Vicarious Experience to Affect Physical Activity in Women: A Randomized Control Trial

Abstract

People compare themselves to others for self-evaluation, practical information, and motivation for healthy behaviors. The effect of active peer models on comparative thinking is unknown. The purpose of this 12-week, randomized, two-group pilot study was to evaluate the effect of a workplace peer modeling intervention on self-efficacy, motivation, and comparative thinking. The attention control group (ACG; n = 24) received general health information. The intervention group (n = 26) met with active peer models, received an exercise prescription and information. No significant group by time interaction effects were found. Comparisons on ability (how well am I doing), opinions (what should I think or believe), future self (think about my future), and modeling (be like someone else) all increased in the intervention group (n = 21) but decreased in the ACG (n = 22). Active peer models may support physical activity behavior change through specific lines of comparative thinking.

A mere 24% of U.S. adults meet the current recommendations for physical activity (PA) known to prevent disease and promote health; 150 minutes of moderate-level aerobic activity and 2 days of strengthening activities per week (Centers for Disease Control and Prevention, 2017). Leading health organizations encourage the use of evidenced-based strategies to improve health behaviors, and self-efficacy development is a recommended strategy to increase PA (Artinian et al., 2010). According to social cognitive theory, there are four known ways to improve self-efficacy of a particular behavior: (a) mastery experiences, (b) verbal persuasion, (c) physiologic feedback, and (d) vicarious experience (Bandura, 1998). Review of self-efficacy and PA literature suggests that of the four sources of self-efficacy, vicarious experience is the least studied and therefore the least understood (Ashford, Edmunds, & French, 2010; Conn, Hafdahl, & Mehr, 2011). Vicarious experience is observing someone else succeed at a given task or behavior. Closely related to self-efficacy and vicarious experience is the theory of social comparison; comparing oneself to another for self-evaluation, self-enhancement, and/or self-improvement (Festinger, 1954). Specifically, a self-assessment is made in comparison to another in either an upward (they are better off than me) or downward (they are worse off than me) manner. This then frames the motivation for behavior change to either gain health or avoid illness. Social comparison is also influenced by the perception of similarity. If the comparison person is considered too dissimilar (age, ability, and socioeconomic status), then a comparison for self-evaluation and subsequent behavior change is not likely (Festinger, 1954).

Social comparison theory has supported effective health behavior change interventions targeting condom use, smoking cessation, weight control, and sunscreen use (Gerrard, Gibbons, Lane, & Stock, 2005; Mueller, Pearson, Muller, Frank, & Turner, 2010; O’Riordan, Geller, Brooks, Zhang, & Miller, 2003; Tigges, 2001). A few studies have used social comparison theory to support PA behavior interventions by targeting the self-evaluation aspect of comparative thinking. Among female employees receiving a pedometer-based PA intervention, those receiving daily individual performance feedback plus comparison to performance of fellow participants had a significantly greater step count and greater compliance with self-monitoring than those who only received individual performance feedback (Chapman, Colby, Convery, & Coups, 2016). Similarly, among community-dwelling older adults, those who received intrapersonal behavior change strategies (included performance comparisons with group members) had significantly greater minutes of weekly PA, compared to those who received interpersonal behavior change strategies (goal setting, barrier management; McMahon et al., 2017). These two studies suggest self-evaluation, in comparison to others making the same PA behavior change, is an effective intervention ingredient.

There is however, more to understand about Festinger’s (1954) other motives for making social comparisons (self-enhancement and self-improvement) and how those motives might support effective PA behavior interventions. Recent meta-analytic evidence indicates larger effects result when study participants: (a) are primed with similarities they share with the target, and (b) compare themselves to a socially “local” versus “distant” target (Gerber, Wheeler, & Suls, 2018). While similar, socially local targets are best for comparison-driven self-evaluations, the best conditions for supporting the self-enhancement and self-improvement motives of social comparison remain unclear. For example, it may be that comparison-driven self-evaluation jumpstarts PA behavior and encouragement and/or practical information from the comparison target sustains the PA behavior. In addition, social comparison research using control conditions is needed. According to Gerber et al. (2018), only 27% (n = 145) of social comparison studies reviewed were designed with a control condition. This randomized attention control PA intervention study evaluates social comparison thinking both in general and specific to motives for PA behavior change.

