Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jan 1.
Published in final edited form as: Health Psychol. 2020 Nov 23;40(1):21–29. doi: 10.1037/hea0001041

Psychosocial Mediators of Physical Activity Change in a Web-Based Intervention for Latinas

Britta Larsen a, Shira I Dunsiger b, Dori Pekmezi c, Sarah Linke a, Sheri J Hartman a, Bess H Marcus a,b
PMCID: PMC8409153  NIHMSID: NIHMS1727615  PMID: 33370154

Abstract

Objective:

To determine whether psychosocial constructs targeted in an online physical activity intervention for Latinas mediated changes in moderate-to-vigorous physical activity (MVPA).

Methods:

Data were taken from a randomized trial of a web-based MVPA intervention for Latina women age 18–65 (N=205) based on Social Cognitive Theory and the Transtheoretical Model. Baseline and 6-month measures included minutes/week of MVPA (ActiGraph GT3X+ accelerometers and 7-Day Physical Activity Recall Interview) and theorized mediators (self-efficacy, behavioral processes, cognitive processes, social support, enjoyment). A multiple mediation model adjusting for baseline MVPA was fit using a products of coefficients method, simultaneously testing all hypothesized mediators.

Results:

MVPA increased more in the intervention group than controls by 50 minutes/week (self-report) and 31 minutes/week (accelerometers). For the self-reported MVPA model, there was an intervention effect (a-path coefficient) on self-efficacy (b=0.43, p<0.01), cognitive processes (b=0.64, p<0.01), behavioral processes (b=0.54, p<0.01), and enjoyment (b=9.91, p=0.01). Changes in self-efficacy (b=24.54, p=0.03), social support from friends (b=2.36, p=0.04), and enjoyment (a=0.74, p=0.08) were associated with changes in MVPA (b-path coefficient). However, only changes in self-efficacy (b=10.49, 95% CI 2.46–24.54) and enjoyment (b=7.30, 95% CI 0.92–21.78) mediated the intervention effect on MVPA (ab-path coefficient). For the accelerometer-measured MVPA model, intervention effects were significant for self-efficacy (b=0.48, p<0.01), cognitive processes (b=0.62, p<0.01), and behavioral processes (b=0.61, p<0.01), yet only self-efficacy was associated with changes in MVPA (b=4.43, p=0.03), and mediated intervention effects on MVPA (b=12.15, 95% CI 11.25–16.34).

Conclusions:

Future MVPA interventions with Latinas should target self-efficacy and enjoyment to maximize efficacy.

Keywords: exercise, self-efficacy, behavioral mechanisms, Hispanics, Internet

National physical activity (PA) guidelines recommend that adults engage in ≥150 minutes per week of moderate to vigorous physical activity (MVPA) to maintain general health (Physical Activity Guidelines for Americans, 2nd edition, 2018; Piercy et al., 2018) and prevent chronic disease, but only a fraction of the population meets this standard (Tucker, Welk, & Beyler, 2011; Villarroel, Blackwell, & Jen, 2019). Only 38.2% of Latinas report meeting physical activity guidelines (vs. 50.9% of non-Latino White women (Schiller, Lucas, Ward, & Peregoy, 2012), making them a particularly vulnerable group that may benefit most from interventions designed to help them increase their PA (Arredondo et al., 2016). Indeed, a growing number of PA interventions have been designed and tailored specifically for Latinas (Marcus et al., 2013). These interventions are grounded in theoretical models of behavior change, which postulate that targeting mediating psychosocial constructs (e.g., social support, self-efficacy) can in turn lead to increases in PA (Bauman, Sallis, Dzewaltowski, & Owen, 2002; S. Linke, Robinson, & Pekmezi, 2014). Web-based PA interventions have particularly been growing due to their convenience among Latinas, who have near universal access to web-based devices (Perrin & Duggan, 2015) and their ability to target theoretical constructs through multiple media channels (e.g. print, music, video). Online interventions also have the benefit of broad reach at relatively low cost (B. Larsen, Marcus, Pekmezi, Hartman, & Gilmer, 2017).

Two theoretical models with strong evidence that has formed the basis for numerous PA behavior change interventions as well as explanatory models for differences in PA behavior are the Social Cognitive Theory (SCT) (Young, Plotnikoff, Collins, Callister, & Morgan, 2014) and the Transtheoretical Model (TTM)(Prochaska, Redding, & Evers, 2008). The SCT posits that behavior change is a result of a combination of personal/cognitive (e.g., knowledge, attitudes, beliefs), behavioral (e.g., rewards, reminders, social support), and environmental (e.g., physical surroundings, community culture) factors that interact with and influence each other through a process called reciprocal determinism (Bandura, 2004). The TTM, also known as the Stages of Change Model, is an integrative, biopsychosocial model used to conceptualize the processes of behavior change (Prochaska, DiClemente, & Norcross, 1992) It incorporates elements such as decisional balance, self-efficacy, and processes of change to describe how individuals move through the stages of change.

Interventions targeting SCT and TTM constructs such as increasing self-efficacy (e.g., confidence in the ability to increase activity), encouraging participants to engage in types of PA that they enjoy to enhance expectancies, supporting cognitive processes (e.g., expectations and knowledge about the benefits of PA), and behavioral processes (e.g., social support for PA) have demonstrated success in increasing PA levels in a variety of populations (Booth, Owen, Bauman, Clavisi, & Leslie, 2000; Stacey, James, Chapman, Courneya, & Lubans, 2015) (Plotnikoff, Costigan, Karunamuni, & Lubans, 2013). However, few PA interventions focused on Latina populations have specifically targeted psychosocial constructs despite evidence that many of these constructs are important to PA behavior among Latinos (Marquez & McAuley, 2006). Furthermore, few PA intervention studies designed to target psychosocial constructs embedded in behavior change theories actually conduct mediation models to evaluate whether or not they successfully accomplished their intended goals (Rhodes & Pfaeffli, 2010). Finally, while past PA interventions have shown changes in SCT and TTM constructs that mediated intervention effects, it is unclear if an online format could effectively change these psychosocial constructs, or whether more direct participant interaction is necessary.

