Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Oct 7.
Published in final edited form as: Health Psychol. 2016 Apr 7;35(7):723–732. doi: 10.1037/hea0000333

Mechanisms of change in diet and activity in the Make Better Choices 1 trial

Kristin L Schneider 1,*, Michael J Coons 2, H Gene McFadden 2, Christine A Pellegrini 2, Andrew DeMott 2, Juned Siddique 2, Don Hedeker 3, Laura Aylward 4, Bonnie Spring 2
PMCID: PMC5055406  NIHMSID: NIHMS762434  PMID: 27054299

Abstract

Objective

The Make Better Choices 1 trial demonstrated that participants with unhealthy diet and activity behaviors who were randomized to increase fruits/vegetables and decrease sedentary leisure achieved greater diet and activity improvement than those randomized to change other pairs of eating and activity behaviors. Participants randomized to decrease saturated fat and increase physical activity achieved the least diet-activity improvement. This study examined which psychological mechanisms mediated the effects of the study treatments on healthy behavior change.

Methods

Participants (n=204) were randomized to one of four treatments: increase fruits/vegetables and physical activity; decrease saturated fat and sedentary leisure; decrease saturated fat and increase physical activity; increase fruits/vegetables and decrease sedentary leisure. Treatments provided three weeks of remote coaching supported by mobile decision support technology and financial incentives. Mediational analyses were performed to examine whether changes in positive and negative affect, and self-efficacy, stages of readiness to change, liking, craving and attentional bias for fruit/vegetable intake, saturated fat intake, physical activity and sedentary leisure, explained the impact of the treatments on diet-activity improvement.

Results

Greater diet-activity improvement in those randomized to increase fruits/vegetables and decrease sedentary leisure was mediated by increased self-efficacy (indirect effect estimate=0.04; 95% bias corrected CI: 0.003, 0.11). All treatments improved craving, stage of change and positive affect.

Conclusion

Accomplishing healthy lifestyle changes for 3 weeks improves positive affect, increases cravings for healthy foods and activities and enhances readiness to make healthy behavior changes. Maximal diet and activity improvement occurs when interventions enhance self-efficacy to make multiple healthy behavior changes.

Keywords: self-efficacy, multiple behavior change, fruit/vegetable intake, mediation

Introduction

A majority of life-threatening diseases are chronic and linked to health risk behaviors such as poor quality diet and physical inactivity (Centers for Disease Control and Prevention, 2011; Sondik, Huang, Klein, & Satcher, 2010; Spring, King, Pagoto, Van Horn, & Fisher, 2015). Moreover, 50% of Americans engage in more than one risk behavior, further heightening their risk for developing chronic disease (Coups, Gaba, & Orleans, 2004; Fine, Philogene, Gramling, Coups, & Sinha, 2004). Together, poor quality diet and physical inactivity are the most prevalent, preventable causes of death in the United States (McGinnis & Foege, 2004; Mokdad, Marks, Stroup, & Gerberding, 2004).

Selecting a few behaviors that patients can change may be an efficient way to facilitate multiple behavior change, particularly when time for counseling patients is limited. Make Better Choices 1 (MBC1) was an experimental clinical trial designed to test the optimal manner to maximize simultaneous healthy changes in diet and activity (Spring et al., 2012). Adult participants who consumed a poor quality diet and were physically inactive were randomly assigned to one of four treatments in which they were asked to change one diet and one activity behavior over three weeks: 1) Increase fruits/vegetables and decrease sedentary leisure screen time, 2) decrease saturated fat and decrease sedentary leisure, 3) decrease saturated fat and increase physical activity, and 4) increase fruits/vegetables and increase physical activity. The primary outcome was change from baseline to post-intervention in a composite diet-activity improvement score.

The results of the MBC1 trial showed that participants who were asked to increase fruits/vegetables and decrease sedentary leisure achieved the greatest composite diet-activity improvement, compared to all other treatments. In contrast, participants who were asked to decrease saturated fat and increase physical activity achieved significantly less composite diet-activity improvement compared to all other treatments (Spring et al., 2012). Diet and activity risk behaviors co-occur, suggesting that shared underlying mechanisms may give rise to risk behavior bundling (Coups et al., 2004). If the mechanisms that underpin risk behavior co-occurrence can be understood, it may become possible to produce change in multiple risk behaviors synergistically by targeting common underlying mechanisms.

Treatment design and study hypotheses for MBC1 were based on social cognitive (Bandura, Adams, & Beyer, 1977), behavioral economic (Bickel, Madden, & Petry, 1998) and self-control (Muraven, Tice, & Baumeister, 1998) theories. Hence, a number of psychological mechanisms were posited as mediators of healthy diet and activity change. Self-efficacy, a core construct in social cognitive theory and a well-established predictor and mediator of physical activity and nutrition behaviors (Anderson, Winett, Wojcik, & Williams, 2010; Anderson, Winett, Wojcik, Winett, & Bowden, 2001; Lewis, Marcus, Pate, & Dunn, 2002), was targeted by the treatments. Each treatment aimed to increase self-efficacy for the targeted behaviors via verbal persuasion and mastery experiences (Bandura et al., 1977). Regular coaching calls and incrementally progressive goals were implemented to foster success experiences that built behavioral skills and increased confidence about being able to perform the targeted behaviors. A handheld mobile device enabled individuals to self-monitor diet and activity behaviors relative to goals, receive performance feedback, and support goal attainment. Since prior successful experiences increase self-efficacy, each self-monitored attainment of a health behavior goal was expected to increase self-efficacy for that behavior (Parschau et al., 2013; Parschau et al., 2014). We examined whether a greater increase in self-efficacy could explain the greater composite improvement in diet and activity behaviors observed for the treatment that increased fruits/vegetables and decreased sedentary leisure, relative to other treatment conditions. Conversely, we examined whether a lesser improvement in self-efficacy could explain why participants targeting decreased saturated fat and increased physical activity achieved less composite diet and activity improvement, relative to the other treatments.

Behavioral choice theory, grounded in behavioral economic principles, was the basis for postulating why the increase fruits/vegetables and decrease sedentary leisure treatment would demonstrate the most diet-activity improvement, compared to the other treatments (Bickel et al., 1998). The theory posits that greater healthy behavior change can arise because a healthy behavior substitutes for (crowds out) an unhealthy behavior, because a decrease in one unhealthy behavior is complemented by a (tag-along) decrease in another, or because an increase in one healthy behavior is accompanied by a same-directional change in another healthy behavior.

