q(df) = 0.46(1), p = 0.499).
We also examined whether differential effects were evident across interventions that included
various combinations of these BCTs to go along with setting goals (i.e., co-occurrence effects;
Peters et al., 2015). Only one intervention fell into the category in which goal setting was done
without also including any of the three BCTs—the effect size of this study was not statistically
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 16
significant (d=.20). Most interventions (k=45) employed one or two BCTs alongside goal setting
with significant effects emerging in these interventions (d=0.58). Specifically, there were
significant effects in goal setting interventions that included feedback and strategy planning
alone or in combination with each other (ds0.51). When rewards were combined with feedback
and incorporated into the goal setting intervention, a comparable but non-significant effect size
was obtained (d=0.58); however, only two studies fell into this category. When all three BCTs
were incorporated into the goal setting intervention (k=6), significant effects emerged (d=0.44).
Discussion
The purpose of this systematic review was to meta-analyze the effects of multi-
component goal setting interventions in relation to PA behaviour. Overall, a medium-sized
positive effect was found, which suggests that these interventions are effective for improving
PA. Subsequent moderator analyses revealed a number of noteworthy findings.
With regard to study characteristics, significant and comparable effects of goal setting
interventions on PA behaviour were observed regardless of the source from which a study was
obtained (i.e., published journal article versus conference/thesis). Similarly, the effects of these
interventions were significant regardless of whether theory was used to guide the intervention or
not (i.e., atheoretical). Significant effects were also seen across all intervention settings, except in
workplace locations; however, conclusions on the latter setting should be tempered given that the
effect approached statistical significance (d=0.211, p=.059). In terms of delivery mode,
interventions were effective irrespective of whether the intervention was delivered in person, via
technology, or as a combination of the two. This is an encouraging finding as it suggests that
interventions utilizing websites or text messaging, for example, as the mode of delivery are just
as effective as those conducted face-to-face. These technology-based goal setting interventions
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 17
also have the added benefit of improving the reach of these interventions beyond that which can
be done in person (cf. Fanning, Mullen, & McAuley, 2012). Moreover, the result of the meta-
regression for duration of interventions is noteworthy as we found that duration did not moderate
the intervention – PA effects. In other words, for improving individuals’ PA behaviour, it does
not seem to matter whether interventions utilize brief protocols (e.g., one week) or prolonged
programs (e.g., year-long). Altogether, these are important and encouraging findings as they
seem to suggest that the utility of goal setting interventions is not limited to the use of theory
(versus atheoretical frameworks), particular research settings, modes of delivery, or lengths of
program durations; positive, significant effects on PA seem to emerge regardless.
With regard to the PA goal that was set, significant intervention effects were found when
aerobic activity or any type of activity was prescribed. On the one hand, these findings are
encouraging as they suggest marked improvements in PA behaviour can be incurred when
participants set goals specifically for aerobic activities or more generally for any type of activity
an individual wishes to undertake. On the other hand, it is unclear whether these findings extend
to other specific activities. For instance, many PA guidelines suggest incorporating bone and
muscle strengthening activities into individuals’ weekly PA regimens (e.g., Tremblay et al.,
2011). It would therefore be prudent for future researchers to assess the effects of goal setting
specifically for these types of health-enhancing activities to determine the generalizability of
these meta-analytic findings to other types of PA. In terms of the type of measure that was used
to assess PA, the results revealed moderate effects irrespective of whether objective (using
technology) or subjective (using self-report) forms of assessment were used. Specifically,
pedometer, PA questionnaire, and activity log measures all yielded significant effects. In
contrast, it should be noted that the effects for accelerometer (only) assessments was non-
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 18
significant; however, caution should be exercise in interpreting these findings as only two studies
used this form of assessment.
With regard to intensity, goal setting interventions were effective when the goals targeted
PA of any intensity or of moderate intensity. However, when goals were directed towards higher
(i.e., moderate-vigorous) intensity PA, the intervention effects were not significant. As one
potential explanation for these latter null findings, it is worth noting that the majority of the
interventions included in this meta-analysis fell into the category in which intensity was not
specified. It is therefore unclear at what intensity participants in these studies were actually
exercising. On the one hand, participants in some of these studies may have exercised at higher
intensities, and, therefore, benefitted from using goal setting to increase their moderate and/or
moderate-to-vigorous PA. On the other hand, it is certainly possible that participants in these ‘not
specified’ studies indeed restricted their PA participation to moderate (or low) intensity PA.
Regardless, based on the effect sizes reported, it appears as though goal setting interventions may
display stronger effects when directed towards achieving moderate intensity PA, rather than high
intensity PA. Further to this point, there may be reasons why goal setting interventions appear to
be less effective when higher-intensity PA is targeted. For instance, it has been suggested that at
high intensities (i.e., above one’s ventilatory threshold), individuals’ affective responses to
exercise (i.e., enjoyment) are more strongly influenced by interoceptive/physiological cues (e.g.,
muscular, respiratory) than by psychological ones (e.g., goals) which play a much greater role at
lower intensities (Ekkekakis, 2003). Some studies have also found that adherence is higher when
lower intensity activities are prescribed compared to higher intensity (cf. Perri et al., 2002). This
may be due to reasons such as a perceived increased risk of injury (Dishman & Buckworth,
1997) and decreased positive affect (Ekkekakis, Parfitt, & Petruzzello, 2011) at higher
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 19
intensities. It should be reiterated that the majority of the participants in the studies within this
meta-analysis were insufficiently active at baseline (i.e., not meeting PA guidelines) and, as
such, it is possible that moderate-intensity PA goals were better suited than higher-intensity goals
for such individuals. In any case, future research is necessary to determine whether (and how)
goal setting interventions are effective for enhancing higher-intensity PA.
With regard to sample characteristics, goal setting interventions were effective
irrespective of a study sample’s age, baseline weight and activity status (prior to the
intervention), and sex. No controlled intervention studies focused on males only; however, given
that comparable effects were found for both females only and those studies that sampled males
and females, goal setting interventions are likely beneficial across sexes. In addition, the findings
that interventions derived significant effects across the age-span (from children to older adults),
involving inactive and active participants, as well as healthy weight and overweight/obese
participants points to the pervasive utility of goal setting interventions. With regard to population
type, the results revealed that the interventions derived comparable effects for specific
populations (e.g., persons with cardiac issues, diabetes, cancer) as well as the general population.
Taken together, goal setting interventions appear to be effective for a wide variety of
populations.
With regard to goal attributes, a number of noteworthy findings emerged. First,
significant effects were found regardless of goal specificity. That is, goal setting interventions
had a medium effect on the targeted PA behaviour when the goal identified an absolute level of
PA behaviour (e.g., to obtain 10,000 steps per day), PA increases relative to one’s current PA
levels (e.g., to increase PA time by 20% from baseline), and even when the goal was vaguely
defined (e.g., to be more active). Although this may be surprising as it is commonly assumed that
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 20
specific goals are superior to vague ones, Latham and Locke (1991) note that “trying for specific,
challenging goals may actually hurt performance in certain circumstances” such as “in the early
stages of learning a new, complex task” (p. 229). It is important to note that the majority of the
samples included in the meta-analysis consisted of participants who were insufficiently active at
baseline and, as such, it is possible that vague goals were advantageous for these participants
who were in the early stages of learning to be physically active. This idea should be tested in
further research.
Second, significant intervention effects were evident regardless of the individual(s) who
prescribed the goal. That is, goal setting interventions had a significant effect on PA behaviour
when the goal was set by the participants themselves, by an interventionist, or when participants
and interventionists collaborated to determine an appropriate PA goal together. These results
corroborate Locke and Latham’s theorizing that goals are just as effective whether they are
“assigned, self-set, or set participatively” (2002, p. 714).
Third, goal setting interventions appeared to be most effective when goals were set in
relation to daily PA or a combination of daily and weekly PA. By contrast, when goals were set
only in relation to weekly PA, these significant effects dissipated. This is an intriguing finding
that warrants future attention as many health promotion organizations around the world advise
obtaining a certain amount of PA per week rather than per day (e.g., World Health Organization,
2010). It is possible that recommending PA behaviour on a weekly timeframe undermines the
potential benefits that can be derived from setting PA goals. Thus, to increase the likelihood of
getting people sufficiently active, perhaps there should be a greater emphasis on daily PA
behaviour rather than—or at least in addition to—weekly PA.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 21
Fourth, in terms of the frequency of goal setting, significant effects were shown for PA
goals set at baseline only, on a weekly or bi-weekly basis, or when they could be modified
whenever the participant felt it was appropriate. By contrast, goals modified on a daily basis did
not have a significant effect on PA, although it is worth noting that only one study had
participants set and revisit their goals on a daily basis. Therefore, it may be premature to
conclude that adjusting goals on a daily basis is ineffective for improving subsequent PA
behaviour. Nevertheless, at present, the results seem to suggest that individuals can derive
significant benefits in PA behaviour if they set their goals at baseline only (i.e., prior to the
program of intervention) and/or if they are able to modify them on a weekly or bi-weekly basis
or as they otherwise deem necessary.