The purpose of this study was to examine the potential mediating effects of psychosocial constructs targeted in a PA intervention grounded in the SCT and TTM and tailored specifically for an insufficiently active but otherwise healthy Latina population. Specifically, we examined whether changes in targeted psychosocial variables explained the variation in MVPA outcomes.

Methods

Design

Detailed study procedures have been published previously (Marcus et al., 2015). In short, Pasos Hacia la Salud was a randomized controlled trial of a web-based Spanish language individually tailored physical activity intervention compared to a contact-matched web-based wellness control condition. Main outcomes (for which the study was powered) were changes in MVPA from baseline to six months measured by both the 7-Day Physical Activity Recall Interview (7-Day PAR) and ActiGraph GT3X+ accelerometers. The intervention was based on Social Cognitive Theory (SCT) and The Transtheoretical Model (Prochaska et al., 2008), and specifically targeted theorized psychosocial mediators of behavior change, including social support, self-efficacy, cognitive processes of change, behavioral processes of change, and enjoyment. These were measured in both the Intervention and Control groups at baseline and at 6-month follow-up. Although additional measurements of the mediators were collected monthly among intervention participants, control group participants only completed these questionnaires at baseline and follow-up.

Participants

Study participants (N=218) were women who self-identified as Hispanic or Latino, aged 18–65, who had regular Internet access (at home, work, or in the community), and who were reported engaging in MVPA in 10 minute bouts for less than 60 minutes per week. We chose this cutoff to target individuals with the most to gain by increasing their physical activity. Exclusion criteria included being unable to read or speak Spanish, current or planned pregnancy, BMI ≥45, or any medical condition which might make unsupervised exercise unsafe. Participants were recruited from newspaper ads, Craigslist.org, and community events.

The study was conducted at the University of California, San Diego, and data were collected between 2011 and 2014. Study protocols were approved by the University of California, San Diego Institutional Review Board, and all participants gave written, informed consent.

Intervention Arms

Tailored Web-Based Physical Activity Intervention (Intervention)

The web-based intervention was centered on theoretical constructs from SCT and the TTM and focused on providing participants with tools to incorporate home-based MVPA into their daily lifestyles. The intervention was tailored for Latinas utilizing with formative research with Latinas (focus groups n = 19) to meet the specific PA intervention needs and preferences of Latinas (See Table 2). Surface structure adaptations included modification of language for literacy and terminology, using pictures of Latinas, and adding features to the website such as a music library and exercise videos including Zumba. Deep structure adaptations addressed concerns about neighborhood safety and alternatives to exercising outside, monetary limitations and ways to be active at no cost, and supporting activity within existing gender roles and power dynamics in families. Participants started with a face-to-face coaching session with a bilingual, bicultural staff member trained in goal setting and motivational interviewing. Participants learned to set specific, realistic goals, track activity, anticipate and solve barriers, and identify sources of social support. They were given their own account on the intervention website, which targeted behavioral and cognitive constructs shown to mediate behavior change. Participants filled out monthly questionnaires online which generated individually tailored reports on 1) current stage of readiness for engaging in regular activity, 2) current self-efficacy for activity, 3) their use of cognitive and behavioral strategies for engaging in behavior change, 4) how they compare to people who currently meet national guidelines (normative feedback), and 5) how they compare to their prior responses (progress feedback). They also were given access to a web-based physical activity manual that was tailored to their current stage of readiness for behavior change based on the TTM. This included information about strategies for increasing MVPA, including goal setting, self-monitoring, rewarding oneself, problem-solving, and finding social support.

Table 2:

Baseline Characteristics

Intervention (Mean and SD or %) (N=104) Control (Mean and SD or %) (N=101) Overall (M and SD or %) (N=205)
Demographics
Female 100% 100% 100%
Hispanic 100% 100% 100%
Age 38.8 (10.6) 39.6 (10.4) 39.2 (10.5)
First Generation in U.S. 86.5% 77.0% 81.9%
BMI (kg/m2) 29.1 (5.8) 28.6 (4.5) 28.8 (5.2)
Mexican 82.7% 86.1% 84.4%
Annual Household Income
<$30,000		 69.3% 63.5% 66.4%
Married 50.0% 57.4% 53.7%
Some college or more education 55.4% 66.4% 60.8%
Language Spoken in the Home
 Only Spanish 40.4% 34.7% 37.6%
 More Spanish than English 30.8% 32.7% 31.7%
 Both Equally 15.4% 23.8% 19.5%
 More English than Spanish 11.5% 5.0% 8.3%
 Only English 1.9% 4.0% 2.9%
Health Literacy (scores of 23–36 “functional”) 34.8 (2.7) 37.3 (22.8) 36.0 (16.1)
Baseline MVPA
Self-report MVPA (minutes/week, 7-Day PAR) 8.01(14.95) 10.44(23.98) 9.20(19.91)
Accelerometer measured MVPA in 10 minute bouts (minutes/week) 35.77(69.65) 28.67 (48.22) 32.25(59.96)
Open in a new tab

In addition to the individually tailored elements, the website included calendars for goal setting and recording activity where participants could compare their planned vs. logged activity and their progress over time, a message board to connect with other participants, a tip of the day, and links to online resources such as aerobic exercise videos on YouTube (e.g. Zumba, cardio dance) and walking route maps. Maps were created of 1–3 mile walking routes throughout the city, clustered in neighborhoods where most participants lived. Participants could view regularly changing online tip sheets on topics such as increasing enjoyment (e.g., trying new activities, avoiding boredom) and juggling family responsibilities.