Self-control theory (Muraven Tice, & Baumeister, 1998) posits that inhibiting responses to overlearned reward self-administration behaviors feels depriving and lessens the self-regulatory strength remaining to continue to resist temptations. Hence, decreasing fat intake and sedentary leisure were expected to result in increased negative affect, decreased positive affect, as well as increased craving, liking and attention bias for unhealthy choices.

The present study extended the findings of the MBC1 trial by examining potential mediators of treatment effects. We hypothesized that changes in self-efficacy, craving, liking, attentional bias and positive and negative affect would mediate the relationship between the treatments and change in the composite diet - activity improvement score. We also explored whether progression or regression in stage of change mediated the relationship between the treatments and composite diet - activity improvement, since these variables have affected diet and activity behavior change in other studies (e.g., Hilderbrand & Betts, 2009).

Methods

Study Design

In the MBC1 trial, participants were randomly assigned to one of four treatments that focused on changing one diet and one activity behavior: 1) increase fruits/vegetables and decrease sedentary leisure, 2) decrease saturated fat and decrease sedentary leisure, 3) decrease saturated fat and increase physical activity, or 4) increase fruits/vegetables and increase physical activity. Participants received a mobile device (personal data assistant) for self-monitoring, receiving feedback and providing in-the-moment decision support. Participants also received remote behavioral coaching and financial incentives for behavioral goal attainment over a 3 week period. Following the intervention, participants transitioned to a five-month maintenance period. During maintenance, participants were encouraged to use the mobile device to maintain behavior changes, but were not financially incentivized to do so. More detailed descriptions of the parent study design (Spring et al., 2010) and results are presented elsewhere (Spring et al 2012). The Northwestern University and University of Illinois at Chicago Institutional Review Boards approved the study procedures.

Study Sample

Participants (n=204) aged 21 to 60 years were recruited through community advertisements and directed to a website where they provided initial informed consent and completed screening assessments including the rapid food screener (Block, Gillespie, Rosenbaum, & Jenson, 2000), physical activity recall questionnaire (PAR; Kohl, Blair, Paffenbarger, Macera, & Kronenfeld, 1988) and a questionnaire paralleling the PAR to assess time devoted to targeted sedentary pastimes. To be eligible, participants had to self-report all four health risk behaviors: < 5 daily servings of fruits/vegetables, > 8% of daily calories from saturated fat, < 60 minutes/day of moderate intensity exercise, and > 120 minutes/day of sedentary leisure screen time. After that initial screening, participants were asked to attend an in-person session where they provided written informed consent and received a mobile device and accelerometer for baseline data collection. Table 1 contains demographic and baseline data for the randomized sample, separated by condition.

Table 1.

Participant Demographics

Total ↑FV    ↑PA ↓Fat    ↓Sed ↑FV    ↓Sed ↓Fat    ↑PA Test P Value
(n=204) (n=48) (n=53) (n=56) (n=47) (by treatment)
Age, years
    Mean (SD) 32.8(11.0) 33.4(10.8) 30.8(10.8) 35.0(12.1) 31.9(9.7) F=1.56 0.20
Body Mass Index, kg/m2
    Mean (SD) 28.3 (7.3) 28.6(7.0) 27.0(6.6) 28.3(6.1) 29.4(9.3) F=0.86 0.46
Gender, No. (%)
    Male 48(23.5) 14(29.2) 12(22.6) 14(25.0) 8(17.0)
    Female 156(76.5) 34(70.8) 41(77.4) 42(75.0) 39(83.0) χ2=2.05 0.56
Ethnicity, No. (%)
    White 109(53.4) 22(45.8) 32(60.4) 33(58.9) 22(46.8)
    Black 47(23.0) 15(31.3) 6(11.3) 12(21.4) 14(29.8)
    Hispanic/Latino 18(8.8) 3(6.3) 5(9.4) 6(10.7) 4(8.5)
    Asian 23(11.3) 7(14.6) 7(13.2) 4(7.1) 5(10.6)
    Other or multiple 7(3 .5) 1(2 .1) 3(5 .7) 1(1 .8) 2(4.3) χ2=10.93 0.54
Education, No. (%)
    College degree 151(74.0) 31(64.6) 41(77.4) 44(78.6) 35(74.5)
    No college degree 53(26.0) 17(35.4) 12(22.6) 12(21.4) 12(25.5) χ2=3.14 0.37
Open in a new tab

Note. ↑FV=increase fruits/vegetables; ↑PA=increase physical activity; ↓Fat=decrease saturated fat; ↓Sed=decrease sedentary leisure screen time.

Baseline Phase

Participants completed a two-week baseline-recording phase when they were asked to wear an accelerometer and self-monitor their diet and activities using the mobile device. Participants also attended a laboratory session to complete additional baseline self-report and performance measures. Participants who no longer met the criteria for all 4 risk behaviors based on their baseline data or who did not attend the laboratory session were deemed ineligible.

Randomization and Intervention Phase

After the baseline phase, eligible participants were randomized to one of the four treatments. Depending on the treatment assigned, participants were encouraged to reach the ultimate goals of: 5 daily servings of fruits/vegetables, <8% of calories from saturated fat, ≥60 minutes per day of physical activity, and ≤ 90 minutes per day of sedentary leisure time. For the first week of intervention, participants were asked to close one half of the gap between their baseline level of the two targeted behaviors and their ultimate daily goal. For the rest of the intervention period, they were asked to attain and maintain the ultimate goal levels for the two targeted behaviors assigned. During the second intervention week, participants completed the second laboratory session.

Outcome Measure

The primary outcome components were assessed by daily self-report recordings obtained from the mobile device. Participants were encouraged to record their diet on the mobile device after each meal and snack by selecting foods from drop-down menus in CyberSoft's Nutribase. Saturated fat and fruit/vegetable consumption were measured from dietary intake recordings. Minutes of physical and sedentary activity were measured cumulatively by an end-of-day 24-hour activity log in which participants accounted for every 15-minute block of the day using a modified compendium of physical activities (Ainsworth et al, 2000) that included sedentary pastimes.