With regard to the BCTs that are theorized as important components of the goal setting
process (Latham & Locke, 1991), the initial moderator results revealed that the interventions
were comparably effective regardless of whether or not participants were provided with feedback
on their goals, planned strategies to help them achieve their goals, or received rewards based on
progress to or attainment of their goals. However, in light of the fact that nearly every
intervention included in this meta-analysis incorporated one or more goal-related BCTs in
addition to goal setting itself, we sought to examine method co-occurrence using procedures
outlined by Peters et al. (2015). Only one study did not explicitly include at least one of the three
goal-related BCTs, which did not significantly benefit subsequent PA behaviour. By contrast,
significant effects emerged when, in addition to setting PA goals, participants received feedback
on their PA goals and/or engaged in strategy planning for how they would reach their goals.
Despite yielding a comparable effect size, the combination of feedback and rewards for being
physically active at a certain level alongside goal setting did not reach statistical significance;
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 22
however, caution should be exercised in interpreting this latter finding as only two interventions
fell into this latter category. Furthermore, when all three techniques were incorporated as part of
a goal setting intervention, significant effects emerged. In summary, when interventions used one
or more goal-related BCTs, the effect sizes were in the medium range. As a final cautionary note,
while 18 comparisons involved goal setting with feedback, and 23 comparisons involved goal
setting with feedback plus planning, the rest of the comparison categories in the co-occurrence
analysis involved a relatively small pool of studies, which makes direct comparisons difficult.
Nevertheless, the results seem to point to the utility of the goal-related BCTs highlighted by
Latham and Locke (1991); that is, if goals are set without these components they tend to be less
effective than if those attributes are included.
Although the results of this meta-analysis provide valuable information on goal setting
interventions, the study is not without its limitations. For one, there was a high degree of
heterogeneity across the included studies, likely due to differences between studies with regard
to variations in the type of PA operationalized, the way in which PA was assessed and
quantified, the timepoints at which PA was measured, the settings in which the interventions
were delivered, and so forth. Although steps were taken to improve the interpretability of the
results (e.g., only including controlled experimental studies, performing sensitivity analyses,
assessing risk of bias, and conducting several moderator analyses) and heterogeneity is not
uncommon in meta-analyses within the social sciences, it can result in conclusions being more
suggestive than indisputable (Higgins & Thompson, 2002). Moreover, the majority of the
samples included in the meta-analysis were with adults who were overweight and whose PA goal
related to aerobic activity. Thus, further research is needed to ascertain the veracity of the results
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 23
of the current meta-analysis, especially for individuals of other weight statuses, in other age
ranges, and for participation in other types of PA.
Another limitation of this study concerns the coding of moderators. First, some of the
moderators required an additional ‘unclear’ or ‘not specified’ category (e.g., weight status, goal
specificity) when it was not explicit to which category the study belonged (based on the details
presented within the respective manuscript). For instance, several interventions were unclear (as
opposed to explicitly defined) with regard to goal specificity (i.e., whether the PA goal was
absolute or relative). This, therefore, makes comparisons—and subsequent conclusions—related
to goal specificity somewhat tentative. Although it may not always be possible for studies to
report exactly what participants did with regard to these moderator variables, stronger
implications could have been provided if this information were consistently available (cf.
Abraham & Michie, 2008; Conn & Groves, 2011). A related point concerns the Cochrane risk of
bias assessment. The majority of studies in this analysis were given an overall code of ‘unclear’
denoting that at least one of the seven potential sources of bias could not be coded as clearly
being ‘low’ risk (see Higgins & Green, 2008). Hence, it is difficult to be certain of (a) whether a
risk of bias was present or absent in these studies (i.e., if the authors of a study had indeed made
information on each source of bias explicit, would this study have been subsequently coded as
having a ‘low’ or ‘high’ risk of bias?), and, in turn, (b) the extent to which potential bias
influenced the overall effect size of goal setting interventions on PA behaviour. Nevertheless, it
is important to note that there were no significant differences between those studies classified as
low and high risk of bias (Higgins and Green, 2008), with regard to the effectiveness of goal
setting interventions in relation to PA behaviour.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 24
Finally, in spite of the wide range of attributes that Locke and Latham (1991; 2002; 2006)
purport would maximize the goal setting effects, some of these attributes were not
operationalized within any of the included studies (e.g., goal difficulty, subconscious priming);
therefore, we were unable to assess the full range of goal setting attributes highlighted by Locke
and Latham (1991; 2002; 2006). Moreover, despite our efforts to examine method co-occurrence
and isolate the effects of individual goal-related BCTs within the interventions in this meta-
analysis, no studies examined goal setting plus rewards only, and very few comparisons aligned
with some of the other co-occurrence categories (e.g., k=2 for goal setting interventions that
included planning only; k=2 for goal setting interventions that involved feedback plus rewards).
As such, we were precluded from concluding which specific combination of goal-related BCTs
(within goal setting theory) are most likely to maximise the effects of goal setting.
In spite of these limitations, the results of this meta-analysis provide evidence that multi-
component goal setting interventions are generally effective in promoting PA behaviour. The
benefits of these interventions are present across a diverse range of settings, populations, and
intervention characteristics. Furthermore, the incorporation of various goal setting attributes
including planning, the use of rewards, and feedback appear to be beneficial in maximizing the
effects of goals setting interventions in relation to PA behaviour.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 25
Endnotes
1. We thank an anonymous reviewer for the suggestion to test method co-occurrence effects
(cf. Peters et al., 2015).
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 26
References
Abraham, C., & Michie, S. (2008). A taxonomy of behavior change techniques used in
interventions. Health Psychology, 27, 379-387. doi: 10.1037/0278-6133.27.3.379
Altenburg, W. A., ten Hacken, N. H., Bossenbroek, L., Kerstjens, H. A., de Greef, M. H., &
Wempe, J. B. (2015). Short-and long-term effects of a physical activity counselling
programme in COPD: A randomized controlled trial. Respiratory Medicine, 109, 112-
121. doi: 10.1016/j.rmed.2014.10.020
Booth, A. O., Nowson, C. A., & Matters, H. (2008). Evaluation of an interactive, Internet-based
weight loss program: A pilot study. Health Education Research, 23, 371-381.
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2005). Comprehensive meta-analysis,
version 2. Englewood, NJ: Biostat.
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2009). Introduction to meta-analysis.
Chichester, England: Wiley.
Cohen, J. (1992). A Power Primer. Psychological Bulletin, 112, 155-159. doi: 10.1037/0033-
2909.112.1.155
Conn, V. S., & Groves, P. S. (2011). Protecting the power of interventions through proper
reporting. Nursing Outlook, 59(6), 318–325. doi:10.1016/j.outlook.2011.06.003
Croteau, K. A. (2004). A preliminary study on the impact of a pedometer-based intervention on
daily steps. American Journal of Health Promotion, 18, 217-219.
de Bruin, M., McCambridge, J., & Prins, J. M. (2015). Reducing the risk of bias in health
behaviour change trials: Improving trial design, reporting or bias assessment criteria? A
review and case study. Psychology and Health, 30, 8-34. doi:
10.1080/08870446.2014.953531
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 27
DeCoster, J., & Claypool, H. M. (2004). A meta-analysis of priming effects on impression
formation supporting a general model of informational biases. Personality and Social
Psychology Review, 8, 2–27. doi:10.1207/S15327957PSPR0801_1
Dishman, R. K., & Buckworth, J. (1997). Adherence to physical activity. In: W.P. Morgan (Ed.),
Physical Activity and Mental Health (pp. 63-80). Philadelphia, PA: Taylor & Francis.
Dishman, R. K., Vandenberg, R. J., Motl, R. W., Wilson, M. G., & DeJoy, D. M. (2010). Dose
relations between goal setting, theory-based correlates of goal setting and increases in
physical activity during a workplace trial. Health Education Research, 25, 620-631.
Ekkekakis, P. (2003). Pleasure and displeasure from the body: Perspectives from exercise.
Cognition & Emotion, 17, 213-239. doi:10.1080/02699930302292
Ekkekakis, P., Parfitt, G., & Petruzzello, S. J. (2011). The pleasure and displeasure people feel
when they exercise at different intensities. Sports Medicine, 41, 641-671.
Fanning, J., Mullen, S. P., & McAuley, E. (2012). Increasing physical activity with mobile
devices: a meta-analysis. Journal of Medical Internet Research, 14, e161. doi:
10.2196/jmir.2171
Field, A. P., & Gillett, R. (2010). How to do a meta-analysis. The British Journal of
Mathematical and Statistical Psychology, 63, 665-694. doi:10.1348/000711010X502733
Fitzsimons, C. F., Baker, G., Gray, S. R., Nimmo, M. A., & Mutrie, N. (2012). Does physical
activity counselling enhance the effects of a pedometer-based intervention over the long-
term: 12-month findings from the Walking for Wellbeing in the West study. BMC Public
Health, 12, 206-218. doi: 10.1186/1471-2458-12-206
Gardiner, P. A., Eakin, E. G., Healy, G. N., & Owen, N. (2011). Feasibility of reducing older
adults’ sedentary time. American Journal of Preventive Medicine, 41, 174-177.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 28
Heron, N., Tully, M. A., McKinley, M. C., & Cupples, M. E. (2013). Steps to a better Belfast:
Physical activity assessment and promotion in primary care. British Journal of Sports
Medicine, 48, 1558–1563. doi: 10.1136/bjsports-2012-091581
Higgins, J. P., Altman, D. G., Gøtzsche, P. C., Jüni, P., Moher, D., Oxman, A. D., ... & Sterne, J.