Participants were sent email prompts to visit the website weekly in month 1, biweekly in months 2 and 3, and monthly in months 4–6. Details on intervention engagement and dose have been published elsewhere (S. E. Linke et al., 2019). Briefly, number of logins and time spent on the website significantly predicted greater increases in MVPA for both self-report and accelerometers, and those meeting national physical activity guidelines spent significantly more time on the website than those who didn’t. Participants also indicated a high degree of satisfaction for the intervention, with nearly all (94%) indicating they found the website motivating.

Wellness Control Group (Control)

The control group received access to a Spanish language website that had information on other wellness topics, including diet, stress, and tobacco use. They were sent email prompts to visit the website on the same tapered schedule as the Intervention group, and filled out questionnaires on wellness topics on the same monthly schedule as the Intervention group.

Measures

Basic demographics were assessed via questionnaire at baseline, and included age, education, income, history of residence, race, marital status, and acculturation. Study staff also measured weight and height at baseline to calculate body mass index (BMI).

Outcome Measures

The primary outcome was change in minutes/week of MVPA from baseline to six months. Power for the trial was calculated using change in self-reported activity measured by the 7-Day PAR, so it served as the primary outcome. Change in activity measured objectively by accelerometers served as an additional primary outcome. Both measures have demonstrated reliability and validity in measuring changes in activity over time, and with Spanish speaking populations (Leenders, Sherman, Nagaraja, & Kien, 2001; Melanson & Freedson, 1995; Prince et al., 2008; Rauh, Hovell, Hofstetter, Sallis, & Gleghorn, 1992). In the current study, both at baseline and 6-month follow-up participants walked on a treadmill for 10 minutes at moderate intensity prior to engaging in the PAR interview to support recall of physical activity.

At baseline and follow-up, participants wore an ActiGraph GT3X+ accelerometer for the 7 days immediately prior to their study visit to overlap with the 7-Day PAR assessment time period. Participants were asked to wear the accelerometer on their left hip for seven days for 12+ hours per day, with valid wear time classified as at least 5 days of at least 600 minutes or at least 3000 minutes of valid wear time across at least 4 days, with at least one weekend day included (Matthews, Ainsworth, Thompson, & Bassett, 2002; Troiano et al., 2008). Accelerometer data was processed using ActiLife software with low frequency extension, and used the Freedson cut point of 1952 as a threshold of at least moderate intensity (Freedson, Melanson, & Sirard, 1998). To be consistent with the 7-Day PAR, activity from accelerometers was included only in bouts of ≥10 minutes, with a 2 minute tolerance.

Proposed Mediators

Table 1 shows hypothesized mediators and how each was targeted and measured in the intervention. Psychosocial mediators of behavior change were assessed via online questionnaire in both the intervention and control groups at baseline and six months. All questionnaires have been validated in past studies and have been used previously in Spanish with Latinas (Marcus et al., 2013; Pekmezi et al., 2009). These included the following:

Table 1.

Intervention components targeting hypothesized mediators

Hypothesized mediators Description How intervention targeted the construct
Self-efficacy for exercise (Marcus, Selby, et al., 1992). Confidence in one’s ability to perform activity under various circumstances (e.g. bad weather, on vacation), was assessed via a 5-item measure Increase self-efficacy through completion of small attainable goals, feedback towards goals, problem solving around barriers, and support from their health coach using motivational interviewing skills to support and build confidence. Participants also received each month a tailored report based on answers to this measure from an expert system that draws from a bank of over 300 messages addressing different levels of self-efficacy.
Processes of Change (Marcus, Rossi, et al., 1992) The Processes of Change measure is a 40-item scale with 10 sub-scales measuring various strategies used to increase physical activity. These include 1) behavioral processes (e.g. using social support, rewarding oneself, reminders), and 2) cognitive processes (e.g. learning about benefits of activity, understanding risks, considering how activity affects others). Participants received each month a tailored report based on answers to this measure from an expert system that draws from a bank of over 300 messages addressing different levels of cognitive and behavioral strategies for change. These motivation-matched reports emphasizes behavioral strategies for increasing activity levels, such as goal-setting, self-monitoring, problem-solving barriers, increasing social support, and rewarding yourself for meeting PA goals as well as cognitive processes highlighting and reinforcing the benefits of being activity.
Social support for exercise (Sallis et al., 1987). Social support for exercise was measured via a 13-item measure with 3 sub scales: support from family, support from friends, and negative support. Social support from other participants and staff was supported through the discussion forum on the website. During the initial health coach session, participants wrote a message on the board to practice posting and so other participants can welcome them to the study. Participants were encouraged to post on the message board at least once a week to interact with other participants and to ask questions and communicate with study staff. Monthly tailored reports also included feedback on current social support and tips for increasing support from friends and family.
Enjoyment for activity (Kendzierski & DeCarlo, 1991) Enjoyment for activity was measured via the Physical Activity Enjoyment Scale (PACES), an 18-item measure that assesses satisfaction derived from participating in physical activity. Based on focus group feedback, music libraries and exercise dance videos (e.g., Zumba) were added to the website to support engaging in enjoyable activities. Personalized activity goals were set for each participant to further support engaging in activities that were enjoyable.
 