Composite Diet - Activity Improvement Score

In the MBC1 main results, the four behaviors were placed on a common scale to quantify overall change and then used to create the “Composite Diet-Activity Improvement Score.” We transformed all variables to better approximate normality, using square root transformation for the count outcomes (fruits/vegetables, physical activity, and sedentary leisure) and arc sine transformation for the percentage outcome, saturated fat (Mosteller & Tukey, 1977). Then we standardized each individual health behavior using a modified z-score (where 1 unit represents a 1-standard deviation change relative to the sample distribution for that behavior at baseline), with higher values representing greater healthy lifestyle improvement. We calculated the mean of all four individual z-scores at baseline and post-treatment, as recommended (Prochaska, Velicer, Nigg, & Prochaska, 2008), to derive a composite index that expressed each participant's overall healthy diet and activity change.

Measures of Candidate Mediating Mechanisms

Assessments of candidate mediators were completed during laboratory sessions at baseline and week 2 of treatment.

Self-Efficacy for Behavior Change

Separate self-efficacy measures for the four behaviors reflected participants’ degree of confidence in performing each behavior. A measure of physical activity self-efficacy (Marcus, Selby, Niaura, & Rossi, 1992) provided the template and was modified to assess self-efficacy for sedentary leisure, saturated fat intake, and fruit and vegetable consumption (e.g., rate how confident you are about eating fruits and vegetables when you are in a rush). Internal consistency for the self-efficacy measures was strong in this sample (Cronbach's α = 0.94 - 0.96). A composite measure of self-efficacy was also calculated using weighted means for each of the self-efficacy measures.

Liking for Foods and Activities

The extent to which participants reported liking foods and activities was assessed using four instruments developed for this study. Each questionnaire was modeled after Rozin and colleagues’ (1991) hedonic scale. Participants completed the Food Liking Questionnaire by indicating their preference for 40 different foods [10 sweet foods high in saturated fat (e.g. cookies), 10 savory foods high in saturated fat (e.g., chips), 10 vegetables (e.g., carrots) and 10 fruits (e.g., bananas)] using a 9-point hedonic scale ranging from 1=dislike extremely to 9= like extremely. On the Activity Liking Questionnaire, participants indicated their preference for 40 different activities [10 moderate-vigorous intensity athletic activities (e.g., running), 10 moderate intensity work or household activities (e.g., climbing stairs), 10 sedentary home-related activities (e.g., surfing the internet), and 10 sedentary recreational activities (e.g., playing videogames)] using the same scale. Internal consistency reliability for the Food and Activity Liking Questionnaires was strong (α = 0.90 and 0.88, respectively). A composite liking measure was calculated by averaging the means of each liking measure, after reverse-scoring sedentary and saturated fat liking to achieve uniform directionality of desirable (healthful) change among the scales.

Craving for Foods and Activities

The four foods and activities that were given the highest liking ratings in each category of the Liking Questionnaires were selected to create the craving measure. For these 16 items, participants were asked to indicate their degree of craving for the foods, or their strength of urge to participate in these activities using a 9-point scale (1=No craving/Urge, 9=Extreme craving/Urge). Internal consistency for the Food and Activity Craving Questionnaires was α = 0.90 and 0.88, respectively. A composite craving measure was calculated by averaging the means of each craving measure, after reverse-scoring sedentary and saturated fat liking to achieve uniform directionality of desirable (healthful) change.

Positive and Negative Affect

The Positive and Negative Affect Schedule (PANAS) contains 10-item subscales for positive affect and negative affect (Watson, Clark, & Tellegen, 1988). The PANAS scales show good construct validity with other measures of affect, with convergent correlations above 0.90 (Watson et al., 1988). In the present study, internal consistency was acceptable (α = 0.95 and 0.92, for positive and negative affect, respectively).

Stages of Readiness for Change

Four instruments assessed stage of readiness to change each of the four behaviors. A fruit/vegetable stage of change measure was modified to create the other measures (e.g., do you intend to decrease your sedentary leisure pastimes in the next 6 months?) (Ling & Horwath, 1999). Participants categorized their readiness to change these behaviors into one of five stages: pre-contemplation, contemplation, preparation, action, or maintenance. Stage of change measures were averaged to create a composite score.

Attention Bias

Attention bias was assessed using a modified version of the STROOP test (Stroop, 1935). Participants completed a series of tasks in which they named the color of the word, while suppressing attention to the semantic meaning of the word. Each task represented one of the following categories: fruits/vegetables, high-saturated fat foods, sedentary leisure pastimes, physical activity, and neutral (furniture) words (e.g., respectively, pear, cake, book, bike and lamp). STROOP words were matched on the number of letters, number of syllables and usage in the English language. The colors of the words was randomly assigned. The STROOP quantifies the degree of cognitive interference (evidenced by response slowing) produced by color-naming the words associated with fruits/vegetables, high-saturated fat foods, sedentary leisure pastimes, and physical activity, as compared to neutral (furniture) words. A composite attention bias measure was calculated by averaging the means of the STROOP interference measures for all food and activity categories.

Analytic Plan

The three planned contrasts described in the primary outcome paper were used to examine the effect of treatment on change in each mediator during the treatment phase (Spring et al., 2012). Two contrasts (the increase fruits/vegetables, decrease sedentary leisure treatment vs. the other three treatments, and the decrease saturated fat, increase physical activity treatment vs. the other three treatments) predicted change in the diet-activity composite score (Spring et al., 2012). The third contrast (the increase fruits/vegetables, increase physical activity treatment vs. the other three treatments) was included in the model to be consistent with the primary outcome paper. Since this contrast did not predict the diet-activity composite improvement score, mediators of this contrast were not examined. Each contrast was coded as 1=targeted treatment and 0=the other 3 treatments.

Using linear mixed models, the effect of each treatment contrast × time interaction on each proposed mediator was examined. The main effect of time from baseline to post-treatment and the treatment contrasts were also included in each model. If neither the increase fruits/vegetables and decrease sedentary leisure treatment contrast by time interaction, nor the decrease saturated fat and increase physical activity treatment contrast by time interaction was significant, all interaction terms were removed from the model to examine whether the main effect of time was significant. For the self-efficacy, liking, craving, stage of change and attention bias measures, the composite measure was first tested using linear mixed models. If significant treatment contrast × time interactions or time main effects emerged, the individual measures for each diet and activity behavior were examined.