Social Comparison Theory as a Tenet of Vicarious Experience

To better understand the effects of self-efficacy for PA development through vicarious experience, measurement of social comparison as a theoretical concept is needed. Only two measures were found that measure comparative thinking; the Iowa-Netherlands Comparison Orientation Measure (INCOM; Gibbons & Buunk, 1999) and the Social Comparison Motives Scale (SCMS; Tigges, 2009). The INCOM measures general tendency to compare oneself to others on self-performance (abilities) and self-beliefs (opinions; Gibbons & Buunk, 1999). The SCMS measures five motives for comparison thinking: distancing (show me what not to do), similarity identification (trusting someone because of shared characteristics like age), self-enhancement (to feel good about myself), modeling (to be like someone admired), and future self (to think about my future). The SCMS was originally developed to measure motives for comparison thinking about pregnancy prevention behavior in adolescents (Tigges, 2009). Items on the SCMS are generic in form and could be applied to various situations in which people might compare themselves to others regarding a behavior change such as increasing PA.

The authors recently conducted a randomized, two-group (intervention and attention control) workplace PA pilot study in inactive women (Rowland et al., 2018). The intervention used active peer models to provide vicarious experience for increasing PA. Both groups increased in ActiGraph measured PA F (df = 1) = 11.4, p = .002 (Rowland et al., 2018). Although there were no significant intervention effects, the intervention group (IG) had greater improvement in cardiorespiratory fitness (primary outcome) and cardiovascular risk variables, compared to the attention control group (ACG; Rowland et al., 2018). The purpose of this study was to evaluate the effect of the active peer model workplace PA intervention on the theoretical variables of self-efficacy for PA, motivation, and comparison thinking. It was hypothesized that compared to the ACG, the IG would have higher self-efficacy for PA, motivation, and comparative thinking.

Methods

Design

A randomized, two-group, repeated-measures design was used. A statistician-generated random assignment schedule was used to allocate eligible participants to either the IG or the ACG. Measures were collected prior to randomization at baseline and repeated at 12 weeks, immediately after intervention.

Sample and Setting

Physically inactive women (self-report ≤60 moderate or ≤20 vigorous minutes PA/week) working at a Midwestern health system were recruited from mid-December 2015 through January 2016 using the organization’s intranet home page, weekly newsletter, and employee health office. The study site employs approximately 3,500 people, 80% of which are women. Inclusion criteria were as follows: women, 19 to 65 years of age, employed at least 20 hours/week, and physically inactive by self-report. Exclusion criteria were as follows: high cardiovascular risk (Arena, Pescatello, Riebe, & Thompson, 2014), body mass index (BMI) ≥45 kg/m², unable to complete cycle fitness testing, pregnant or planned pregnancy during the study to control for metabolic heterogeneity among participants, night shift worker, currently participating in a weight-loss or exercise program, and currently taking a beta-blocker medication.

Sample size was determined using G*Power (Faul, Erdfelder, Lang, & Buchner, 2007) using univariate repeated-measures analysis of variance (ANOVA) to test within-between interaction (time*group effect). With alpha = 0.1, two groups over two time points, a sample size of 50 has 80% power to detect an effect size of f = 0.178, which is equivalent to a (d) of 0.357. The alpha level used is appropriate for a preliminary study (Hertzog, 2008). The medium effect size was consistent with the effect sizes found in workplace PA interventions using maximal oxygen consumption VO_2max to measure fitness outcomes (Conn, Hafdahl, Cooper, Brown, & Lusk, 2009).