Open in a new tab
Barrier Self-Efficacy.

Self-efficacy for exercise, or confidence in one’s ability to perform activity under various circumstances (e.g. bad weather, on vacation), was assessed via a 5-item measure develop by Marcus and colleagues. This measure has been used in numerous studies and has shown good reliability and is strongly correlated with stage of change for physical activity (Marcus, Selby, Niaura, & Rossi, 1992).

Processes of Change.

The Processes of Change measure is a 40-item scale with 10 sub-scales measuring various strategies used to increase physical activity. These include 1) behavioral processes (e.g. using social support, rewarding oneself, reminders), and 2) cognitive processes (e.g. learning about benefits of activity, understanding risks, considering how activity affects others). This measure has shown high internal consistency and has been used in numerous physical activity studies (Marcus, Rossi, Selby, Niaura, & Abrams, 1992).

Social Support.

Social support for exercise was measured via a 13-item measure with 3 sub scales: support from family, support from friends, and negative support. Each of the sub scales have shown acceptable internal consistency (Sallis, Grossman, Pinski, Patterson, & Nader, 1987).

Enjoyment.

Enjoyment for activity was measured via the Physical Activity Enjoyment Scale (PACES), an 18-item measure that assesses satisfaction derived from participating in physical activity. It has shown strong internal consistency and criterion validity in past studies (Kendzierski & DeCarlo, 1991).

Analysis

A full description of the study sample has been presented elsewhere (Marcus et al., 2016). Descriptive statistics for baseline demographics, baseline physical activity and psychosocial constructs are presented in Table 2.

For the purpose of this study, mediators of treatment effects on both self-reported and objectively measured MVPA are presented. Hypothesized mediators were continuous measures of the targeted psychosocial constructs (behavioral processes, cognitive processes, self-efficacy, social support and enjoyment). Analyses were conducted to determine whether targeted psychosocial constructs were the mechanism through which participants in the Intervention arm significantly increased their MVPA from baseline to 6 months compared to those in the Control Condition. As a preliminary step, mediators were assessed in univariate models. Constructs for which the indirect effect of the intervention trended towards significant (p<.10), were included in a multivariate mediators model.

Next, using the product of coefficients method, we fit a multiple mediation model in order to simultaneously test the effects of the potential mediators. An important advantage of using a multiple mediation model is that it allows researchers to simultaneously test the effect of a set of mediators, while controlling for the effects of the others (VanderWeele & Vansteelandt, 2014). In addition, a multiple mediation model allows for a comparison of the effects of the mediators to determine the relative influence of each mediator. This approach mimics theory in that it postulates that the mediators work together to influence outcome and do not work in isolation (as is suggested by single mediation models). Finally, a multiple mediation model can be thought to account for the potential collinearity among mediators. Corresponding standard errors were calculated using bootstrapping (10,000 bootstrapped samples).

Mediation models yield path coefficients; the a path is the effect of intervention vs control on the mediator(s) at 6m controlling for baseline. The b path is the effect of the mediator on MVPA outcome controlling for baseline values of both the mediator(s) and MVPA. Finally, the ab path is represents the indirect effect of the intervention (versus control) on the outcome (MVPA) through the relevant mediator, controlling for baseline values, and the ab-path coefficient is the product of coefficients from the aforementioned a- and b-paths. Following the guidelines presented in Preacher and Hayes (MacKinnon, Fairchild, & Fritz, 2007; Preacher & Hayes, 2008; Preacher & Kelley, 2011), a construct was considered a mediator of the intervention effect if the indirect effect of treatment through that construct was significantly different than zero. The study was not originally powered for mediation. Post-hoc power calculations were completed and suggest that given the sample size and mediated effect as presented in the current study, we had 74% power to detect effects of this size or greater.

Analyses were conducted on the intent to treat sample, with all randomized participants included in the analysis. A likelihood based approach to estimation was used which provides consistent estimates of path coefficients without directly imputing missing values (for missing outcomes or mediators or both). All analyses were carried out in SAS 9.3 and significance value set at .05 a priori.

Results

Participants were 205 Latinas, with the majority being first generation in the U.S. (81%), overweight or obese (76.5%), and identifying as having a Mexican background (84%), with a small number identifying as Colombian (3.5%), Guatemalan (1%), and Puerto Rican (1%), or mixed backgrounds (10.5%). At baseline, participants reported low levels of self-efficacy for physical activity, cognitive and behavioral processes of change, enjoyment, and social support from family and friends (See Table 2). The overall effects of the intervention on physical activity have been published (Marcus et al., 2016). Briefly, significantly greater increases in minutes/week of MVPA were seen in the intervention group for both the PAR and accelerometers (see Table 3). Retention rates at 6 months were 88.2% with no significant between-group differences.

Table 3.