For significant treatment contrast by time interactions, mediation analyses were then conducted to examine whether any of the candidate psychological mechanisms mediated the relationship between the treatment contrast and the diet-activity composite improvement score. Prior to testing mediation, change scores (baseline score subtracted from post-treatment score ) were computed for the composite measure of each candidate mechanism (e.g., composite change in self-efficacy) and for the candidate mechanism specific to each diet and activity behavior (e.g., change in self-efficacy for fruits/vegetables). To calculate change scores, participants were required to have data at both time points (i.e., missing post-treatment data was not imputed). The composite measures were used first in these mediational analyses; if significant, the individual measures for each diet and activity outcome were then tested. The Preacher and Hayes (2008) SPSS Macro for Multiple Mediation was used to test for mediation. This analysis examines the effect of the independent variable on the mediator, the effect of the mediator on the outcome variable, the total effect of the independent variable on the outcome variable and the direct effect of the independent variable on the outcome variable. Results are provided for the indirect effect of the independent variable on the outcome variable through the mediator based on 1000 bootstrapping samples. All analyses were conducted using SPSS version 21.

Power

The main trial was powered to detect differences between conditions on the primary outcome variable. This study had > .80 power to detect mediation using the bias-corrected bootstrap procedure when the alpha and beta path estimates were ≥ 0.26 based on Fritz & MacKinnon's (2007) procedure for calculating the sample size required for mediation.

Results

Examining the treatment contrast × time interaction for proposed mediators

The first mixed model examining the composite self-efficacy measure revealed significant treatment contrast × time interactions for the increase fruits/vegetables and decrease sedentary leisure (B=0.20, 95% confidence interval (CI): 0.04, 0.36; t=2.45, p=.02) and the decrease saturated fat and increase physical activity (B=−0.21, 95% CI: −0.38, −0.04; t=−2.39, p=.02) treatment contrasts. During treatment, participants randomized to the increase fruits/vegetables and decrease sedentary leisure treatment reported increased self-efficacy from baseline to post-treatment (MΔ=0.18; SD=0.69), compared to the other treatments (MΔ=−0.06; SD=0.67). Participants randomized to the decrease saturated fat and increase physical activity treatment experienced a significant decrease in self-efficacy from baseline to post-treatment (MΔ=−0.20; SD=0.70), compared to all other treatments (MΔ=0.06; SD=0.67).

Next, we examined the effect of the treatment contrast × time interactions on self-efficacy for the individual behaviors. Participants randomized to the increase fruits/vegetables and decrease sedentary leisure treatment reported increased self-efficacy for fruits/vegetables (B=0.37, 95% CI: 0.16, 0.58; t=3.48, p= .001; MΔ=0.39, SD=0.99) and sedentary leisure (B=0.33, 95% CI: 0.13, 0.52; t=3.27, p=.001; MΔ=0.31, SD=0.84), compared to all other treatments (MΔ=−0.08, SD=0.88 for fruits/vegetables and MΔ=−0.08, SD=0.82 for sedentary leisure; Figure 1). In contrast, participants randomized to the decrease saturated fat and increase physical activity treatment reported decreased self-efficacy for fruits/vegetables (B=−0.37, 95% CI: −0.59, −0.14; t=−3.20, p=.002; MΔ=−0.33, SD=0.91) and sedentary leisure (B=−0.32, 95% CI: −0.53, −0.11; t=−3.04, p=.003; MΔ=−0.30, SD=0.77), compared to all other treatments (MΔ=0.16, SD=0.91 for fruits/vegetables; MΔ=0.12, SD=0.84 for sedentary leisure). Changes in self-efficacy for physical activity and saturated fat did not differ significantly among the treatments.

Figure 1.

Figure 1

Open in a new tab

Change in Each Self-Efficacy Measure by Treatment

Note.. PostTxmt= End of intervention phase. ↑FV=increase fruits/vegetables, ↑PA=Increase physical activity, ↓Fat=Decrease saturated fat, ↓Sed=Decrease sedentary leisure screen time.

** p < .01, *** p < .001

An examination of the composite measure of liking for healthy foods and activities revealed no statistically significant treatment contrast × time interactions and no change in liking over time (all p's > 0.45). For the composite measure assessing craving for healthful foods and activities, the treatment contrast × time interactions were not significant. However, the time main effect was significant, such that craving for healthy behaviors increased from baseline to post-treatment (B=0.19, 95% CI: 0.05, 0.33; t=2.61, p=.01; MΔ=0.19; SD=1.01). Craving for fruits/vegetables (B=0.44, 95% CI: 0.25, 0.64; t=4.43, p<.001; MΔ=0.46; SD=1.39) and physical activity (B=0.38, 95% CI: 0.19, 0.57; t=3.81, p<.001; MΔ=0.37; SD=1.36) increased, whereas craving for saturated fat (p=0.32) or sedentary leisure (p=0.49) did not change.

The treatment contrast × time interactions were not significant for positive (p's>0.36) or negative affect (p's>0.67). There was a significant main effect of time such that positive affect increased from baseline to end of treatment (B=1.38, 95% CI: 0.30, 2.46; t=2.57, p=0.01; MΔ=0.96; SD=5.05); negative affect did not change (p=0.13). The treatment contrast × time interactions were not significant for stage of change (p's >0.07), but readiness to make healthy changes increased significantly from baseline to end of treatment (B=0.46, 95% CI: 0.37, 0.54; t=10.88, p<0.001; MΔ=0.45; SD=0.57). Stage of change increased for readiness to improve saturated fat (B=0.65, 95% CI: 0.48, 0.82; t=7.52, p<0.001; MΔ=0.65; SD=1.17), fruits/vegetables (B=0.65, 95% CI: 0.50, 0.79; t=8.72, p=<0.001; MΔ=0.64; SD=1.03) and physical activity (B=0.69, 95% CI: 0.55, 0.83; t=9.80, p<0.001; MΔ=0.66; SD=0.96). The treatment contrast × time interactions for attentional bias were not significant (p's > 0.40), nor was the main effect of time (p=0.47).