From the 187 women who attended informational sessions and/or emailed the principle investigator (PI) about participation, 52 completed informed consent and were randomized to the ACG (n = 26) or the IG (n = 26). Two participants were removed from the ACG following baseline assessments because they were considered high cardiovascular risk and not appropriate for cycle fitness testing in the community setting; one for BMI ≥45 kg/m² and the other for a new diagnosis of type II diabetes. Despite more in the IG receiving the allocated intervention compared to the ACG (IG n = 26; ACG n = 24), drop-out and loss to follow up were higher in the IG. The resulting sample for analysis was ACG n = 22 and IG n = 21.

Intervention

Preparation.

Active female employees from the study site were recruited to serve as active peer models to provide vicarious experience for living an active lifestyle. A total of 21 women were interviewed about their PA and 7 were selected to deliver the intervention. Per self-report, these peer models engaged in ≥150 minutes of moderate-level PA/week, for at least the past 6 months and were diverse in age, ethnicity, occupation, and experiences with PA. Following training on ethical research conduct and orientation to the study, the peer models were asked to create a 20-minute PowerPoint describing their motivation, barriers, facilitators, self-monitoring, goal setting, and technology used for PA.

The theoretical basis for the intervention was social comparison theory (self-improvement based on comparison to others; Festinger, 1954) and self-efficacy theory (confidence in one’s capabilities for a given behavior; Bandura, 1998). In this study, peer models provided social comparison/vicarious experience to participants by describing themselves and their experience with PA. Participants could then compare themselves to someone similar (i.e., age, occupation, and family demands), yet successful with a healthy level of PA.

Intervention group.

The IG participants had a one-time, 20-minute private meeting with a nurse practitioner trained in exercise testing and health coaching, to review fitness testing and receive an exercise prescription. An appropriate level of PA intensity was determined for IG participants based on their baseline cycle fitness test: 40% to 60% of HR_reserve for low fit and 60% to 70% of HR_reserve for moderately fit participants. The exercise prescription included: (a) a prescribed target heart rate (HR) calculated from cycle test results (HR_max–HR_rest × % intensity desired + HR_rest; Arena et al., 2014), (b) teaching on the rating of perceived exertion (RPE) scale (Borg, 1982), (c) direction on submitting weekly logs documenting type and minutes of PA, peak HR, and peak RPE during moderate-level PA lasting ≥10 minutes, and (d) direction to engage in ≥150 minutes of moderate-level (RPE 12 15) PA/week.

Over the 12-week intervention, the IG group met every other week at the workplace for a 45-minute group lunch and learn presentation. Each of the six sessions included: 10 minutes of PI presenting the on the health benefits of PA, 20 minutes of the peer model presenting their PA story, and 10 to 15 minutes of the peer model taking questions from participants. The IG group encountered seven different peer models; one per session except for the last session. The last session featured two peer models as the extra peer model was not needed as a substitute during the study.

Attention control group.

The ACG did not receive results of their fitness test or an exercise prescription until after postintervention measures were collected. They were asked to maintain their baseline level of PA throughout the study. Over the 12-week intervention, the ACG met every other week for six lunch and learn sessions. The 45-minute health topic presentations were delivered by four different master’s prepared nurses (including PI) with expertise in the topic areas of healthy diet, cancer prevention, stress management, pain management, and sleep.

Procedures.

The study was approved by the study site and the University’s Institutional Review Board. Following informed consent, participants were fitted with an ActiGraph to wear for 7 consecutive days and given paper questionnaires to complete and return in a sealed envelope when they returned 7 to 10 days later for the physical assessments with the nurse practitioner. Following the completion of all baseline assessments, participants were randomized to either the ACG or the IG. Postintervention, participants were again fitted with an ActiGraph to wear for 7 consecutive days, given paper questionnaires, and then returned 7 to 10 days later to return the ActiGraph, questionnaires in a sealed envelope, and complete the physical assessment with the nurse practitioner. An advanced practice nurse investigator trained in fitness testing and not blinded to group, completed all physical assessments and collected questionnaires in a private area of the employee fitness center.

Measures

Study measures collected at baseline and postintervention at 12 weeks (Table 1).

Table 1.

Study Measures, Concepts, and Theoretical Definitions.