A. (2011). The Cochrane Collaboration’s tool for assessing risk of bias in randomised
trials. BMJ, 343-352. doi: 10.1136/bmj.d5928
Higgins, J. P. & Green, S. (Ed.). (2008). Cochrane handbook for systematic reviews of
interventions (Vol. 5.1.0). The Cochrane Collaboration. Retrieved from www.cochrane-
handbook.org.
Higgins, J. P. T., & Thompson, S. G. (2002). Quantifying heterogeneity in a meta-analysis.
Statistics in Medicine, 21, 1539–1558. doi:10.1002/sim.1186
Higgins, J., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003). Measuring inconsistency in
meta-analyses. BMJ, 327, 557-560. doi: 10.1136/bmj.3277414.557
Horne, P. J., Hardman, C. A., Lowe, C. F., & Rowlands, A. V. (2009). Increasing children’s
physical activity: A peer modelling, rewards and pedometer-based intervention.
European Journal of Clinical Nutrition, 63, 191-198.
Kok, G., Gottlieb, N. H., Peters, G. J. Y., Mullen, P. D., Parcel, G. S., Ruiter, R. A., ... &
Bartholomew, L. K. (in press). A Taxonomy of Behavior Change Methods; an
Intervention Mapping Approach. Health Psychology Review, 1-32.
Kyllo, L. B., & Landers D. M. (1995). Goal setting in sport and exercise: A research synthesis to
resolve the controversy. Journal of Sport and Exercise Psychology, 17, 117-137.
Latham, G. P., & Locke, E. A. (1991). Self-regulation through goal setting. Organizational
Behavior and Human Decision Processes, 50(2), 212-247.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 29
Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal setting and
task motivation: A 35-year odyssey. American Psychologist, 57, 705-717.
doi:10.1037//0003-066X.57.9.705
Locke, E. A., Shaw, K. N., Saari, L. M., & Latham, G. P. (1981). Goal setting and task
performance: 1969–1980. Psychological Bulletin, 90(1), 125-152.
Locke, E. A., & Latham, G. P. (2006). New directions in goal-setting theory. Current Directions
in Psychological Science, 15, 265-268. doi:10.1111/j.1467-8721.2006.00449.x
Matthews, C. E., Wilcox, S., Hanby, C. L., Der Ananian, C., Heiney, S. P., Gebretsadik, T., &
Shintani, A. (2007). Evaluation of a 12-week home-based walking intervention for breast
cancer survivors. Supportive Care in Cancer, 15, 203-211. doi: 10.1007/s00520-006-
0122-x
Moher, D., Liberati, A., Tetzlaff, J., & Altman, D. G. (2009). Preferred reporting items for
systematic reviews and meta-analyses: the PRISMA statement. Annals of Internal
Medicine, 151, 264-269. doi: 10.1371/journal.pmed1000097
Moy, M. L., Janney, A. W., Nguyen, H. Q., Matthess, K. R., Cohen, M., Garshick, E., &
Richardson, C. R. (2010). Use of pedometer and Internet-mediated walking program in
patients with chronic obstructive pulmonary disease. Journal of Rehabilitation Research
and Development, 47, 485-496.
Moy, M. L., Weston, N. A., Wilson, E. J., Hess, M. L., & Richardson, C. R. (2012). A pilot
study of an internet walking program and pedometer in COPD. Respiratory Medicine,
106, 1342-1350.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 30
Narayan, A., & Mazzola, J. (2014). Encouraging employees to be more active: Taking “steps”
toward personal goals. Journal of Psychological Issues in Organizational Culture, 4, 5-
17. doi: 10.1002/jpoc.21121
Penn, W. L. (2010). Use of pedometers to increase physical activity in African American females
(Master’s thesis). Arkansas State University. Retrieved from ProQuest Dissertation and
Theses.
Perri, M. G., Anton, S. D., Durning, P. E., Ketterson, T. U., Sydeman, S. J., Berlant, N. E., ... &
Martin, A. D. (2002). Adherence to exercise prescriptions: effects of prescribing
moderate versus higher levels of intensity and frequency. Health Psychology, 21, 452-
458. doi: 10.1037//0278-6133.21.5.452
Peters, G-J. Y., de Bruin, M., & Crutzen, R. (2015). Everything should be as simple as possible,
but no simpler: Towards a protocol for accumulating evidence regarding the active
content of health behaviour change interventions. Health Psychology Review, 9(1) 1-14.
doi: 10.1080/17437199.2013.848409
Rosenberg, M. S. (2005). The file-drawer problem revisited: A general weighted method for
calculating fail-safe numbers in meta-analysis. Evolution: International Journal of
Organic Evolution, 59, 464-468. doi: 10.1554/04-602
Rosenthal, R. (1979). The file drawer problem and tolerance for null results. Psychological
Bulletin, 86, 638-641. doi: 10.1037/0033-2909.86.3.638
Ryan, T. A. (1970). Intentional behavior. New York: Ronald Press.
Sidman, C. L., Corbin, C. B., & Le Masurier, G. (2013). Promoting physical activity among
sedentary women using pedometers. Research Quarterly for Exercise and Sport, 75, 122-
129.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 31
Strath, S. J., Swartz, A. M., Parker, S. J., Miller, N. E., Grimm, E. K., & Cashin, S. E. (2011). A
pilot randomized controlled trial evaluating motivationally matched pedometer feedback
to increase physical activity behavior in older adults. Journal of Physical Activity and
Health, 8, S267-S274.
Talbot, L. A., Gaines, J. M., Huynh, T. N., & Metter, E. J. (2003). A home-based pedometer-
driven walking program to increase physical activity in older adults with osteoarthritis of
the knee: A preliminary study. Journal of the American Geriatrics Society, 51, 387-392.
Tilley, F. (2006). Effects of goal setting for children with an intellectual disability on physical
activity and dietary habits (Master’s thesis). Wingate, NC: Wingate University. Retrieved
from ProQuest Dissertation and Theses.
Tremblay, M. S., Warburton, D. E., Janssen, I., Paterson, D. H., Latimer, A. E., Rhodes, R. E., ...
& Duggan, M. (2011). New Canadian physical activity guidelines. Applied Physiology,
Nutrition, and Metabolism, 36(1), 36-46. doi: 10.1139/H11-009
Trinh, L., Wilson, R., Williams, H. M., Sum, A. J., & Naylor, P. J. (2012). Physicians promoting
physical activity using pedometers and community partnerships: A real world
trial. British Journal of Sports Medicine, 46, 284-290. doi: 10.1136/bjsm.2009.069567
Tubbs, M. E. (1986). Goal setting: A meta-analytic examination of the empirical evidence.
Journal of Applied Psychology, 71, 474-483. doi:10.1037//0021-9010.71.3.474
Wang, S. (2004). The effects of goal setting on female middle school students’ physical activity
levels and motivation toward exercise (Doctoral dissertation). Tallahassee, FL: Florida
State University. Retrieved from ProQuest Dissertation and Theses.
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 32
World Health Organization (2007). A guide for population-based approaches to increasing
levels of physical activity: Implementation of the WHO global strategy on diet, physical
activity and health. Geneva, Switzerland: World Health Organization.
World Health Organization (2009). Global health risks: mortality and burden of disease
attributable to selected major risks. Geneva, Switzerland: World Health Organization.
World Health Organization (2010). Global recommendations on physical activity for health.
Geneva, Switzerland: World Health Organization.
Zetik, D. C., & Stuhlmacher, A. F. (2002). Goal setting and negotiation performance: A meta-
analysis. Group Processes & Intergroup Relations, 5, 35–52. doi:
10.1177/1368430202005001537
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 33
Figure 1. Results of literature search.
Records identified through database
searching
(n = 40,139)
Screenin
g
Included Eli
g
ibilit
y
Identification
Records screened after duplicates
removed
(n = 22,199)
Records excluded based on
review of titles and abstracts
(n = 21,698)
Full-text articles assessed for
eligibility
(n = 501)
Records excluded based on full-text
review (n = 456)
-Not a goal setting intervention (n =
209)
-Not a controlled experimental
study (n = 53)
-Goal setting was not an individual
PA goal (n = 120)
-Insufficient statistics (n = 44)
-Not relevant (n = 30)
Studies included in meta-analysis
(n = 45)
Records identified from manual
bibliographic searches
(n = 899)
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 34
Table 1. Summary descriptions of studies included in meta-analysis.
Study Design Sample Characteristics Physical Activity Measures Effect Size Calculation
Aittasalo et al.
(2004)
52-week RCT (2
experimental
conditions, 1
control)
Employees from various industries in
Finland (n: 155, age: 44±9, 56% female)
Pedometer: daily step count
Activity log: LTPA sessions per week; LTPA
minutes per week; Kcal expenditure per week;
MVPA sessions per week; MVPA minutes per
week
6 effect sizes: 2 experimental conditions
merged as 1 in analyses vs control;
differences in PA changes between
conditions from baseline to 12 months
Aittasalo et al.
(2012)
26-week RCT (1
experimental
condition, 1
control)
Employees from various industries in
Finland; Control (n: 118, age: 45.3 ±9.1),
Intervention (n: 123, age: 44.1±9.4)
Pedometer: daily step count
Activity log: number of steps in various
activities
5 effect sizes: differences between
conditions in PA at baseline and week 26
Altenburg et al.