Psychosocial mediators at baseline and 6 months by study condition

Baseline 6 Months
Intervention Mean (SD) Control Mean (SD) Intervention Mean (SD) Control Mean (SD)
Psychosocial Mediators
Self-Efficacy (range 1–5) 2.27 (0.75) 2.40 (0.82) 2.71 (0.96) 2.24 (0.75)
Behavioral Processes (range 1–5) 1.98 (0.64) 2.00 (0.58) 2.86 (0.90) 2.26 (0.76)
Cognitive Processes (range 1–5) 2.41 (0.85) 2.49 (0.79) 3.00 (0.88) 2.58 (0.80)
Social Support
 Family (range 10–50)
 Friends (range 10–50)
17.59 (7.43)
15.17 (7.30)
17.96 (7.81)
14.67 (5.59)
21.46 (9.92)
17.05 (7.64)
19.78 (8.86)
15.72 (7.15)
Enjoyment (range 18–126) 86.51 (21.69) 87.83 (18.75) 100.61 (19.45) 94.58 (21.79)
Physical Activity
Self-reported MVPA (minutes per week) 8.0 (15.0) 10.44 (23.98) 112.8 (97.1) 63.5 (88.7)
Accelerometer-measured MVPA (minutes per week) 35.77 (69.65) 28.67 (48.22) 75.8(91.00) 43.0 (60.90)
Open in a new tab

Psychosocial & Theoretical Mediators

Descriptive statistics (means and standard deviations) of physical activity and psychosocial constructs at baseline and 6-month follow-up are shown in Table 3. For all measures, higher scores corresponded to higher values of the measured construct. Self-reported and accelerometer-measured MVPA increased significantly more in the intervention group than controls, as did self-efficacy, behavioral processes, cognitive processes, and enjoyment (for all p< 0.04) (Marcus et al., 2016).

Univariate mediation models suggested significant or trending indirect effects for self-efficacy, behavioral processes, cognitive processes, and enjoyment (p’s<.10). As such, all four constructs were included in the multiple mediation model.

Mediation path coefficients are shown in Table 4 (self-reported MVPA) and Table 5 (accelerometer-measured MVPA). For the multiple mediation model including self-reported MVPA as the outcome, there were significant a path coefficients such that Intervention participants had higher mean self-efficacy (b = 0.43, SE= 0.15, p=.004), behavioral processes (b= 0.64, SE = 0.14, p<.001), cognitive processes (b = 0.54, SE= 0.13, p<.001) and enjoyment scores (b= 9.91, SE= 3.57, p=0.01) at 6 months controlling for baseline compared to Control. For the multiple mediation model including accelerometer-measured MVPA as the outcome, results were similar, with significant a path coefficients for self-efficacy (b = 0.48, p=.003), behavioral processes (b= 0.61, p<.001), and cognitive processes (b = 0.62, p<.001).

Table 4.

Self-reported MVPA multiple mediator models

Psychosocial Mediator a path b path ab path
Self-Efficacy 0.43(0.15),p=.004 24.54(10.98),p=0.03 10.49(2.46–24.54)
Behavioral Processes 0.64(0.14), p<.001 16.36(15.90), p=0.31 10.50(−1.82–25.06)
Cognitive Processes 0.54(0.13), p<.001 17.79(15.19), p=.24 9.67(−4.69–32.94)
Social Support 1.53(1.55),p=.32
Family
0.87(1.23), p=.48
Friends		 −0.67(1.04), p=.52
Family
2.36(1.23),p=.04
Friends		 −1.02(−12.35–1.66)
Family
2.05(−2.11–15.85)
Friends
Enjoyment 9.91(3.57), p=0.01 0.74(0.42), p=0.08 7.30(0.92–21.78)
Open in a new tab

Columns represent coefficients (standard errors) and p-values for a-paths and b-paths; Final column includes coefficients and 95% confidence intervals for ab-paths

Table 5.

Accelerometer-measured MVPA multiple mediator models

Psychosocial Mediator a path b path ab path
Self-Efficacy 0.48(0.16),p=.003 4.43(2.6),p=0.03 12.15(11.25–16.34)
Behavioral Processes 0.61(0.15), p<.001 15.64(16.36), p=0.34 9.61 (−7.85–37.46)
Cognitive Processes 0.62 (0.13), p<.001 8.64 (16.07), p=0.59 5.34 (−13.21–32.54)
Social Support Friends 1.91(1.62), p=0.24
Family
1.88(1.44), p=0.19
Friends		 −0.18(1.04), p=0.86
Family
0.91(1.10), p=0.41
Friends		 −0.35(−9.44–2.93)
Family
1.72(−1.51–17.11)
Friends
Enjoyment 2.98(3.73), p=0.43 0.74(0.47), p=0.12 2.20(−1.70–13.53)
Open in a new tab

Columns represent coefficients (standard errors) and p-values for a-paths and b-paths; Final column includes coefficients and 95% confidence intervals for ab-paths

There were significant b path coefficients such that higher self-efficacy (b= 24.54, SE= 10.98, p=0.03), social support from friends (b=2.36, SE= 1.23, p=.04) and enjoyment scores (b=0.74, SE=0.42, p=0.08) at 6 months were associated with greater minutes of self-reported MVPA at 6 months controlling for baseline. Higher self-efficacy at 6 months was significantly associated with higher accelerometer measured MVPA at 6 months controlling for baseline (b=4.43, SE=2.6, p=0.03).

Finally, results suggested self-efficacy mediated the intervention effects on MVPA for self-reported (b=10.49, 95%CI =2.46, 24.54) and objectively (b=12.15, 95% CI 11.25–16.34) measured MVPA. Enjoyment of physical activity also mediated the intervention effects on self-reported MVPA (b= 7.30, 95%CI = 0.92, 21.78). See Figure 1 for mediation models.

Figure 1.

Figure 1.

Open in a new tab

Mediation models for self-efficacy and self-reported MVPA (A), enjoyment and self-reported MVPA (B), and self-efficacy and accelerometer-measured MVPA (C)

Discussion

The Pasos Hacia la Salud intervention strategies were grounded in strong behavioral science theory (SCT, TTM) and produced significantly greater increases in physical activity in adult Latina women (age 18–65) relative to a control condition (Marcus et al., 2016). Findings from the current study indicated that the intervention also produced significantly greater improvements in the targeted SCT constructs from baseline to six months than the control arm and these SCT improvements were associated with increased physical activity. Moreover, physical activity self-efficacy and enjoyment were shown to be mediators of intervention efficacy in this sample.