Measure	Concept/Subconcepts	Definition
Adapted Self-Efficacy and Exercise Habits Survey	Self-efficacy for PA	Confidence to continue PA despite barriers
	• Sticking to it
	• Making time	Confidence to dedicate time to PA
Investigator-developed questions	Motivation for PA	Motivated by another to be physically active
INCOM	Social Comparison Orientation
	• Abilities	Self-evaluation of performance compared to others
	• Opinions	Self-evaluation of thoughts and feelings compared to others
Adapted SCMS	Social Comparison Motives
	• Distancing	To want to be unlike someone else
	• Similarity identification	Recognition of shared attributes with someone else (age, job, and family status)
	• Self-enhancement	To feel good about oneself
	• Modeling	To follow the example of someone else
	• Future self	To think about oneself in the future

Note. PA = physical activity; INCOM = Iowa-Netherlands Comparison Orientation Measure; SCMS = Social Comparison Motives Scale.

Self-efficacy for PA.

Confidence for adopting and maintaining moderate-intensity PA for at least 6 months was measured using the 12-item Self-Efficacy and Exercise Habits Survey (Sallis, Pinski, Grossman, Patterson, & Nader, 1988). This questionnaire has two subscales: “making time for exercise” and “sticking to exercise.” Higher scores on “making time” indicated greater confidence to dedicate time to PA. Higher scores on “sticking to it” indicated greater confidence to engage in PA despite barriers. This tool has test–retest reliability (0.68) and internal consistency (alpha coefficients 0.83 and 0.85; Sallis et al., 1988). The questionnaire was modified by changing the word “exercise” to “physical activity.” At baseline, Cronbach’s alpha for “sticking to it” was 0.86 and “making time” was 0.80; at postintervention “sticking to it” was 0.88 and “making time” was 0.62. Concurrent criterion-related validity and convergent validity was established by using a multinomial logistic regression model to evaluate the association between self-efficacy and PA category (inactive, irregularly active, or regularly active) as measured by the Behavioral Risk Factor Surveillance System, Wald chi-square (1, N = 1,931) = 266.52, p < .001 (Wilcox, Sharpe, Hutto, & Granner, 2005).

Motivation.

Motivation for PA was measured using three investigator-developed Likert-type questions. Higher scores reflect higher motivation for PA. Cronbach’s alpha at baseline was 0.86 and postintervention was 0.84. Three senior-level health researchers reviewed the items to establish face validity.

Social comparison.

General tendency to compare oneself to others was measured with the INCOM (Gibbons & Buunk, 1999). This 11-item Likert-type questionnaire has two subscales measuring comparisons of “ability” (5 items) and “opinions” (6 items). The tool has high reliability (Cronbach’s alpha, 0.78-0.85) of abilities and opinions, respectively (Gibbons & Buunk, 1999). Higher scores indicate more comparison thinking of “abilities” and “opinions.” Cronbach alphas at both time points were 0.74 for “abilities” and 0.82 and 0.69 for “opinions.”

Reasons for making comparison with others were measured using an adapted Social Comparison Motive Scale (SCMS; Tigges, 2009). The 19-item Likert-type questionnaire has five subscales to measure motives for comparison: distancing (three items), similarity identification (three items), enhancement (four items), modeling (three items), and future self (six items). Higher scores indicate more comparisons for that motive. The questionnaire was tailored to PA behavior with the instructional sentence “Think about comparisons you make with other people when thinking about becoming physically active.” The original tool was reliable (Cronbach’s alpha = 0.91) and valid (content validity index = 1.0) in adolescents (Tigges, 2009). In this study, all social comparison subscales had acceptable reliability and Cronbach alphas ranged from 0.80 to 0.92. The SCMS has convergent validity with the INCOM (r = .50, p = .000) and divergent validity with the Rosenberg Self-Esteem Scale (r = .15, p = .003; Tigges, 2009).

Analysis

Participants were examined on an intent-to-treat basis. Statistical Package for Social Sciences (SPSS) software (v. 24) was used to screen and analyze demographic and outcome variables. Analysis using repeated-measures ANOVA was initially planned. However, due to participant drop-out (IG = 9; ACG = 3), hierarchical linear modeling was used to examine group by time interactions for the outcome variables. Statistical significance was determined with a p value < 0.10.

