CIVIC ENGAGEMENT’S ROLE IN WELL-BEING: EXAMINING A MEDIATIONAL MODEL ACROSS TWO SAMPLES

Introduction: Generally, civic engagement seems to produce positive health outcomes for those who participate. However, there is conflicting evidence as to whether civic engagement may be more of an added burden for individuals from lower socioeconomic (SES) communities, or among women, perhaps due to the differing types of civic activity in which men and women may choose to engage. Type of activity may also be critical in determining whether certain forms of engagement more strongly predict well-being than others. This study explored several possible factors that inform for whom and under what circumstances civic engagement may improve health outcomes among college students. Methods: Participants were recruited at a mid-sized university in the Northeast. Consented participants completed an online survey consisting of demographics, the Civic Engagement Quiz, Mental Health Continuum-SF, Meaning in Life Questionnaire, Self-Efficacy Towards Service scale, and the Interpersonal Support Evaluation List-12. Mediational, direct, and full effects models were conducted to investigate whether SES predicts civic activity, and if type of civic activity predicts subjective well-being, mediated by meaning in life and self-efficacy. Social support was added to the model as a covariate, and all models were conducted across samples of men and women. Results: Participants (N=438) were primarily White (78%), female (72%) and 68% reported at least one parent having a Bachelor’s degree or higher. Students reported mean levels of well-being that indicated flourishing (M=45.42, SD=14.23), high self-efficacy toward service (M=20.17, SD=3.63), adequate social support (M=25.76, SD=6.87), and moderate meaning in life (M=50.22, SD=8.98). Overall, fit indices showed that the full model provided the best fit in men, women, and both samples combined (low χ to df ratio, nonsignificant p values, CFI > .95, RMSEA < .10). Due to power concerns, a smaller model was examined using SES as the first predictor variable, mediated by engagement type, with well-being as the dependent variable. Multisample analysis across gender groups showed that the mediational model was the best fit [χ(8)=17.19, p=.03;CFI=.95;RMSEA=.08, 90%CI [.03,.14]; R=.04(W), .05(M)], and there was no significant difference in model fit by gender. Conclusion: Results from this sample showed that the full model in the larger path analysis and the mediational model in the smaller path analysis provided a good fit for both men and women. Thus, contrary to prior literature, men and women did not behave in statistically different ways in this New England sample. Type of activity showed mixed results in relation to well-being, perhaps due to model complexity. Meaning in life and self-efficacy partially mediated the relationship between engagement and well-being, although both factors warrant further investigation. Given sample demographics and limits to generalizability, replication studies using more diverse populations, particularly in terms of age, race and ethnicity could greatly enhance the field if civic engagement is to be considered as a health promotional tool.


LIST OF TABLES
Illness, 2017; Rhodan, 2016). To reduce the rising prevalence of mental illness, the World Health Organization (2017) has advocated for increased efforts centered around health promotion and disorder prevention. One such strategy for health promotion could be through encouraging civic behaviors. Civic engagement-often conceptualized as individual or collective action taken for the betterment of a community-has been researched across many populations with varying results in terms of its costs and benefits to health and well-being (Chan, Ou, & Reynolds, 2014).
Prior literature has linked development of one's civic identity in adolescence to factors like school and peer influences, which become particularly pertinent to college students living on their own for the first time (Amna, 2012). Given that the age of most college students falls within a developmental period known to be critical for social, political, and civic identity development, more information is needed to disentangle the effects of civic engagement on health and well-being for this population if it is to be used as a point of intervention (Amna, 2012;Verba, Schlozman, & Brady, 1995). The current study explored several possible factors to inform for whom and under what circumstances civic engagement may be beneficial to improve health outcomes among college students.

Defining Civic Engagement
Before delving into the evidence for and against civic activity, it is important to first define and understand civic engagement as a component of its umbrella construct, social capital. Currently, many researchers utilize Robert Putnam's (1993) framework when discussing social capital theory. Putnam defines social capital along three domains: 1) networks of civic engagement (both quality and quantity of networks), 2) norms of reciprocity (I will help you today with the expectation that you will help me in the future), and 3) social trust (trust in individuals and institutions).
Putnam emphasizes civic involvement as a form of collective action, which then facilitates trust and collaboration to work together to solve future issues of public concern. According to Putnam's research, a thriving community does not generate greater social capital; rather, social capital generates thriving communities (1993).
Since Putnam's work, researchers have defined civic engagement in numerous ways, highlighting the complexities of studying this construct. Nevertheless, some instructive definitions have emerged from the literature relevant to address the question at hand. Adler & Goggin's (2005) review of the literature defines civic engagement as "the ways in which citizens participate in the life of a community in order to improve conditions for others or to help shape the community's future." Similarly, Chan et al. (2014) posited that "civic engagement includes individual and collective activities intended to identify and address issues of public concern, and enhance the well-being of one's community and the society." They go on to clarify that civic engagement can be performed at the individual level (e.g., writing a letter to city council representatives) or at the group level (e.g., volunteering with a church group at a local soup kitchen). Civic engagement behaviors typically fall within three categories: "civic activities" that involve improving the local community, "electoral activities" such as voting or political campaign canvassing, and "political voice" (referred to as "sociopolitical voice" henceforth), which involves expressing opinions pertaining to social causes (Adler & Goggin, 2005;Andolina, Keeter, Zukin, & Jenkins, 2003;Pancer, 2005). While there may be countless ways to engage, the overarching goal is for the betterment of society at large.  Kawachi, Kennedy, Lochner, and Prothrow-Stith (1997) found that social capital mediated the relationship between income inequality and mortality rates.