Mediation

Because the increase fruits/vegetables and decrease sedentary leisure treatment contrast and the decrease saturated fat and increase physical activity treatment contrast by time interactions significantly predicted change in composite self-efficacy, we examined whether increased self-efficacy mediated the differential treatment effects. Bootstrapped results suggested that change in the self-efficacy composite significantly mediated the greater improvement in the composite diet-activity score produced by the increase fruits/vegetables and decrease sedentary leisure treatment (indirect effect estimate=0.04; 95% bias corrected CI: 0.003, 0.11; p<.0001; total effect estimate=0.41; Table 2). Participants randomized to the increase fruits/vegetables and decrease sedentary leisure treatment reported increased self-efficacy, which accounted for their greater improvement in the composite diet-activity score (Figure 2).

Table 2.

Summary of results testing the mediation effects of psychological factors on the relationship between the treatments and diet-activity improvement score

Mediator Mediation criteria Effect of mediation (95% bias corrected CI)
Estimate of time X treatment contrast on mediator (95% CI) Estimate of mediator on DV (95% CI) Estimate of time X treatment contrast adjusting for mediator on DV (95% CI)
Treatment contrast: Increase fruits/vegetables and decrease sedentary leisure
Self-efficacy composite .22 (0.44, 0.006) .18 (0.05, 0.30) 0.37 (0.55, 0.18) 0.04 (0.01, 0.11)
Self-efficacy-FV 0.37, (0.16, 0.58) 0.59 (0.28, 0.89) 2.35 (2.99, 1.71) 0.28 (0.11, 0.61)
Self-efficacy-Sed 0.33, (0.13, 0.52) NS NS NA
Self-efficacy-Fat NS NA NA NA
Self-efficacy-PA NS NA NA NA
Treatment contrast: Decrease saturated fat and increase physical activity
Self-efficacy composite −0.21 (−0.38, −0.04) 0.18 (0.06, 0.31) −0.33 (−0.12, −0.53) −0.04 (−0.11, 0.0005)a
Self-efficacy-FV −0.37 (−0.59, −0.14) NA NA NA
Self-efficacy-Sed −0.32 (−0.53, −0.11) NA NA NA
Self-efficacy-Fat NS NA NA NA
Self-efficacy-PA NS NA NA NA
No significant treatment contrast x time interactions
Liking NS NA NA NA
Craving NS NA NA NA
Positive affect NS NA NA NA
Negative affect NS NA NA NA
Stages of change NS NA NA NA
Attentional bias NS NA NA NA
Open in a new tab

Note. N's range from 175-188 due to missing data on mediator variables; NA=not applicable because at least one mediation criterion was not met; NS= non-significant.

a

Mediation for individual behaviors not examined because composite did not fulfill mediation criteria.

Figure 2.

Figure 2

Open in a new tab

Composite Self-Efficacy Mediates the Relationship Between the Increase Fruit/Vegetable and Decrease Sedentary Leisure Treatment Contrast and the Diet-Activity Composite Score.

Note. Change (Δ) was calculated as change from end of treatment – baseline. ↑FV=Increase fruits/vegetables, ↓Sed=Decrease sedentary leisure screen time

**p<0.01, ****p<0.0001.

An examination of the individual component behaviors showed that fruits/vegetables self-efficacy significantly mediated the relationship between the increase fruits/vegetables and decrease sedentary leisure treatment contrast and fruit/vegetable intake (indirect effect estimate=0.28; 95% bias corrected CI: 0.11, 0.61; p <.0001; total effect estimate=2.63). Participants randomized to the increase fruit/vegetable and decrease sedentary leisure treatment reported increased self-efficacy for eating fruits/vegetables, which explained their increased fruit/vegetable intake (Figure 3). Change in self-efficacy for decreasing sedentary leisure activities did not mediate the relationship between the increase fruit/vegetable and decrease sedentary leisure treatment contrast and reduced sedentary leisure time (estimate=0.002; 95% bias corrected CI: −0.005, 0.0002). Self-efficacy did not mediate the relationship between the decrease saturated fat and increase physical activity treatment contrast and that treatment's lesser improvement in the composite diet-activity score (estimate=−0.04; 95% bias corrected CI: −0.11, 0.0005).

Figure 3.

Figure 3

Open in a new tab

Fruit/Vegetable Self-Efficacy Mediates the Relationship between the Increase Fruit/Vegetable and Decrease Sedentary Leisure Treatment Contrast and Fruit/Vegetable Intake Note. Change (Δ) was calculated as change from end of treatment – baseline. ↑FV=Increase fruits/vegetables, ↓Sed=Decrease sedentary leisure screen time.

**p<0.01, ***p<0.001, ****p<0.0001.

Post-hoc analyses

Since self-efficacy, craving, positive affect and stage of change improved over time, we conducted post-hoc analyses examining the impact of change in these variables on improvement in the composite diet-activity score, across all treatments. Separate linear regression models were used to test the effect of change in each of these psychological mechanisms on the composite diet-activity improvement score. Only self-efficacy impacted the composite diet-activity improvement score with increases in self-efficacy predicting improvement in the composite diet-activity score (B=0.21, 95% CI: 0.09, 0.34, R2=0.06).

Discussion

This study examined psychological processes as potential mechanisms to explain the differential effects of four different diet and activity interventions on composite improvement in diet and activity behaviors. Those randomized to increase fruits/vegetables and decrease sedentary leisure showed significant increases in self-efficacy for both behaviors, and their greater overall improvement in self-efficacy accounted for greater composite improvement in healthy diet and activity behaviors, compared to the other treatments. Results support self-efficacy as a mechanism through which a brief, technology-supported intervention that targets increasing fruits/vegetables and decreasing sedentary leisure time, improves multiple healthy diet and activity behaviors by increasing self-efficacy, although the effect was small.

The finding that the increase fruits and vegetables and decrease sedentary leisure treatment surpassed other treatments in the amount of diet and activity improvement it generated supports behavioral choice theory, as does the finding that this treatment produced complementary, correlated reductions in saturated fat intake and sedentary leisure (Spring et al., 2012). Importantly, the benefits of this treatment were mediated by improved self-efficacy for healthy lifestyle behaviors, which is consistent with social-cognitive theory. The increase fruits and vegetables, decrease sedentary leisure treatment significantly increased self-efficacy for both of these targeted behaviors, and the rise in fruits and vegetable intake was mediated by increased self-efficacy. Even though the same treatment also improved sedentary leisure and saturated fat intake, those behavior changes were not mediated by increased self-efficacy, perhaps because those improvements were either too subtle or too automatized to be consciously experienced.