(2015)
12-week RCT (1
experimental
condition, 1
control)
Persons with COPD from the
Netherlands (n: 155, age: 62, 66% male)
Pedometer: daily step count; daily PA 2 effect sizes: differences between
conditions in PA changes from baseline to
week 12
Annesi (2002) 52-week RCT (1
experimental
condition, 1
control)
Members of a northern Italian fitness
facility; Control (n: 50, age: 39.3±8.7,
70% female), Experimental (n: 50, age:
40.7±9.4, 68% female)
Attendance: % of each condition attending
fitness facility 3 times per week
1 effect size: difference in attendance
across the 52 weeks
Araiza et al.
(2006)
6-week RCT (1
experimental
condition, 1
control)
Individuals with type 2 diabetes mellitus;
Control (n: 15, age: 51±10), Intervention
(n: 15, age: 49±11)
Pedometer: daily step count 1 effect size: differences between
conditions in PA at baseline and week 6
Babazano et al.
(2007)
52-week RCT (1
experimental
condition, 1
control)
Older adults from Japan; Control (n: 41,
age: 65.7±7.8, 63% female),
Experimental (n: 46, age: 65.3±7.6, 54%
female)
Pedometer: daily step count 1 effect size: differences between
conditions in PA at baseline and week 52
Baker et al.
(2008)
12-week RCT (1
experimental
condition, 1
control)
Scottish community sample not meeting
current PA recommendations (n: 79,
age: 49.2 ± 8.8, 80% female)
Pedometer: daily step count,
1 effect size: differences between
conditions in PA at baseline and week 12
Baker et al.
(2011)
4-week RCT (2
experimental
condition, 1
control)
Adults from Scottish University (n: 61,
age: 42.1±10.6 years, 72% female )
Pedometer: daily step count 2 effect sizes: differences between
conditions in PA at baseline and week 4
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 35
Bickmore et al.
(2013)
52-week RCT (1
experimental
condition, 1
control)
Older adults from outpatient clinics in
Boston, USA; Control (n: 131, age:
70.8±5.2, 72% female), Experimental (n:
132, age: 71.7±5.6)
Pedometer: daily step count 1 effect size: difference between conditions
in step count changes from baseline to
week 52
Butler et al.
(2009)
6-week RCT (1
experimental
condition, 1
control)
Cardiac rehabilitation patients from New
South Wales, AUS; Control (n: 55, age:
64.5±11.2, 82% male), Experimental (n:
55, age: 63±10.4, 69% male)
Pedometer: mean and median PA minutes, PA
sessions, walking minutes, and walking
sessions
8 effect sizes; differences between
conditions in PA changes from baseline to
week 6
Bycura (2009) 4-week RCT (1
experimental
condition, 1
control)
Undergraduate students at University in
Southwest USA; Control (n: 15, age:
19.9), Experimental 1 (n: 29, age: 20),
Experimental 2 (n:40, age: 19.9)
Questionnaire: days exercised per week 2 effect sizes; differences between
conditions in PA changes at weeks 4 and
week 7
Devi et al.
(2014)
6-week RCT (1
experimental
condition, 1
control)
Persons with stable angina from a
primary care setting in one region of
England; Experimental (n: 48; age:
66.3±8.4; 71% male), Control (n: 46;
age: 66.2±10.0; 78% male)
Pedometer: daily step count 1 effect size: differences between
conditions at baseline and week 6
Dishman et al.
(2009)
12-week RCT (1
experimental
condition, 1
control)
Employees of 16 worksites across USA
and Toronto, CA (n: 965)
Questionnaire: moderate PA METS; vigorous
PA METS
Pedometer: walking METS
3 effect sizes: differences between
conditions in PA at baseline and week 12
Duncan &
Pozehl (2003)
24-week RCT (1
experimental
condition, 1
control)
Patients with heart failure in Nebraska,
USA (n: 14; age: 66.4; 86% male)
Activity log: exercise sessions completed 1 effect size: difference between conditions
in adherence changes from week 12 to
week 24
Fjeldsoe et al.
(2010)
12-week RCT (1
experimental
condition, 1
control)
Australian postnatal women (n: 88, age:
30 ±6 )
Questionnaire: MVPA and walking frequency 2 effect sizes: differences between
conditions in PA changes from baseline to
week 12
Furber et al.
(2008)
2-week RCT (1
experimental
condition, 1
control)
Persons with type 2 diabetes or glucose
intolerance from an Australian diabetes
service; Experimental (n: 121; age:
58.3±12.6; 46% female), Control (n:
105; age: 61.6±12.3; 48.6% female)
Questionnaire: walking time 1 effect size: differences between
conditions in walking time at baseline and
week 2
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 36
Furber et al.
(2010)
6-week RCT (1
experimental
condition, 1
control)
Australian cardiac patients who did not
attend a Cardiac Rehabilitation program;
Experimental (n: 104; age: 66.7±10.6;
71% male), Control (n: 111; age:
65.4±11.5; 69% male)
Questionnaire: Sessions and minutes of
walking and total PA
4 effect sizes: differences between
conditions in PA at baseline and week 6
Hallmark et al.
(2005)
6-week RCT (2
experimental
conditions,
control)
Adult women (n: 38) Pedometer: daily step count 2 effect sizes: differences between each
experimental condition and control
condition in step count at baseline and
week 6
Hancock (2005) 3-week RCT (1
experimental
condition, 1
control)
Primary school children in USA (n: 163) Pedometer: daily step count 1 effect size: difference between conditions
in step count at baseline and week 3
Hatchett (2008) 12-week RCT (1
experimental
condition, 1
control)
Women recovering from breast cancer
(n: 74, age: 53.35)
Questionnaire: total, moderate, and vigorous
PA
3 effect sizes: differences between
conditions in PA at baseline and week 12
Havenar (2007) 52-week RCT (1
experimental
condition, 1
control)
Adult community sample (n: 41, age:
45.6±12.6, 84% female)
Questionnaire: total minutes of MVPA/week,
total energy expenditure
2 effect sizes: differences between
conditions in PA at baseline and week 52
Hawkins et al.
(2014)
12-week RCT (1
experimental
condition, 1
control)
Pregnant women at risk for gestational
diabetes mellitus in Northeastern USA
(n: 290, age: 26.5)
Questionnaire: total PA 1 effect size: difference between conditions
in PA at baseline and week 12
Horne et al.
(2009)
14-week RCT (1
experimental
condition, 1
control)
Primary school children in Wales;
Experimental (Boys - n: 15, age: 9.9±0.7;
Girls - n: 23, age: 10±0.7), Control (Boys
- n: 28, age: 10.2±0.6; Girls - n: 23, age:
9.9±0.6)
Pedometer: daily step count 2 effect sizes: differences between
conditions in step counts at week 16 for
boys and for girls
Hospes et al.
(2009)
12-week RCT (1
experimental
condition, 1
control)
Persons with COPD in the Netherlands;
Experimental (n: 18, age: 63.1±8.3, 56%
male), Control (n: 17, age: 61.2±9.1,
65% male)
Pedometer: daily step count 1 effect size: difference between conditions
in step count at baseline and week 12
Houle et al.
(2011)
12-month RCT
(1 experimental
condition, 1
control)
Patients hospitalized for an acute
coronary syndrome in Quebec;
Experimental (n: 32, age: 58±8, 81%
male), Control (n: 33, age: 59±9, 76%
male)
Pedometer: daily step count 1 effect size: difference between conditions
in step count at baseline and week 52
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 37
Irvine et al.
(2013)
12-week RCT (1
experimental
condition, 1
control)
Sedentary adults from USA (n: 368, age:
60.3±4.9; 69% female)
Questionnaire: PA minutes per day in balance,
cardiovascular, strengthening, stretching, and
total activities
5 effect sizes: differences between
conditions in PA changes from baseline to
week 12
Kaminsky et al.
(2013)
8-week RCT (1
experimental
condition, 1
control)
Patients in a maintenance cardiac
rehabilitation program; Experimental (n:
10, age: 53.3±8.1, 80% male), Control
(n: 8, age: 59.4±9.9, 75% male)
Pedometer: daily step count 1 effect size: difference between conditions
in step count at baseline and week 8
Kovelis et al.
(2012)
1-month RCT (1
experimental
condition, 1
control)
Adult smokers in Brazil; Experimental
(n: 23, age: 51, 35% male), Control (n:
17, age: 52, 59% male)
Pedometer: daily step count 2 effect sizes: differences between
conditions in step count changes among
active and inactive (at baseline)
participants from baseline to week 4
Lombard (1994) 16-week RCT (1
experimental
condition, 1
control)
Community sample near a University in
Southeastern USA (n: 74, age: 37±10.5,
86% female)
Activity log: number of days active per week;
number of minutes active per week; average
minutes active per week; active or not
(dichotomous variable); meeting ACSM PA
guidelines
5 effect sizes: difference between
conditions in PA at baseline and week 16
Martin (1998) 3-week RCT (1
experimental
condition, 1
control)
Female undergraduate students at a
university in Eastern USA (n: 30, age:
25.8±8.6)
Questionnaire: number of exercise sessions,
minutes, and intensity per week
3 effect sizes: differences between
conditions in PA at baseline and week 4
Matthews et al.