SCT improvements have also been reported in past studies. For example, similar increases in self-efficacy and cognitive and behavioral processes of change were found in a recent print-based physical activity intervention study in Latinas (Marcus et al., 2013), along with enhanced family and friend support (unlike the current study). The web-based intervention delivery in the current study may have felt less personal to Latinas than mail/print approaches and thus had less influence upon perceived social support. Future directions could include incorporating social media components into the website to further target this construct in this population.

Not all changes in SCT variables were associated with changes in physical activity in the current study. More specifically, higher physical activity self-efficacy, social support from friends and enjoyment were associated with increased physical activity at 6 months, but cognitive and behavioral processes were not. Several past studies with mostly non-Hispanic white participants, including two theory tailored physical activity distance-delivery interventions (Lewis, Williams, Martinson, Dunsiger, & Marcus, 2013; Papandonatos et al., 2012) and a physician counseling program (Baruth, Wilcox, Dunn, et al., 2010), cited behavioral processes of change as a key mediator. If our findings are replicated, some intervention targets (e.g., behavioral processes) may not be as critical to advancing physical activity behavior change in Latinas and thus could be less of a treatment focus, which has implications for developing more streamlined, potent approaches for this population.

Self-efficacy and enjoyment emerged as effective mediators in the current study which corroborated findings from some past studies (and not others). Self-efficacy was also identified as an important psychosocial mediator of physical activity behavior in an Internet-based physical activity intervention study for women with a family history of breast cancer (Marinac et al., 2018) and some past reviews in this area (Lewis, Marcus, Pate, & Dunn, 2002; Teixeira et al., 2015), but not in other reviews (Murray et al., 2018; Olander et al., 2013). Less research has been conducted on the SCT construct of enjoyment (vs. self-efficacy), but it did not meet criteria for mediation in the previously mentioned 12-month motivationally tailored physical activity intervention among mostly white non-Hispanic adults (Papandonatos et al., 2012) or in a church-based intervention for African American adults (Baruth, Wilcox, Blair, et al., 2010). Despite the mixed findings in the literature, results from the current study may signal the relevance of these particular constructs to our target population and, merit further investigation.

There are many possible reasons for the different roles psychosocial constructs played in mediating physical activity changes in the current study compared to other previous studies. Results may have varied due to differences in 1) intervention approaches (face to face vs. phone-, Internet-, print-delivery), 2) assessment of physical activity (self reported vs. objective) and theoretical constructs, 3) type of mediation analyses conducted, and 4) target population/s across studies. Given the cultural/linguistic differences and unique physical activity intervention barriers and preferences reported by Latinas (B. A. Larsen, Pekmezi, Marquez, Benitez, & Marcus, 2013) and other underserved populations, more research will be needed to corroborate findings from the current study and gain consensus on what mediates physical activity change in Latinas.

Strengths of the current study include the large sample of minority women and use of accelerometer data. Most past studies in this area have relied on self-report physical activity outcomes and focused on non-Hispanic participants. The current findings shed light on what influences physical activity in an underserved, at risk population and will prove valuable in terms of informing future intervention efforts to address related health disparities. Moreover including objective data (vs. only self report) and controlling for baseline values in our analyses gives us more confidence in our conclusions. Finally, this study was also novel in that it investigated an online intervention. The reduced direct participant contact that accompanies this format increases its potential for broad dissemination; importantly, this investigation showed that the online format appears no less effective in influencing psychosocial mediators of PA change.

As for limitations, there are issues in the temporal precedence of variables in the causal chain given the chronological nature of the constructs. More specifically, the mediation analyses were conducted using psychosocial data from baseline and six-month time points (same as the physical activity outcomes), rather than an interim time point, as recommended. Also, SCT and TTM cover a wide range of constructs, some of which (e.g. outcome expectancies, intentions) were not included in these analyses. To be included, constructs needed to both be targeted by the intervention, and have a valid, reliable measure in Spanish given to participants at multiple time points. Future studies should investigate the potential role of additional SCT and TTM constructs in mediating changes in MVPA in this population using the web and other media channels.

Conclusion

Findings from the current study help highlight theoretical mechanisms of action through which the intervention worked. Such results provide insight into physical activity behavior change in minority women and have important intervention implications. Future researchers and practitioners are encouraged to pay careful attention to self-efficacy and enjoyment and target these constructs when promoting physical activity among Latinas.

Supplementary Material

Supplemental Material

Acknowledgments

This research was funded by grant R01CA159954 from the National Institutes of Health/National Cancer Institute