Self-efficacy is one of the most extensively studied predictors of positive health behavior change. Covariation between self-efficacy and health behaviors has been demonstrated in cross-sectional and prospective studies (Higgins et al., 2014; Shaikh, Yaroch, Nebeling, Yeh, & Resnicow, 2008; Teixiera et al., 2015), even though the association is not universally observed (Guillaumieab, Godinac, Manderscheidb, Spitzd, & Muller, 2012; Olander et al., 2013). Importantly, the present study showed that self-efficacy decreased among participants who received the traditional dieting recommendation to decrease saturated fat and increase physical activity, even though the reduction in self-efficacy did not account for the traditional dieting condition's smaller composite diet and activity improvement, compared to the other treatments. It is noteworthy that practicing traditional dieting behaviors did not increase self-efficacy for decreasing fat and increasing physical activity, even though participants were successful in changing these behaviors. Moreover, following traditional dieting recommendations significantly reduced participants’ self-efficacy for being able to increase fruits/vegetables and decrease sedentary leisure. Negative intervention effects on self-efficacy also have been reported in a technology-supported weight loss intervention (Laing et al., 2014), as well as in a physical activity intervention involving obese adolescents (Bergh et al., 2012). The reason why self-efficacy failed to increase despite successful behavior change in the traditional dieting group is unclear. Perhaps prior lack of success at maintaining weight loss highlighted the inherent difficulty of restricting energy intake and increasing physical activity, undermining participants’ confidence about being able to sustain the behaviors. The findings suggest the potential for uniquely disempowering effects to occur when following a regimen that intentionally restricts dietary intake and increases physical activity, as would be done when following a traditional weight loss diet.

Even though we found relatively few mediators of differential treatment effects on diet and activity improvement, several measures of psychological processes showed significant improvement over time. Presumably these improvements reflect the fact that all treatment conditions were designed to improve diet and activity behaviors (i.e., no treatment condition was an inactive control). The observation that all treatments increased cravings for fruits/vegetables and physical activity suggests that different types of healthy lifestyle intervention have the general effect of increasing the desire to consume healthful foods and engage in healthful activities. Even when fruit/vegetables and physical activity behaviors are not specifically targeted by treatment, increased cravings for them occur. Notably, the increase in craving for fruits/vegetables and physical activity was no greater in the treatment conditions that targeted these behaviors, as compared to those that targeted saturated fat and sedentary leisure.

Contrary to the prediction, based on self-control theory, that craving and attentional bias toward unhealthier alternatives might increase when participants were asked to decrease fat and sedentary leisure, neither adverse effect occurred. Instead, increased cravings for healthy foods and activity appeared to track the increasing strength of the habit to engage in healthy lifestyle behaviors. Positive mood also increased over time in all treatment conditions, paralleling the increase in healthy lifestyle behaviors. Hence, the findings suggest that making healthy diet and activity changes of various types may have led participants to feel happier.

Participants’ stage of readiness to improve dietary fat, fruits and vegetables and physical activity increased during the course of treatment; only readiness to decrease sedentary leisure did not change. Improvements in stage of change for three of the health behaviors are encouraging in light of the short intervention duration.

Strengths of the study include the prospective design, diverse sample, examination of multiple co-occurring chronic disease risk behaviors, and breadth of hypothesized psychosocial mediating mechanisms. A limitation is that well-validated, reliable measures did not exist for several of our constructs of interest. However, well-validated measures were adapted to assess those mechanisms and demonstrated acceptable reliability. Only self-report measures were used to assess the outcomes, which are vulnerable to biased reporting. Bogus pipeline procedures were used in the MBC1 trial to reduce that threat and facilitate accurate reporting. Participants were informed that accelerometry and urinary assays were being used to validate their self-reported food intake and activity. Also, notably, the treatment effects observed in the parent trial held when social desirability was included as a covariate (Spring et al., 2012).

A number of individual differences (e.g., personality traits), interpersonal influences (e.g., social support) and environmental factors (e.g., neighborhood walkability), besides the mechanisms we hypothesized, might have mediated or moderated differential treatment effects in the MBC trial. Replication is warranted, especially since the present study was not powered to detect mediation. The use of technology, short length of treatment and the financial incentives provided to participants for treatment adherence may limit generalizability to short, reward-based interventions supported by mobile technology that are becoming widely used in worksite health promotion (e.g., Mattila et al., 2013). Since mediators were not measured during follow-up appointments, this study could not examine if the short-term beneficial effects of the increase fruits and vegetables, decrease sedentary leisure treatment on self-efficacy were maintained and continued to facilitate healthy diet and activity behaviors. Notably, the superior effects of this treatment condition on diet and activity improvement did hold during follow-up in the parent study, suggesting that mediator effects might also have maintained.

This study examined a variety of theoretical mechanisms hypothesized to underlie healthful change in diet and physical activity. Several psychological processes showed positive changes across all of the healthy lifestyle treatments. Observed improvements included increased craving for healthy behaviors, improved mood, and increased readiness to improve diet and activity. In contrast, the treatments produced differential changes in self-efficacy that mediated treatment effects on the magnitude of healthy diet and activity change. The greater impact of the increase fruit/vegetable intake and decrease sedentary leisure time treatment on the composite healthy diet and activity improvement score, as compared to other treatments, was mediated by its more positive impact on self-efficacy.

These findings are important in demonstrating that not all healthy lifestyle changes feel equally empowering. Success in increasing fruits/vegetables intake and decreasing leisure screen time increased participants’ confidence about being able to improve these behaviors. On the other hand, success in decreasing fat intake and increasing physical activity did not increase confidence about being able to change those behaviors, even though participants did in fact improve them. Moreover, success at traditional dieting behaviors undermined self-confidence about being able to improve sedentary leisure screen time and fruits and vegetables intake. Even given comparably successful initial attainment of diet and activity goals, behavior changes that maximize self-efficacy appear more likely to be maintained (Spring et al., 2012). Because health behaviors are often interrelated, changing several behaviors simultaneously is particularly efficient (Evers & Quintiliani, 2013; Spring et al., 2015). To help patients initiate and maintain multiple healthy lifestyle changes, clinicians should help patients change specific behaviors that most greatly increase self-efficacy. Strategies that enhance self-efficacy include ensuring that the patient is interested in changing the targeted behavior, having the patient discuss past successful attempts at altering the behavior, setting achievable and specific goals, tracking behavior change, making a plan for possible obstacles, and providing positive reinforcement for behavior change. Even though falling short of a goal may impede self-efficacy, clinicians can temper a potential decline in self-efficacy by helping patients understand what they did well and what they can do to improve. Pairing these self-efficacy promoting strategies with specific behavior changes that are easy to accomplish and empowering might increase the success of behavior change interventions.