(2007)
12-week RCT (1
experimental
condition, 1
control)
Breast cancer survivors in Eastern USA;
100% female; Experimental (n: 22, age:
51.3±9), Control (n: 14, age: 56.9±12.3)
Questionnaire: social activities; household
activities; lawn and garden work; non walking
exercise; self-reported walking; total activity
Accelerometer: activity counts; daily step
counts; moderate walking; % of activity at
light intensity; and % of activity at moderate-
vigorous intensity
11 effect sizes: differences between
conditions in PA at baseline and week 12
Maturi et al.
(2011)
12-week RCT (1
experimental
condition, 1
control)
Females in the postpartum period in Iran;
Experimental (n: 32, age: 25.7±4.6),
Control (n: 34, age: 24.8±3.7)
Pedometer: energy expenditure, and light,
moderate, vigorous PA
4 effect sizes: differences between
conditions in PA at baseline and week 12
Ornes (2006) 4-week RCT (1
experimental
condition, 1
control)
Female university students in Western
USA; Experimental (n: 60, age:
20.7±2.8), Control (n: 61, age: 20.6±3.5)
Pedometer: daily step count 1 effect size: differences between
conditions in step count at baseline and
week 4
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 38
Petersen et al.
(2012)
13-week RCT (1
experimental
condition, 1
control)
Adults from Denmark; Experimental (n:
192); Control (n: 173)
Questionnaire: total walking time 1 effect size: difference between conditions
in total walking time at week 13
Purath (2002) 6-week RCT (1
experimental
condition, 1
control)
Female university employees from
Midwestern USA (n: 271)
Questionnaire: minutes walked per week to
work, on errands, during work, and for
exercise; total minutes walked per week;
flights of stairs taken per day; blocks walked
per day; hours of weekday and weekend
MVPA
9 effect sizes: differences between
conditions in PA at baseline and week 6
Reijonsaari et
al. (2012)
52-week RCT (1
experimental
condition, 1
control)
Employees from an insurance company
in Finland (n: 43, age: 43, 64% female)
Questionnaire: weekly volume of PA 1 effect size: differences between
conditions in PA at baseline and week 52
Ribeiro et al.
(2014)
3-month RCT (2
experimental
conditions, 2
control)
Middle-aged women in Brazil;
Experimental 1 (n: 53, age: 45±3),
Experimental 2 (n: 48, age: 45±3),
Control 1 (n: 47, age: 45±3), Control 2
(n: 47; age: 45±3)
Pedometer: total and moderate steps 4 effect sizes: differences between each
experimental condition and control
condition in PA changes from baseline and
week 13
Sawchuk et al.
(2011)
6-week RCT (1
experimental
condition, 1
control)
American Indian/Alaska Native elders;
Control condition (n: 19, age: 61±8.4,
68% female), Experimental condition (n:
17, age: 62±9.8, 71% female)
Questionnaire: calorie expenditure (all
activities), calorie expenditure (moderate PA);
exercise bouts per week, bouts of moderate
PA per week
Pedometer: daily step count, total steps over 5
weeks; daily step count for 5 weeks
7 effect sizes: differences between
conditions in PA at week 6
Schofield et al.
(2005)
12 week RCT
(2 experimental
conditions, 1
control)
Adolescent females from Australian high
schools; Experimental 1 (n: 23, age:
15.9±0.8), Experimental 2 (n: 21, age:
15.7±0.8), Control (n: 24, age: 15.9 ±0.8)
Pedometer: daily step count
Questionnaire: vigorous and moderate-
vigorous PA
6 effect sizes: differences between each
experimental condition and control
condition in PA at baseline and week 12
Strath et al.
(2011)
12-week RCT (2
experimental
conditions, 1
control)
Older adults from Central USA; Control
(n: 20, age: 64.9±7.1, 80% female),
Experimental 1 (n: 20, age: 63.3±5.8,
85% female), Experimental 2 (n: 20, age:
63.6±4.2, 86% female)
Pedometer: daily step count 2 effect sizes: differences between each
experimental condition and control
condition in step count at week 12
Talbot et al.
(2003)
12-week RCT (1
experimental
condition, 1
control)
Older adults from Eastern USA;
Experimental (n: 17, age: 69.6±6.7, 77%
female), Control (n: 17, age: 70.8±4.7,
77% female)
Pedometer: daily step count, total vector 2 effect sizes: differences between
conditions in PA at baseline and week 12
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 39
Tudor-Locke et
al. (2004)
16-week RCT (1
experimental
condition, 1
control)
Adults with type 2 diabetes;
Experimental (n: 24, age: 52.8±5.2, 50%
male), Control (n: 23, age: 52.5±4.8,
61% male)
Pedometer: daily step count 1 effect size: differences between
conditions in step count at baseline and
week 16
Wadsworth
(2005)
6-week RCT (1
experimental
condition, 1
control)
Female university students from
Southeastern USA; Experimental (n: 45),
Control (n: 46)
Accelerometer: days and minutes of moderate
and vigorous PA
4 effect sizes: differences between
conditions in PA at baseline and week 6
Wang (2004) 6-week RCT (1
experimental
condition, 1
control)
Female middle school students in USA;
Experimental (n: 24), Control (n: 22)
Pedometer: daily step count 1 effect size: differences between
conditions in step count at baseline and
week 6
Watson et al.
(2012)
12-week RCT (1
experimental
condition, 1
control)
Adults from Boston, USA; Experimental
(n: 27; age: 44.1; 89% female), Control
(n: 30; age: 40.6; 80% female)
Pedometer: daily step count 1 effect size: difference between conditions
in step count changes from baseline to
week 12
Note: Although some studies had more experimental conditions than reported here, some of those conditions did not meet eligibility criteria. Therefore, the results
given in this meta-analysis only reflect outcomes from the experimental conditions that met our inclusion criteria; RCT = Randomized Controlled Trial; PA =
physical activity; MET: Metabolic Equivalence Test; LTPA = leisure time physical activity; MVPA = moderate-vigorous physical activity; kcal = kilocalories; NR
= not reported
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 40
Table 2. Summary results of studies included in meta-analysis.
Study Relative
Weight
Effect Size
(SE)
95% CI
(lower, upper)
Z-value p-value ES with study
removed
Aittasalo et al. (2004) 2.55 -.111 (.17) -.45, .22 -0.65 .513 0.568
Aittasalo et al. (2012) 2.65 .163 (.15) -.14, .46 1.06 .289 0.563
Altenburg et al. (2015) 2.45 .618 (.19) .25, .99 3.30 .001 0.551
Annesi (2002) 2.24 1.339 (.22) .91, 1.77 6.05 <.001 0.530
Araiza et al. (2006) 1.41 .902 (.38) .15, 1.65 2.35 0.019 0.547
Babazano et al. (2007) 2.22 .900 (.23) .46, 1.34 3.99 <.001 0.544
Baker et al. (2008) 2.17 .779 (.23) .32, 1.24 3.34 .001 0.547
Baker et al. (2013) A 1.40 .346 (.39) -0.41, 1.10 0.89 .371 0.555
B 1.34 .909 (.40) 0.12, 1.70 2.27 .023 0.547
Bickmore et al. (2013) 2.79 .328 (.13) .08, .58 2.58 .010 0.560
Butler et al. (2009) 2.33 .573 (.21) .17, .98 2.77 .006 0.552
Bycura (2009) A 1.34 -.311 (.40) -1.10, .48 -0.78 .437 0.563
B 1.40 -.146 (.39) -.91, .61 -0.38 .707 0.562
Devi et al. (2014) 2.16 .475 (.24) .02, .94 2.02 .043 0.554
Dishman et al. (2009) 3.03 -.048 (.08) -.19, .10 -0.65 .517 0.566
Duncan & Pozehl (2003) 0.70 1.908 (.65) .64, 3.17 2.96 .003 0.542
Fjeldsoe et al. (2010) 2.27 .847 (.22) .05, .90 2.19 <.029 0.554
Furber et al. (2008) 2.72 .914 (.14) .64, 1.19 6.52 <.001 0.540
Furber et al. (2010) 2.72 .320 (.14) .04, .60 2.27 .023 0.560
Hallmark et al. (2005) A 1.02 .437 (.50) -.54, 1.41 0.88 .381 0.553
B 0.98 .857 (.51) -.15, 1.86 1.67 .095 0.549
Hancock (2005) 2.16 .683 (.24) .22, 1.15 2.89 .004 0.549
Hatchett (2008) 2.11 .091 (.24) .42, 1.38 3.69 <.001 0.544
Havenar (2007) 1.54 1.07 (.35) .38, .1.76 3.04 .002 0.543
Hawkins et al. (2014) 2.81 .314 (.13) .07, .56 2.51 .012 0.560
Horne et al. (2009) 1.71 .625 (.32) .003, 1.25 1.97 .049 0.551
Hospes et al. (2009) 1.58 .596 (.35) -.08, 1.27 1.75 .085 0.552
Houle el al. (2011) 2.05 .617 (.25) .12, 1.11 2.43 .015 0.551
Irvine et al. (2013) 2.83 .649 (.12) .42, .88 5.44 <.001 0.550
Kaminsky et al. (2013) 1.01 .961 (.50) -.02, 1.94 1.92 .055 0.548
Kovelis (2012) A 1.01 -.297 (.50) -1.28, .69 -.59 .555 0.560
B 1.20 .878 (.44) .02, 1.74 1.99 .046 0.548
Lombard (1994) 2.31 .636 (.21) .23, 1.05 3.03 .002 0.550
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 41
Martin (1998) 1.44 .349 (.38) -.39, 1.09 0.93 .354 0.555
Matthews et al. (2007) 1.48 .710 (.37) -.01, 1.43 1.94 .053 0.550
Maturi et al. (2011) 2.06 .255 (.25) -.24, .75 1.01 .313 0.559
Ornes (2006) 2.38 .900 (.20) .51, 1.29 4.53 <.001 0.543
Petersen et al. (2012) 2.86 .296 (.12) .07, .52 2.58 .010 0.561
Purath (2002) 2.82 .207 (.12) -.03, .45 1.68 .093 0.563
Reijonsaari et al. (2012) 2.90 -.068 (.11) -.28, .14 -0.64 .525 0.569
Ribeiro et al. (2014) A 1.97 1.184 (.27) .66, 1.71 4.40 <.001 0.538
B 2.05 .793 (.25) .30, 1.29 3.12 .002 0.547
Sawchuk et al. (2011) 1.62 -.104 (.34) -.76, .55 -0.31 .755 0.563
Schofield et al. (2005) A 1.46 .559 (.37) -.17, 1.29 1.51 .132 0.552
B 1.50 .479 (.36) -.23, 1.91 1.32 .187 0.553
Strath et al. (2011) A 1.04 1.636 (.49) .67, 2.60 3.32 .001 0.540
B 1.04 1.636 (.49) .67, 2.60 3.32 .001 0.540
Talbot et al. (2003) 1.57 .350 (.35) -.33, 1.03 1.01 .312 0.556
Tudor-Locke et al. (2004) 1.73 1.102 (.31) .49, 1.72 3.52 <.001 0.542
Wadsworth (2005) 2.17 .191 (.23) -.27, .65 0.82 .414 0.560
Wang (2004) 1.71 1.108 (.32) .49, 1.73 3.50 <.001 0.542
Watson et al. (2012) 1.96 .572 (.27) .04, 1.10 2.11 .034 0.552
OVERALL 100 .552 (0.06) 0.43, 0.67 9.03 <.001
Note: A and B = experimental groups within study; SE = standard error; CI = confidence interval; ES = effect size
Effect Size
-2. 00 0.00 2.00
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 42
Table 3. Results of moderator analyses for study, sample, intervention characteristics, and components of the goal setting process.