Results

At baseline, both groups were similar in age: ACG (M = 43.4 ± 8.8 years) and IG (M = 43.5 ± 12.4 years). Both groups were mostly Caucasian, married/partnered and worked full-time (Table 2). More participants in the IG worked in a clinical position (direct patient care) than in the ACG. At baseline, there were no significant group differences in self-efficacy for PA, motivation, and comparative thinking. In general, scores on each of the measures fell at or just above the mid-point for each scale indicating moderate levels of self-efficacy, motivation, and social comparison for PA at baseline (Table 3).

Table 2.

Baseline Female Participant Demographics by Group.

	Attention Control Group (n = 24)		Intervention Group (n = 26)		Test Statistic p value
Characteristic	n	%	N	%
White/Caucasian	22	91.7	24	92.3	.99a
Married/partnered	19	79.2	18	69.2	.42b
Has children	18	75.0	20	76.9	.87b
Works full-time	22	91.7	26	100	.22a
Works in direct patient care	7	29.2	17	65.4	.01b

^aFischer’s exact test used when expected cell counts <5.

^bChi-square test.

Table 3.

Mean Values and Group X Time Interaction Effects on Self-efficacy for PA, Motivation, and Social Comparison.

		Attention Control Group				Intervention Group
		Pre (n = 24)		Post (n = 22)		Pre (n = 26)		Post (n = 21)		Model Estimates
Measure	Range	M	SD	M	SD	M	SD	M	SD	Beta	T	p
Self-efficacy for PA
Sticking to it	7-35	3.79	0.81	3.49	0.85	3.95	0.75	3.78	0.74	0.046	0.215	.830
Making time	4-20	4.00	0.84	3.71	0.83	4.18	0.81	3.75	0.74	−0.186	−0.794	.432
Motivation	3-15	9.21	3.34	9.27	2.89	9.77	3.94	9.81	2.96	−0.430	−0.473	.639
Social Comparison Orientation (INCOM)
Abilities	5-25	15.63	3.96	15.09	3.97	15.19	3.83	15.52	3.17	0.419	0.572	.570
Opinions	6-30	19.50	4.61	18.82	3.94	18.31	4.68	19.17	3.14	0.596	0.701	.487
Motive for social comparison (adapted SCMS)
Distancing	4-16	12.88	3.35	13.18	3.75	12.77	3.02	14.10	3.14	0.923	1.090	.282
Similarity	3-12	7.58	2.94	7.86	3.22	7.73	2.94	9.05	2.39	0.796	0.907	.371
Enhancement	4-16	10.38	4.20	10.68	4.32	10.54	4.16	11.76	3.83	0.528	0.536	.595
Modeling	3-12	8.46	3.33	8.41	2.77	8.65	2.62	9.95	2.50	1.282	1.617	.113
Future	6-24	18.71	5.40	18.45	5.82	18.58	5.27	20.24	4.19	1.468	1.058	.296

Note. Hierarchical linear modeling used to test group by time interaction effect. PA = physical activity; INCOM = Iowa-Netherlands Comparison Orientation Measure; SCMS = Social Comparison Motives Scale.

There were no group by time interactions for self-efficacy, motivation, and comparative thinking for PA (Table 3). Over time, self-efficacy for PA scores fell in both groups while motivation scores remained stable. On the adapted social comparison motives measure (SCMS) both groups increased in comparisons to distance themselves from someone who is inactive (distancing), to find similarities with someone who is active (similarity identification) and to feel better about themselves (self-enhancement). Though not significant, there were notable group differences in four of the seven areas of comparison thinking. Comparisons on ability (how well am I doing), opinions (what should I think or believe), future self (to think about my future), and modeling (be like someone else) all increased in the IG but decreased in the ACG (Table 3).