Acknowledgements

The Make Better Choices trial was supported by National Institutes of Health (NIH) grant HL075451 to Dr. Spring, by the Robert H. Lurie Comprehensive Cancer Center grant (NIH P30 CA060553), by NIH K07 CA154862 to Dr. Siddique, and by NIH F31 MH070107 to Dr. Schneider.

References

  1. Ainsworth BE, Haskell WL, Whitt MC, Irwin ML, Swartz AM, Strath SJ, O'Brien WL, Bassett DR, Jr, Schmitz KH, Emplaincourt PO, Jacobs DR, Jr, Leon AS. Compendium of physical activities: An update of activity codes and MET intensities. Medicine and Science in Sports and Exercise. 2000;32(9 Suppl):S498–504. doi: 10.1097/00005768-200009001-00009. [DOI] [PubMed] [Google Scholar]
  2. Anderson ES, Winett RA, Wojcik JR, Williams DM. Social cognitive mediators of change in a group randomized nutrition and physical activity intervention: Social support, self-efficacy, outcome expectations and self-regulation in the guide-to-health trial. Journal of Health Psychology. 2010;15(1):21–32. doi: 10.1177/1359105309342297. [DOI] [PubMed] [Google Scholar]
  3. Anderson ES, Winett RA, Wojcik JR, Winett SG, Bowden T. A computerized social cognitive intervention for nutrition behavior: Direct and mediated effects on fat, fiber, fruits, and vegetables, self-efficacy, and outcome expectations among food shoppers. Annals of Behavioral Medicine. 2001;23(2):88–100. doi: 10.1207/S15324796ABM2302_3. [DOI] [PubMed] [Google Scholar]
  4. Bandura A, Adams NE, Beyer J. Cognitive processes mediating behavioral change. Journal of Personality and Social Psychology. 1977;35(3):125–139. doi: 10.1037//0022-3514.35.3.125. [DOI] [PubMed] [Google Scholar]
  5. Bergh IH, van Stralen MM, Grydeland M, Bjelland M, Lien N, Andersen LF, Anderssen SA, Ommundser Y. Exploring mediators of accelerometer assessed physical activity in young adolescents in the health in adolescents study - a group randomized controlled trial. BMC Public Health. 2012;12(814) doi: 10.1186/1471-2458-12-814. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bickel WK, Madden GJ, Petry NM. The price of change: The behavioral economics of drug dependence. Behavior Therapy. 1998;29(4):545–565. [Google Scholar]
  7. Block G, Gillespie C, Rosenbaum EH, Jenson C. A rapid food screener to assess fat and fruit and vegetable intake. Am J Prev Med. 2000;18:284–288. doi: 10.1016/s0749-3797(00)00119-7. [DOI] [PubMed] [Google Scholar]
  8. Centers for Disease Control and Prevention Prevalence and trends data 2009: Behavioral risk factor surveillance system. 2011 Retrieved from http://apps.nccd.cdc.gov/brfss/ [Google Scholar]
  9. Coups EJ, Gaba A, Orleans T. Physician screening for multiple behavioural health risk factors. American Journal of Preventive Medicine. 2004;27:34–41. doi: 10.1016/j.amepre.2004.04.021. [DOI] [PubMed] [Google Scholar]
  10. Evers KE, Quintiliani LM. Advances in multiple health behavior change research. Translational Behavioral Medicine. 2013;3(1):59–61. doi: 10.1007/s13142-013-0198-z. doi: 10.1007/s13142-013-0198-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Fine LJ, Philogene GS, Gramling R, Coups EJ, Sinha S. Prevalence of multiple chronic disease risk factors. 2001 national health interview survey. American Journal of Preventive Medicine. 2004;27:18–24. doi: 10.1016/j.amepre.2004.04.017. [DOI] [PubMed] [Google Scholar]
  12. Fritz MS, Mackinnon DP. Required sample size to detect the mediated effect. Psychological Science. 2007;18(3):233–239. doi: 10.1111/j.1467-9280.2007.01882.x. doi:PSCI1882. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Guillaumieab L, Godinac G, Manderscheidb J, Spitzd E, Muller L. The impact of self-efficacy and implementation intentions-based interventions on fruit and vegetable intake among adults. Psychology and Health. 2012;27(1):30–50. doi: 10.1080/08870446.2010.541910. [DOI] [PubMed] [Google Scholar]
  14. Higgins TJ, Middleton KR, Winner L, Janelle CM. Physical activity interventions differentially affect exercise task and barrier self-efficacy: A meta-analysis. Health Psychology. 2014;33(8):891–903. doi: 10.1037/a0033864. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hildebrand DA, Betts NM. Assessment of stage of change, decisional balance, self-efficacy, and use of processes of change of low-income parents for increasing servings of fruits and vegetables to preschool-aged children. Journal of Nutrition Education and Behavior. 2009;41(2):110–119. doi: 10.1016/j.jneb.2008.09.007. [DOI] [PubMed] [Google Scholar]
  16. Kohl HW, Blair SN, Paffenbarger RS, Jr, Macera CA, Kronenfeld JJ. A mail survey of physical activity habits as related to measured physical fitness. American Journal of Epidemiology. 1988;127:1228–39. doi: 10.1093/oxfordjournals.aje.a114915. [DOI] [PubMed] [Google Scholar]
  17. Laing BY, Mangione CM, Tseng C, Leng M, Vaisberg E, Mahida M, Bholat M, Glazier E, Morisky DE, Bell DS. Effectiveness of a smartphone application for weight loss compared with usual care in overweight primary care patients: A randomized, controlled trial. Annals of Internal Medicine. 2014;161(10_supplement):S5–S12. doi: 10.7326/M13-3005. doi:10.7326/M13-3005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lewis BA, Marcus BH, Pate RR, Dunn AL. Psychosocial mediators of physical activity behavior among adults and children. American Journal of Preventive Medicine. 2002;23(2):26–35. doi: 10.1016/s0749-3797(02)00471-3. [DOI] [PubMed] [Google Scholar]