Moderator k Effect size
(SE)
95% CI Z-value P-value Q value (df),
p-value
Study Characteristics
Publication Source 0.01(1), p=0.933
Peer-reviewed journal 39 0.555 (0.07) 0.42, 0.69 7.90 <0.001
Other (thesis or conference) 13 0.543 (0.13) 0.29, 0.79 4.27 <0.001
Theoretical Framework 1.67(3), p=0.433
Atheoretical/not specified 21 0.487 (0.10) 0.29, 0.68 4.83 <0.001
Theoretical 31 0.593 (0.08) 0.44, 0.75 7.54 <0.001
-single theory 19 0.541 (0.09) 0.36, 0.73 5.73 <0.001
-multiple theories 12 0.709 (0.14) 0.43, 0.99 5.03 <0.001
Intervention Setting 17.12(6), p=0.009
Workplace 7 0.211 (0.11) -0.01, 0.43 1.89 0.059
Medical/rehabilitation/primary care center 9 0.653 (0.12) 0.43, 0.88 5.62 <0.001
University 5 0.434 (0.17) 0.10, 0.77 2.51 0.012
Grade school 3 0.792 (0.22) 0.36, 1.22 3.61 <0.001
Fitness facility 1 1.339 (0.33) 0.69, 1.99 4.02 <0.001
Home 7 0.566 (0.13) 0.32, 0.82 4.45 <0.001
Not specified 20 0.574 (0.09) 0.39, 0.75 6.43 <0.001
Mode of Intervention Delivery 2.41(3), p=0.492
In person 22 0.622 (0.09) 0.44, 0.80 6.68 <0.001
Technology 10 0.421 (0.12) 0.18, 0.66 3.46 0.001
Multiple methods 11 0.627 (0.13) 0.36, 0.89 4.68 <0.001
Unclear 9 0.457 (0.16) 0.15, 0.76 2.91 0.004
Type of Physical Activity 15.60(2), p<0.001
Aerobic activity 35 0.635 (0.06) 0.51, 0.76 9.95 <0.001
Any/multiple types 10 0.607 (0.11) 0.39, 0.82 5.49 <0.001
Not specified 7 0.142 (0.11) -0.07, 0.36 1.29 0.196
PA Intensity Prescribed 2.30(3), p=0.513
Moderate 7 0.709 (0.17) 0.39, 1.03 4.29 <0.001
Moderate-vigorous 5 0.323 (0.20) -0.07, 0.72 1.60 0.109
Any 4 0.494 (0.19) 0.12, 0.87 2.60 0.009
Not specified 36 0.562 (0.08) 0.41, 0.71 7.41 <0.001
PA Measure 16.85(4), p=0.002
Technology 20 0.671 (0.09) 0.49, 0.85 7.25 <0.001
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 43
-pedometer 18 0.717 (0.09) 0.54, 0.90 7.69 <0.001
-accelerometer 2 0.333 (0.25) -0.15, 0.82 1.34 0.179
Self-report 18 0.525 (0.09) 0.35, 0.70 5.83 <0.001
-PA questionnaire 16 0.449 (0.09) 0.28, 0.62 5.15 <0.001
-activity log/diary 2 1.450 (0.31) 0.85, 2.05 4.73 <0.001
Multiple 18 0.525 (0.11) 0.19, 0.61 3.77 <0.001
Sample Characteristics
Age 3.15(4), p=0.534
Children 2 0.657 (0.31) 0.04, 1.27 2.10 0.036
Youth 3 0.735 (0.29) 0.18, 1.29 2.58 0.010
Adults 36 0.523 (0.07) 0.38, 0.67 7.17 <0.001
Older adults 5 0.417 (0.18) 0.06, 0.77 2.31 0.021
Adults and older adults 6 0.849 (0.22) 0.43, 1.27 3.95 <0.001
Sex 0.05(1), p=0.831
Female only 16 0.573 (0.11) 0.35, 0.79 5.11 <0.001
Both males and females 36 0.544 (0.07) 0.40, 0.69 7.37 <0.001
Baseline Weight Status (Sample Mean BMI) 1.49(3), p=0.684
Healthy 5 0.488 (0.23) 0.03, 0.95 2.09 0.037
Overweight 26 0.567 (0.09) 0.40, 0.73 6.68 <0.001
Obese 3 0.843 (0.27) 0.31, 1.38 3.10 0.002
Not specified 18 0.502 (0.10) 0.30, 0.70 4.88 <0.001
Baseline Activity Levels (Sample Mean) 0.14(1), p=0.712
Does not meet PA guidelines at baseline 46 0.561 (0.07) 0.43, 0.69 8.51 <0.001
Meets PA guidelines at baseline 6 0.491 (0.18) 0.14, 0.84 2.74 0.006
Population Type 0.63(1), p=0.428
General population 31 0.512 (0.08) 0.36, 0.66 6.61 <0.001
Special population 21 0.609 (0.10) 0.42, 0.80 6.38 <0.001
Goal Content
Goal Specificity 2.13(3), p=0.545
Specific 31 0.589 (0.08) 0.43, 0.75 7.01 <0.001
-absolute 7 0.447 (0.22) 0.03, 0.87 2.08 0.038
-relative 19 0.673 (0.11) 0.47, 0.88 6.34 <0.001
-absolute and relative 5 0.422 (0.20) 0.03, 0.82 2.08 0.037
Vague/unclear 21 0.511 (0.10) 0.33, 0.70 5.36 <0.001
Source of Goal Prescription 0.76(3), p=0.860
Participant 9 0.625 (0.14) 0.36, 0.90 4.54 <0.001
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
Accepted Manuscript
Goal Setting Interventions 44
Interventionist 23 0.583 (0.11) 0.38, 0.79 5.56 <0.001
Collaborative 9 0.473 (0.14) 0.20, 0.75 3.38 0.001
Unclear 11 0.517 (0.14) 0.25, 0.78 3.81 <0.001
Goal Timeframe 9.89(3), p=0.020
Daily 28 0.600 (0.08) 0.44, 0.76 7.48 <0.001
Weekly 7 0.152 (0.15) -0.14, 0.45 1.02 0.310
Daily and weekly 2 0.947 (0.26) 0.45, 1.45 3.71 <0.001
Unclear/other 15 0.562 (0.09) 0.38, 0.75 6.02 <0.001
Frequency of Goal Setting/Modifications 1.55(4), p=0.819
At baseline only 6 0.568 (0.20) 0.17, 0.97 2.79 0.005
Daily 1 0.328 (0.36) -0.38, 1.04 0.91 0.366
Weekly 16 0.641 (0.11) 0.42, 0.86 5.70 <0.001
Bi-weekly 5 0.634 (0.23) 0.19, 1.08 2.78 0.005
Any time 24 0.499 (0.08) 0.33, 0.66 5.92 <0.001
Goal-Related Behaviour Change Techniques
Feedback 0.16(1), p=0.693
No 3 0.646 (0.24) 0.17, 1.12 2.66 0.008
Yes 49 0.547 (0.06) 0.42, 0.68 8.50 <0.001
Strategy planning 0.50(1), p=0.480
No 21 0.498 (0.10) 0.30, 0.70 4.95 <0.001
Yes 31 0.588 (0.08) 0.43, 0.74 7.41 <0.001
Rewards 0.46(1), p=0.499
No 44 0.570 (0.07) 0.44, 0.70 8.52 <0.001
Yes 8 0.462 (0.15) 0.18, 0.75 3.15 0.002
Method co-occurrence effects 3.42(5), p=0.635
Goal setting only 1 0.207 (0.37) -0.52, 0.93 0.56 0.575
Partial application – goal setting plus: 45 0.582 (0.07) 0.45, 0.71 8.70 <0.001
-feedback 18 0.517 (0.11) 0.30, 0.73 4.72 <0.001
-strategies 2 0.970 (0.32) 0.35, 1.59 3.05 0.002
-feedback and strategies 23 0.600 (0.09) 0.42, 0.78 6.48 <0.001
-feedback and rewards 2 0.541 (0.31) -0.07, 1.16 1.73 0.084
Complete application – goal setting plus all three
of feedback, strategies, and rewards
6 0.440 (0.17) 0.11, 0.77 2.61 0.009
Downloaded by [The University of British Columbia] at 08:13 13 October 2015
... Goal-setting is one of the most widely used and highly effective behavioral change strategies for promoting peoples' PA and helping them to adopt more health-related behaviors (McEwan et al., 2016;Pearson, 2012;Shilts et al., 2004;Swann et al., 2021). For example, in a meta-analysis of McEwan et al. (2016, p. 67) results showed "a medium, positive effect of goal setting interventions in relation to PA behaviour". ...