Footnotes

This study was registered at clinicaltrials.gov #NCT01834287

The authors have no conflicts to disclose

Trial Registration: NCT01834287

References

  1. Arredondo EM, Sotres-Alvarez D, Stoutenberg M, Davis SM, Crespo NC, Carnethon MR, … Evenson KR (2016). Physical Activity Levels in U.S. Latino/Hispanic Adults: Results From the Hispanic Community Health Study/Study of Latinos. Am J Prev Med, 50(4), 500–508. doi: 10.1016/j.amepre.2015.08.029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bandura A. (2004). Health promotion by social cognitive means. Health Educ Behav, 31(2), 143–164. doi: 10.1177/1090198104263660 [DOI] [PubMed] [Google Scholar]
  3. Bauman AE, Sallis JF, Dzewaltowski DA, & Owen N. (2002). Toward a better understanding of the influences on physical activity: the role of determinants, correlates, causal variables, mediators, moderators, and confounders. American Journal of Preventive Medicine, 23(2 Suppl), 5–14. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12133733 [DOI] [PubMed] [Google Scholar]
  4. Booth M, Owen N, Bauman A, Clavisi O, & Leslie E. (2000). Social-cognitive and perceived environmental influences associated with physical activity in older Australians. Preventive Medicine, 31, 15–22. [DOI] [PubMed] [Google Scholar]
  5. Freedson PS, Melanson E, & Sirard J. (1998). Calibration of the Computer Science and Applications, Inc. accelerometer. Medicine and Science in Sports and Exercise, 30(5), 777–781. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9588623 [DOI] [PubMed] [Google Scholar]
  6. Kendzierski D, & DeCarlo KJ (1991). Physical activity enjoyment scale: two validation studies Journal of Sport & Exercise Psychology, 13(1), 50–64. doi: 10.1123/jsep.13.1.50 [DOI] [Google Scholar]
  7. Larsen B, Marcus B, Pekmezi D, Hartman S, & Gilmer T. (2017). A Web-Based Physical Activity Intervention for Spanish-Speaking Latinas: A Costs and Cost-Effectiveness Analysis. J Med Internet Res, 19(2), e43. doi: 10.2196/jmir.6257 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Larsen BA, Pekmezi D, Marquez B, Benitez TJ, & Marcus BH (2013). Physical activity in Latinas: social and environmental influences. Womens Health (Lond Engl), 9(2), 201–210. doi: 10.2217/whe.13.9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Leenders NY, Sherman WM, Nagaraja HN, & Kien CL (2001). Evaluation of methods to assess physical activity in free-living conditions. Medicine and Science in Sports and Exercise, 33(7), 1233–1240. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11445774 [DOI] [PubMed] [Google Scholar]
  10. Lewis BA, Marcus BH, Pate RR, & Dunn AL (2002). Psychosocial mediators of physical activity behavior among adults and children. Am J Prev Med, 23(2 Suppl), 26–35. doi: 10.1016/s0749-3797(02)00471-3 [DOI] [PubMed] [Google Scholar]
  11. Linke S, Robinson C, & Pekmezi D. (2014). Applying psychological theories to promote healthy lifestyles American Journal of Lifestyle Medicine, 8(1), 4–14. [Google Scholar]
  12. Linke SE, Dunsiger SI, Gans KM, Hartman SJ, Pekmezi D, Larsen BA, … Marcus BH (2019). Association Between Physical Activity Intervention Website Use and Physical Activity Levels Among Spanish-Speaking Latinas: Randomized Controlled Trial. J Med Internet Res, 21(7), e13063. doi: 10.2196/13063 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. MacKinnon DP, Fairchild AJ, & Fritz MS (2007). Mediation analysis. Annu Rev Psychol, 58, 593–614. doi: 10.1146/annurev.psych.58.110405.085542 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Marcus BH, Dunsiger SI, Pekmezi DW, Larsen BA, Bock BC, Gans KM, … Tilkemeier P. (2013). The seamos saludables study: a randomized controlled physical activity trial of latinas. American Journal of Preventive Medicine, 45(5), 598–605. doi: 10.1016/j.amepre.2013.07.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Marcus BH, Hartman SJ, Larsen BA, Pekmezi D, Dunsiger SI, Linke S, … Rojas C. (2016). Pasos Hacia La Salud: a randomized controlled trial of an internet-delivered physical activity intervention for Latinas. Int J Behav Nutr Phys Act, 13(1), 62. doi: 10.1186/s12966-016-0385-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Marcus BH, Hartman SJ, Pekmezi D, Dunsiger SI, Linke S, Marquez B, … Rojas C. (2015). Using interactive Internet technology to promote physical activity in Latinas: Rationale, design, and baseline findings of Pasos Hacia La Salud. Contemp Clin Trials. doi: 10.1016/j.cct.2015.08.004 [DOI] [PMC free article] [PubMed]
  17. Marcus BH, Rossi JS, Selby VC, Niaura RS, & Abrams DB (1992). The stages and processes of exercise adoption and maintenance in a worksite sample. Health Psychol, 11(6), 386–395. doi: 10.1037//0278-6133.11.6.386 [DOI] [PubMed] [Google Scholar]
  18. Marcus BH, Selby VC, Niaura RS, & Rossi JS (1992). Self-efficacy and the stages of exercise behavior change. Res Q Exerc Sport, 63(1), 60–66. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/1574662 [DOI] [PubMed] [Google Scholar]
  19. Marquez DX, & McAuley E. (2006). Social cognitive correlates of leisure time physical activity among Latinos. Journal of Behavioral Medicine, 29(3), 281–289. [DOI] [PubMed] [Google Scholar]
  20. Matthews CE, Ainsworth BE, Thompson RW, & Bassett DR Jr. (2002). Sources of variance in daily physical activity levels as measured by an accelerometer. Med Sci Sports Exerc, 34(8), 1376–1381. doi: 10.1097/00005768-200208000-00021 [DOI] [PubMed] [Google Scholar]
  21. Melanson EL Jr., & Freedson PS (1995). Validity of the Computer Science and Applications, Inc. (CSA) activity monitor. Medicine and Science in Sports and Exercise, 27(6), 934–940. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/7658958 [PubMed] [Google Scholar]