Discussion

The intervention in this study was designed to provide vicarious experience for living a physically active lifestyle. Although there was not a significant intervention effect on social comparison thinking, there were noteworthy group differences. The IG increased comparative thinking on all seven measures by the end of the study. The ACG had mixed changes; increasing on three measures and decreasing on four measures of comparative thinking. At the end of the study, general comparisons on abilities (how am I doing) and opinions (what should I think or feel) were increased in the IG and decreased in the ACG. Similarly, two motives for making social comparisons were increased in the IG and decreased in the ACG; modeling (to be like someone else) and future self (to think about my future).

This study begins to address what people may attend to or be motivated by in a vicarious experience to change a health behavior. In this study, comparative thinking may increase PA by identifying with a trusted model and thinking about future self as an active person. Although not significant, the IG did have greater improvements in measures of cardiovascular fitness and risk compared to the ACG (Rowland et al., 2018). A prior observational study on PA behavior in a large sample of adolescents (n = 2,387) found higher levels of INCOM-measured social comparison correlated with higher levels of self-reported PA (Luszczynska, Gibbons, Piko, & Tekozel, 2004). There is a relationship between social comparison thought and other health behaviors. Smokers who have higher levels of social comparison have greater success with cessation than smokers with lower levels of social comparison (Gerrard et al., 2005). Adolescents with contact to HIV-positive peers had increased motivation to get tested for HIV (Misovich, Fisher, & Fisher, 1997).

In this study, both groups decreased in self-efficacy for PA. Although PA increased in both groups, the IG did have greater fitness and less cardiovascular risk than the ACG, postintervention (Rowland et al., 2018). These findings suggest: (a) the social comparison peer modeling interventions may be an effective strategy for increasing PA behavior, (b) a different tool to measure self-efficacy may have had different results, and (c) the timing of when self-efficacy is measured may matter. Perhaps, participants in the IG had a decline in self-efficacy once they started trying to meet the PA goals of the study. In other words, the baseline self-efficacy scores may reflect participant’s self-confidence for being active when they are not, and postintervention scores may reflect more realistic PA self-efficacy after actually trying to meet the goals of the study (150 minutes moderate-level PA at prescribed HR and RPE). It remains unclear if self-efficacy is the best indicator of health behavior change. A positive, but non-significant relationship between PA behavior and PA self-efficacy has been found (Ashford et al., 2010).

At baseline, both groups were similar in level of motivation for PA, and both groups maintained this level through the study. It is noteworthy that the IG maintained motivation for PA despite dropping in self-efficacy for PA. Stable motivation with a drop in self-efficacy suggests the need for multiple measures of the theoretical constructs of health behaviors.

A strength of this study is that it is the first known study to use a randomized, two-group intervention design to affect PA behavior through vicarious experience using peer models. Although not significant in this small sample, group differences on some measures of social comparison suggest the intervention has promise for affecting some motives for comparative thinking. Use of two different measures of social comparison, a poorly understood concept in health behavior, is another strength. However, the SCMS only has validation on pregnancy avoidance behavior in the adolescent population. Although the SCMS demonstrated acceptable internal consistency reliability in this study, additional psychometric testing of the SCMS on PA behavior in the adult population would strengthen support for its use.

The generalizability of study findings is limited by the small sample size, predominately Caucasian sample, and single workplace setting for the intervention. Motivation assessment was limited by use of three investigator-developed items. Additional time point collection of all variables (social comparison, self-efficacy, and motivation) may strengthen evaluation of these theoretical concepts over time. For example, all or some of these theoretical concepts may be overestimated at baseline and reset to a lower level once the reality of increasing PA behavior is experienced early in the intervention.

In the workplace, peer models may be a source of vicarious experience to increase PA through social comparison processes. Comparative thinking may increase PA by identifying with a trusted model and thinking about future self as an active person. More needs to be understood about the relationship between making social comparisons and making health behavior change and is that relationship different based on age ( adolescent vs. adult), type of health behavior, or stage of change readiness. Further research in this area would help strengthen current general recommendations to use vicarious experience in interventions targeting health behaviors like PA, by providing guidance on who and how to best use the strategy.