... Thus, it could be suitably recommended for use and application by exercise specialists and/or health care professionals as a strategy to change or improve health behaviors of MS patients. The above results are in line with the vast majority of studies showing that goal-setting strategy is probably one of the strongest facilitators of improving physical activity levels, dietary and smoking behaviors (Lorencatto et al., 2016;McEwan et al., 2016;Shilts et al., 2004;Swann et al., 2021). ...
Article
Full-text available
p> Background. A healthy lifestyle is crucial for patients with multiple sclerosis (MS). Objective. The purpose of the study was to examine the effect of a combined exercise and goal-setting program of Greek patients with MS on increasing physical activity (PA) level, adopting healthier nutritional behaviors and reduce smoking. Methods. The sample consisted of 30 patients with multiple sclerosis, 15 men and 15 women, aged 23 to 65 years, randomly assigned into two equal (experiment and control) groups. The experiment group participated in an 8-week exercise program combined with nutrition and goal-setting strategies with a purpose to adopt and maintain a healthier lifestyle. The control group did not participate in any of the intervention procedures. Both groups completed -pre and post intervention- questionnaires measuring their leisure time PA, nutritional behaviors and smoking dependence. Results. The results showed higher rates of leisure time PA, improved eating habits and reduced smoking behaviors of experiment group participants following the 8-week intervention program compared to control group participants. Conclusions. Based on findings, further recommendations were made concerning PA levels, nutrition and smoking cessation of patients with multiple sclerosis. Article visualizations: </p
... Meta-analysis on health behavior change techniques used in interventions concludes that self-monitoring, especially when combined with prompts for intention formation, specific goal setting, review of behavioral goals and feedback on performance, is the most effective technique for behavior change in healthy eating and physical activity (Michie et al., 2009). Using goal setting theory as theoretical framework, McEwan et al. (2016) conduct systematic review and metaanalysis to evaluate the effectiveness of multicomponent goal setting interventions. This type of intervention has significant effect sizes when daily targets, feedback and strategy planning are used. ...
... In a recent study comparing the effects of open, SMART and do-your-best goals , the authors reported that SMART goals might be detrimental to intrinsic motivation, stress and affect. According to systematic review and metaanalysis of multi-component goal-setting interventions, the specificity of goals also seemed rather irrelevant (McEwan et al., 2016). Based on support for them from both application domains, we hypothesize that specific goals have a positive influence on any kind of goal, i.e., on both process-and outcome-focused goals: ...
Preprint
Full-text available
Commercial mobile applications implement several theory-driven features, such as feedback, self-monitoring and goal-setting, but they often fail to drive long-lasting changes in health behavior. The nature of users' goals and how current systems support them might be one reason for that. This study uses the lens of the exerciser self-schema, a physical activity-specific identity variable, to study how commercial self-monitoring systems support process-and outcome-focused goals. A survey of mobile app users monitoring physical activity resulted in 238 valid responses. 59% of the respondents were identified as exerciser schematics. Exercisers differed significantly from those without exercise self-schema (unschematics) in all measured constructs except for outcome-focused goals. Exercisers evaluated their systems as more persuasive and seemed to be more focused on process goals. Process goals contributed positively to outcome goals in all studied subgroups. Exercisers preferred process goals, but the feedback and self-monitoring features of systems do not support this goal focus. Process goals can be considered a means through which to reach outcome goals. Process goals also mediated the well-documented effects of goal specificity. Improving the ability to set and monitor specific process goals seems promising to improve these systems.
... Within the current review, the absence of the BCT '1.1 goal setting' was associated with a statistically significant moderate effect size. Goal setting is a commonly used BCT within physical activity and sedentary behaviour intervention research (McEwan et al., 2016). However, research often produces conflicting results in relation to this BCT, with some studies indicating goal setting enhanced intervention effectiveness (Dishman et al., 2010) and others find limited support for its use (Shilts et al., 2004). ...
Thesis
Full-text available
Globally, physical activity levels have declined sharply and it has been estimated that up to 42% of individuals within developed countries are classified as being physically inactive. Insufficient physical activity is a substantial health risk and has been associated with negative psychophysiological outcomes including cardiovascular disease, diabetes and depression. Whilst there are many contributors to physical inactivity the workplace has been identified as a particularly significant contributor. Consistently high levels of sedentary behaviour have been documented within many modern workplaces, with employees spending up to 81% of working hours seated in white collar roles. Given that approximately 58% of global workforce will spend one third of their adult life at work, the workplace has been identified as a key domain in which researchers can deliver interventions to promote physical activity. Despite this, evidence for the efficacy of workplace physical activity interventions has been mixed. One potential explanation for this is an underutilisation of participatory approaches during intervention design. Within organisational research, concerns have been expressed regarding a widening gap between research and practice.Whilst interventions may be academically robust they may lack sufficient relevancy to the employees that they are intended to support. To address these issues this thesis adopted a pragmatic, participatory stance and drew upon co-creation methodologies to develop a new workplace physical activity intervention that would meet the needs of employees.
... A systematic review by Epton et al. [69] of 141 papers on behavioral interventions through GS found that GS has positive intervention effects, is an effective approach to behavior change, and can be considered an essential component of conducting successful interventions. Desmond et al. [70], after a systematic review of 45 experimental literature on interventions for physical activity behaviors through multicomponent GS, noted that GS interventions related to physical activity behaviors had moderate positive effects (Cohen's d = 0.552). Short-term goals are the most likely to have an immediate motivational impact on human action, and clear, specific, measurable goals generate a greater motivational drive and lead to good grades [71,72]. ...
This study aimed to analyze the impacts of a 12-week core strength training (CST) and goal-setting (GS) program on the core endurance, agility, sprinting, jumping, grip strength, and exercise attitude in a group of adolescents. This study followed a randomized parallel design in which 362 adolescents (age: 14.5 ± 1.07 years; body mass index: 19.82 ± 3.64) were allocated to a GS (n = 89), CST (n = 92), or GS + CST (n = 90) program or to a control group (n = 91). Participants were assessed two times (baseline and postintervention) for the following tests: (i) 50 m dash, (ii) grip strength, (iii) long jump, (iv) 1000 m running for boys and 800 m for girls, (v) core endurance, and (vi) exercise attitude. Significant differences (p < 0.05, η2p = 0.035-0.218) were found between the four groups of the six components of physical fitness and the three components of attitude toward exercise (target attitudes, behavioral habits, and sense of behavioral control). Between-group analysis revealed that the GS + CST had significant advantages (p < 0.05) over the CON in terms of the 50 m dash (Cohen's d = 0.06), grip strength (Cohen's d = 0.19_left, 0.31_right), 800/1000 m running (Cohen's d = 0.41), core endurance (Cohen's d = 0.95), and sense of behavioral control (Cohen's d = 0.35). Between-group analysis also revealed that the CST had significant advantages over the CON in terms of grip strength (Cohen's d = 0.27_left, 0.39_right), 50 m (Cohen's d = 0.04), long jump (Cohen's d = 0.21), 800/1000 m (Cohen's d = 0.09), and core stability (Cohen's d = 0.63), which were significantly different from CON (p < 0.05). GS differed from CON only on 50 m (Cohen's d = 0.02) and core stability (Cohen's d = 0.13) with a small effect (p < 0.05). We conclude that the combined intervention of GS and CST is more effective in promoting fitness in adolescents, i.e., GS + CST > CST and GS + CST > GS.
... Self-affirmation was also effective with students and adults (for a review, see Cohen & Sherman, 2014). Goal setting was successfully applied in the fields of sport (for a review, see Jeong et al., 2021), health (for a review, McEwan et al., 2016), education (Bruhn et al., 2016) and behavior change (Epton et al., 2017). ...