  19. Ling AM, Horwath C. Self-efficacy and consumption of fruit and vegetables: Validation of a summated scale. American Journal of Health Promotion. 1999;13(5):290–298. doi: 10.4278/0890-1171-13.5.290. [DOI] [PubMed] [Google Scholar]
  20. Marcus BH, Selby VC, Niaura RS, Rossi JS. Self-efficacy and the stages of exercise behavior change. Research Quarterly for Exercise and Sport. 1992;63(1):60–6. doi: 10.1080/02701367.1992.10607557. [DOI] [PubMed] [Google Scholar]
  21. Mattila E, Orsama AL, Ahtinen A, Hopsu L, Leino T, Korhonen I. Personal health technologies in employee health promotion: Usage activity, usefulness, and health-related outcomes in a 1-year randomized controlled trial. JMIR Mhealth and Uhealth. 2013;1(2):e16. doi: 10.2196/mhealth.2557. doi: 10.2196/mhealth.2557. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. McGinnis JM, Foege WH. The immediate vs the important. The Journal of the American Medical Association. 2004;291(10):1263–1264. doi: 10.1001/jama.291.10.1263. doi:10.1001/jama.291.10.1263. [DOI] [PubMed] [Google Scholar]
  23. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United States, 2000. The Journal of the American Medical Association. 2004;291(10):1238–45. doi: 10.1001/jama.291.10.1238. [DOI] [PubMed] [Google Scholar]
  24. Mosteller F, Tukey JW. Data analysis and regression: A second course in statistics. Addison-Welsey; Reading, MA: 1977. [Google Scholar]
  25. Muravan M, Tice DM, Baumeister RF. Self-control as limited resource: Regulatory depletion patterns. Journal of Personality and Social Psychology. 1998;74(3):774–789. doi: 10.1037//0022-3514.74.3.774. [DOI] [PubMed] [Google Scholar]
  26. Olander EK, Fletcher H, Williams S, Atkinson L, Turner A, French DP. What are the most effective techniques in changing obese individuals' physical activity self-efficacy and behaviour: a systematic review and meta-analysis. The International Journal of Behavioral Nutrition and Physical Activity. 2013;10(29) doi: 10.1186/1479-5868-10-29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Parschau L, Fleig L, Koring M, Lange D, Knoll N, Schwarzer R, Lippke S. Positive experience, self-efficacy, and action control predict physical activity changes: A moderated mediation analysis. British Journal of Health Psychology. 2013;13(2):395–406. doi: 10.1111/j.2044-8287.2012.02099.x. [DOI] [PubMed] [Google Scholar]
  28. Parschau L, Fleig L, Warner LM, Pomp S, Barz M, Knoll N, Schwarzer R, Lippke S. Positive exercise experience facilitates behavior change via self-efficacy. Health Education & Behavior. 2014;41(4):414–422. doi: 10.1177/1090198114529132. [DOI] [PubMed] [Google Scholar]
  29. Preacher KJ, Hayes AF. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods. 2008;40:879–891. doi: 10.3758/brm.40.3.879. [DOI] [PubMed] [Google Scholar]
  30. Prochaska JJ, Velicer WF, Nigg CR, Prochaska JO. Methods of quantifying change in multiple risk factor interventions. Preventive Medicine. 2008;46(3):260–265. doi: 10.1016/j.ypmed.2007.07.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Rozin P, Levin E, Stoess C. Chocolate craving and liking. Appetite. 1991;17(3):199–212. doi: 10.1016/0195-6663(91)90022-k. [DOI] [PubMed] [Google Scholar]
  32. Shaikh AR, Yaroch AL, Nebeling L, Yeh MC, Resnicow K. Psychosocial predictors of fruit and vegetable consumption in adults a review of the literature. American Journal of Preventive Medicine. 2008;34(6):535–543. doi: 10.1016/j.amepre.2007.12.028. [DOI] [PubMed] [Google Scholar]
  33. Sondik EJ, Huang DT, Klein RJ, Satcher D. Progress toward the healthy people 2010 goals and objectives. Annual Review of Public Health. 2010;31:271–281. doi: 10.1146/annurev.publhealth.012809.103613. doi: 10.1146/annurev.publhealth.012809.103613. [DOI] [PubMed] [Google Scholar]
  34. Spring B, King A, Pagoto S, Van Horn L, Fisher J. Fostering Multiple Healthy Lifestyle Behaviors for Primary Prevention of Cancer. American Psychologist. 2015;70(2):75–90. doi: 10.1037/a0038806. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Spring B, Schneider K, McFadden HG, Vaughn J, Kozak AT, Smith M, Moller AC, Epstein L, Russell SW, DeMott A, Hedeker D. Make better choices (MBC): Study design of a randomized controlled trial testing optimal technology-supported change in multiple diet and physical activity risk behaviors. BMC Public Health. 2010;10:586–2458-10-586. doi: 10.1186/1471-2458-10-586. doi:10.1186/1471-2458-10-586. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Spring B, Schneider K, McFadden HG, Vaughn J, Kozak AT, Smith M, Moller AC, Epstein LH, Demott A, Hedeker D, Siddique J, Lloyd-Jones DM. Multiple behavior changes in diet and activity: A randomized controlled trial using mobile technology. Archives of Internal Medicine. 2012;172(10):789–796. doi: 10.1001/archinternmed.2012.1044. doi:10.1001/archinternmed.2012.1044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Stroop JR. Studies of interference in serial verbal reactions. Journal of Experimental Psychology. 1935;18:643–662. [Google Scholar]
  38. Teixeira PJ, Carraca EV, Marques MM, Rutter H, Oppert JM, De Bourdeaudhuij I, Lakerveld J, Brug J. Successful behavior change in obesity interventions in adults: A systematic review of self-regulation mediators. BMC Medicine. 2015;13:84–015-0323-6. doi: 10.1186/s12916-015-0323-6. doi:10.1186/s12916-015-0323-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology. 1988;54(6):1063–1070. doi: 10.1037//0022-3514.54.6.1063. [DOI] [PubMed] [Google Scholar]

RESOURCES