  22. Murray JM, Brennan SF, French DP, Patterson CC, Kee F, & Hunter RF (2018). Mediators of Behavior Change Maintenance in Physical Activity Interventions for Young and Middle-Aged Adults: A Systematic Review. Ann Behav Med, 52(6), 513–529. doi: 10.1093/abm/kay012 [DOI] [PubMed] [Google Scholar]
  23. Olander EK, Fletcher H, Williams S, Atkinson L, Turner A, & French DP (2013). What are the most effective techniques in changing obese individuals’ physical activity self-efficacy and behaviour: a systematic review and meta-analysis. Int J Behav Nutr Phys Act, 10, 29. doi: 10.1186/1479-5868-10-29 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Pekmezi DW, Neighbors CJ, Lee CS, Gans KM, Bock BC, Morrow KM, … Marcus BH (2009). A culturally adapted physical activity intervention for Latinas: a randomized controlled trial. Am J Prev Med, 37(6), 495–500. doi: 10.1016/j.amepre.2009.08.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Perrin A, & Duggan M. (2015). Americans’ Internet Access: 2000–2015. Retrieved from Physical Activity Guidelines for Americans, 2nd edition. (2018). Washington, DC: Department of Health and Human Sercies [Google Scholar]
  26. Piercy KL, Troiano RP, Ballard RM, Carlson SA, Fulton JE, Galuska DA, … Olson RD (2018). The Physical Activity Guidelines for Americans. JAMA, 320(19), 2020–2028. doi: 10.1001/jama.2018.14854 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Plotnikoff RC, Costigan SA, Karunamuni N, & Lubans DR (2013). Social cognitive theories used to explain physical activity behavior in adolescents: a systematic review and meta-analysis. Prev Med, 56(5), 245–253. doi: 10.1016/j.ypmed.2013.01.013 [DOI] [PubMed] [Google Scholar]
  28. Preacher KJ, & Hayes AF (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods, 40(3), 879–891. doi: 10.3758/brm.40.3.879 [DOI] [PubMed] [Google Scholar]
  29. Preacher KJ, & Kelley K. (2011). Effect size measures for mediation models: quantitative strategies for communicating indirect effects. Psychol Methods, 16(2), 93–115. doi: 10.1037/a0022658 [DOI] [PubMed] [Google Scholar]
  30. Prince SA, Adamo KB, Hamel ME, Hardt J, Connor Gorber S, & Tremblay M. (2008). A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review. Int J Behav Nutr Phys Act, 5, 56. doi: 10.1186/1479-5868-5-56 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Prochaska JO, DiClemente CC, & Norcross JC (1992). In search of how people change: applications to addictive behaviors. Am Psychol, 47, 1102–1114. [DOI] [PubMed] [Google Scholar]
  32. Prochaska JO, Redding CA, & Evers KE (2008). The Transtheoretical Model and Stages of Change. In Glanz K, Rimer BK, & Viswanath K (Eds.), Health Behavior and Health Education (4th ed., pp. 97–122). San Francisco: Jossey-Bass. [Google Scholar]
  33. Rauh MJ, Hovell MF, Hofstetter CR, Sallis JF, & Gleghorn A. (1992). Reliability and validity of self-reported physical activity in Latinos. International Journal of Epidemiology, 21(5), 966–971. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/1468861 [DOI] [PubMed] [Google Scholar]
  34. Rhodes RE, & Pfaeffli LA (2010). Mediators of physical activity behaviour change among adult non-clinical populations: a review update. Int J Behav Nutr Phys Act, 7, 37. doi: 10.1186/1479-5868-7-37 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Sallis J, Grossman R, Pinski R, Patterson T, & Nader P. (1987). The development of scales to measure social support for diet and exercise behaviors. Preventive Medicine, 16(6), 825–836. [DOI] [PubMed] [Google Scholar]
  36. Schiller JS, Lucas JW, Ward BW, & Peregoy JA (2012). Summary health statistics for U.S. adults: National Health Interview Survey, 2010. Vital and Health Statistics. Series 10: Data from the National Health Survey(252), 1–207. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/22834228 [PubMed]
  37. Stacey FG, James EL, Chapman K, Courneya KS, & Lubans DR (2015). A systematic review and meta-analysis of social cognitive theory-based physical activity and/or nutrition behavior change interventions for cancer survivors. J Cancer Surviv, 9(2), 305–338. doi: 10.1007/s11764-014-0413-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Teixeira PJ, Carraca EV, Marques MM, Rutter H, Oppert JM, De Bourdeaudhuij I, … Brug J. (2015). Successful behavior change in obesity interventions in adults: a systematic review of self-regulation mediators. BMC Med, 13, 84. doi: 10.1186/s12916-015-0323-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, & McDowell M. (2008). Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc, 40(1), 181–188. doi: 10.1249/mss.0b013e31815a51b3 [DOI] [PubMed] [Google Scholar]
  40. Tucker JM, Welk GJ, & Beyler NK (2011). Physical activity in U.S.: adults compliance with the Physical Activity Guidelines for Americans. Am J Prev Med, 40(4), 454–461. doi: 10.1016/j.amepre.2010.12.016 [DOI] [PubMed] [Google Scholar]
  41. VanderWeele TJ, & Vansteelandt S. (2014). Mediation Analysis with Multiple Mediators. Epidemiol Methods, 2(1), 95–115. doi: 10.1515/em-2012-0010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Villarroel M, Blackwell DL, & Jen A. (2019). Tables of Summary Health Statistics for U.S. Adults: 2018 National Health Interview Survey Retrieved from http://www.cdc.gov/nchs/nhis/SHS/tables.htm
  43. Young MD, Plotnikoff RC, Collins CE, Callister R, & Morgan PJ (2014). Social cognitive theory and physical activity: a systematic review and meta-analysis. Obes Rev, 15(12), 983–995. doi: 10.1111/obr.12225 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
NIHMS1727615-supplement-Supplemental_Material.pdf (107.7KB, pdf)

RESOURCES