Article
Full-text available
Increasing well-being is a prominent worldwide goal that can be achieved primarily through social support and environmental factors. However, in times of social distancing or isolation, it is important to also rely on self-managed activities. This study aimed to (a) test the effectiveness of a seven-week well-being intervention, in increasing need satisfaction, self-compassion, emotion regulation, and grateful disposition by curbing need frustration, self-derogation, and emotional suppression, and (b) examine the maintenance and long-term effects of the practices based on recall, elaboration, and writing. One hundred and twenty university students weekly recalled and elaborated for seven consecutive weeks on three recent episodes of gratitude, self-affirmation, goal setting, or meaningful things, according to the group to which they were assigned. Before the intervention, immediately after and one month later, they filled in questionnaires to assess need satisfaction/frustration, self-compassion/derogation, emotion regulation and grateful disposition. The results confirmed an increase in well-being and a decrease in ill-being for all groups (Cohen d for the significant differences ranging from 0.18 to 0.53). The effects were maintained one month later and even increased for self-compassion, self-derogation, need frustration, and emotional reappraisal. A follow-up assessment revealed that a third of the participants continued with the well-being practices. Implications and suggestions for future well-being interventions are discussed.
... Recent research on goal setting revealed that interventions that set weekly or daily goals produced greater effects on PA than goals set over a longer time frame. 67 Moreover, it appears better to consider the achievement of the goals in 'percentage of objective achieved' rather than in a binary way (success/fail) in order to inform that the objective is reached or close to being reached. 68 Following these recommendations, the initial step goal at the beginning of the programme will be based on the daily step count of the evaluation week. ...
Article
Full-text available
Introduction Physical inactivity and excessive sedentary behaviours are major preventable causes in both the development and the treatment of obesity and type 2 diabetes mellitus (T2DM). Nevertheless, current programmes struggle to engage and sustain physical activity (PA) of patients over long periods of time. To overcome these limitations, the Digital Intervention Promoting Physical Activity among Obese people randomised controlled trial (RCT) aims to evaluate the effectiveness of a group-based digital intervention grounded on gamification strategies, enhanced by social features and informed by the tenets of the self-determination theory and the social identity approach. Methods and analysis This trial is a two-arm parallel RCT testing the effectiveness of the Kiplin digital intervention on obese and patients with T2DM in comparison to the usual supervised PA programme of the University Hospital of Clermont-Ferrand, France. A total of 50 patients will be randomised to one of the two interventions and will follow a 3-month programme with a 6-month follow-up postintervention. The primary outcome of the study is the daily step count change between the baseline assessment and the end of the intervention. Accelerometer data, self-reported PA, body composition and physical capacities will also be evaluated. To advance our understanding of complex interventions like gamified and group-based ones, we will explore several psychological mediators relative to motivation, enjoyment, in-group identification or perceived weight stigma. Finally, to assess a potential superior economic efficiency compared with the current treatment, we will conduct a cost–utility analysis between the two conditions. A mixed-model approach will be used to analyse the change in outcomes over time. Ethics and dissemination The research protocol has been reviewed and approved by the Local Human Protection Committee (CPP Ile de France XI, No 21 004-65219). Results will inform the Kiplin app development, be published in scientific journals and disseminated in international conferences. Trial registration number NCT04887077 .
... Further, self-monitoring showed some effectiveness in the reduction of substance use (Gass et al., 2021) and the promotion of PA (Izawa et al., 2005;Noland, 2013). Goal setting has been effectively used as a motivational technique in PA interventions (Kyllo & Landers, 1995;McEwan et al., 2015) and by means of motivational interviewing (MI; Miller & Rollnick, 2013) in SUD treatment. Meanwhile, pharmacological support was most frequently used in smoking cessation interventions. ...
Article
Full-text available
Objective: Increasing regular physical activity (PA) behaviour may be an effective adjunct intervention for substance use disorder (SUD) treatment. This systematic review aims to identify promising behaviour change techniques (BCTs), namely, BCTs present in the design of interventions evidencing significant short-term and/or long-term (d ≥ 0.15 for objective measures and d ≥ 0.36 for self-report measures) increase in PA and/or reduction of substance use, secondary psychological measures, and retention in the PA intervention. Method: PRISMA guidelines were followed, and the search was performed on March 11, 2021 across databases including MEDLINE, PsycINFO, SPORTDiscus, Cochrane Library, CINAHL, ProQuest, Web of Science Core Collection, Google Scholar; Open Grey, and ProQuest Dissertations & Theses. Studies were included if they measured PA, included participants aged ≥18 years, were randomised control trials, and if participants were diagnosed with SUDs. The Cochrane RoB 2.0 Tool was used to assess risk of bias. BCTs from eligible studies were extracted, coded, and ranked according to their proportional presence across studies. Results: The final synthesis included k = 61 studies with N = 12,887 participants. High heterogeneity across outcome measures, interventions and control conditions was found. In total, 477 applications of BCTs were identified. Instruction on how to perform the behaviour, social support (unspecified), behavioural practice/rehearsal, problem solving, pharmacological support, goal setting (behaviour), self-monitoring (behaviour), and biofeedback were the eight most frequently used promising BCTs across studies. Conclusions: Incorporating the eight most promising BCTs identified in this review in future PA interventions in SUD populations may improve SUD outcomes.
... Goal setting is considered a fundamental component of successful behavior change and is the most frequently used component in health behavior interventions [23]. Evidence from systematic reviews and meta-analyses has shown goal-setting interventions to have small [24] to moderate [25] positive effects on PA. As part of the initial onboarding of the JitaBug app, we implemented a goal-setting feature that allowed participants to choose a step count or activity minutes goal depending on their preference. ...
... Goal setting is considered a fundamental component of successful behavior change and is the most frequently used component in health behavior interventions [23]. Evidence from systematic reviews and meta-analyses has shown goal-setting interventions to have small [24] to moderate [25] positive effects on PA. As part of the initial onboarding of the JitaBug app, we implemented a goal-setting feature that allowed participants to choose a step count or activity minutes goal depending on their preference. ...
Article
Background: Just-in-time adaptive interventions (JITAIs) provide real time in-the-moment behavior change support to people when they need it most. JITAIs could be a viable way to provide personalized physical activity (PA) support to older adults in the community. However, it is unclear how feasible it is to remotely deliver a PA intervention through a smartphone to older adults or how acceptable they would find a JITAI targeting PA in everyday life.
Article
Grounded in social cognitive theory (SCT), this study sought to examine whether parents perceived social cognitive factors regarding children's physical activity (PA) behaviors were associated with preschool children's moderate-to-vigorous PA (MVPA) levels. A total of 142 Hong Kong parent-child pairs from five preschools/childcare centers completed all assessments in the cross-sectional study. Children’s (42% girls; mean age = 4.52 ± 0.67 years) PA was measured through accelerometers. Parents (74% mothers; mean age = 37.38 ± 4.63 years) completed a paper-based questionnaire assessing the social cognitive factors on their children’s PA participation. The data were analyzed using latent variable structural equation modeling. Findings revealed that the model showed acceptable fit with the data: χ² (23) = 38.14, p = .025, χ²/df = 1.66, CFI = 0.955, TLI = 0.929, RMSEA = 0.068, 90% CI [0.025, 0.106], and SRMR = 0.072. The model accounted for 39.1% of the variance in the PA behavior of preschool-aged children. Structural equation modelling revealed parental self-efficacy (β = 0.29, 95% CI [0.95, 0.49]) and goal setting (β = 0.25, 95% CI [0.06, 0.44]) were directly associated with children’s MVPA. Outcome expectations (β = 0.09, 95% CI [0.01, 0.03]) and goal setting (β = 0.18, 95% CI [0.05, 0.32]) mediated the association between parental self-efficacy and children’s MVPA. Indirect associations of parental self-efficacy from setting goals via parental support (β = 0.15, 95%CI [0.02, 0.30]) and perceived barriers (β = 0.15, 95% CI [0.05, 0.28]) were uncovered. Results supported the use of SCT in understanding how the parents perceived social cognitive factors predict the PA behaviors of young children. This study provides insight into whether these theoretical variables could be modified or promoted in future intervention programs. Enhancing parents’ abilities to ensure preschool-aged children are physically active is of great importance given the global decline in PA among children.
Article
Full-text available
Systematic reviews should build on a protocol that describes the rationale, hypothesis, and planned methods of the review; few reviews report whether a protocol exists. Detailed, well-described protocols can facilitate the understanding and appraisal of the review methods, as well as the detection of modifications to methods and selective reporting in completed reviews. We describe the development of a reporting guideline, the Preferred Reporting Items for Systematic reviews and Meta-Analyses for Protocols 2015 (PRISMA-P 2015). PRISMA-P consists of a 17-item checklist intended to facilitate the preparation and reporting of a robust protocol for the systematic review. Funders and those commissioning reviews might consider mandating the use of the checklist to facilitate the submission of relevant protocol information in funding applications. Similarly, peer reviewers and editors can use the guidance to gauge the completeness and transparency of a systematic review protocol submitted for publication in a journal or other medium.
Conference Paper
Full-text available
BACKGROUND Mental health disorders and self-assessed mental health problems are common among students in tertiary education in Western countries. On average, one in three university students suffer from depressive symptoms and more female students are affected than males. Student mental health services are relevant settings for promoting mental health and preventing mental ill-health through interventions at the organizational-, group-and individual levels. However, student's problems differ across campuses and many intervention programs and policies are not based on the best available evidence. More over, sound interventions may remain without effect unless they are thoroughly implemented.