Social Capital and Health Outcomes
Not surprisingly, as income inequality rose, social mistrust also rose, and economically disadvantaged communities saw higher mortality rates. By contrast, every single-unit increase in average per capita group membership was associated with a decline in mortality of 83.2 deaths per 100,000 people (Kawachi et al., 1997). Taken together, these studies suggest that social capital may play a pivotal role as a protective factor in bolstering health and well-being, particularly among low-income communities. If components of social capital are to be used in adaptive interventional work with underserved populations, additional research in this field could bolster claims from Kawachi et al.'s (1997) De Silva, McKenzie, Harpham, and Huttley (2005) found that at the individual level, greater social capital was related to lower risk for mental illness among adults, with moderate evidence to support an inverse association between cognitive social capital and child mental illness. However, at the ecological level, social capital and mental illness saw mixed results: two studies found an inverse association, two found a positive association, and ten found no association.
A more recent systematic review of 31 cross-sectional (primarily observational quantitative designs) and 8 cohort studies by Ehsan and De Silva (2015) showed that high cognitive social capital was associated with lower risk of developing a common mental disorder like anxiety and depression. However, results were varied for structural social capital: most studies found no association, one found a negative association, and two studies conducted with individuals from low-to middle-income countries showed that higher structural capital correlated with higher risk of developing a common mental disorder. Thus, it appears that the quality of social networks seems to yield positive results while the quantity of networks elicits more mixed outcomes. Furthermore, Ehsan and De Silva found that there seem to be certain situations, particularly among low-income countries or for those of lower socioeconomic (SES) status, in which group memberships may place an additional strain on individuals already juggling many responsibilities. This finding seems counter to Kawachi et al.'s (1997) result in which higher per capita group membership was associated with lower mortality rates, although these studies drew from completely different samples at different points in history. Moreover, authors of the two systematic reviews acknowledged the complexities inherent when comparing multiple study designs across differing cultures. Nevertheless, more evidence is needed to better understand the role of civic networks on well-being among individuals from lower SES backgrounds.

Considered
As a component of social capital, researchers have also examined civic engagement as a stand-alone concept for its potential to influence health outcomes. In older adults, civic engagement over time has been positively correlated with better self-rated health, lower functional impairment, and reduced mortality rates (Hinterlong, Morrow-Howell, & Rozario, 2007;Post, 2017). Participants randomized to a weekly volunteer group for 10 weeks showed decreased BMI, cholesterol, and inflammation pre-to post-testing when compared to waitlist controls; volunteers who reported greater levels of altruism and mental health also saw greatest reductions in biomarker levels of inflammation, cholesterol, and body mass index (Post, 2017).
Similarly, African American women participating in a weight loss promotion program utilizing civic engagement as a complementary booster of behavior change saw positive pre-post changes in dietary behaviors, physical activity, cardiovascular fitness, and blood pressure (Brown et al., 2017). However, while participants reported enjoying the experience overall, they saw a non-significant increase in perceived stress, perhaps due to the additional time commitment. The social capital literature suggests that engagement may produce differing health outcomes for individuals from lower SES backgrounds; this study indicates that involvement in civic networks may have differing outcomes for women and/or racial-ethnic minority populations as well.
In concordance with findings from Brown et al.'s (2017) weight loss intervention, refugees from poor communities in Jordan were more likely to report lower levels of perceived health if they were a non-club member as opposed to those involved in club memberships (Khawaja et al., 2006). In other words, involved individuals reported greater perceived health. However, this relationship disappeared among women but not for men after controlling for demographic, socioeconomic, and health risk factors, indicating potential for gender differences in civic engagement health benefits among highly patriarchal cultures (Khawaja et al., 2006). While this study is limited to cross-sectional data from a single timepoint, it does align with other longitudinal studies (Brown et al., 2017;Landstedt, Almquist, Eriksson, & Hammarstrom, 2016), suggesting that perhaps gender and SES play a role in determining health outcomes.

Types of Engagement
As the evidence presented thus far has illustrated, civic engagement has not always engendered positive health outcomes, or often times there are specific qualifiers when it does. For example, a longitudinal study in Sweden found that while civic engagement was correlated with lower levels of depressive symptoms for both men and women, higher civic engagement was predictive of lower depressive symptoms at a later timepoint only for males (Landstedt et al., 2016). Men more often reported participation in sports or motor activities, while women reported greater membership in humanitarian organizations. The authors suggested that perhaps men benefit from the social influence and cohesion of sport, but humanitarian work can have more of an emotionally draining effect. Thus, it seems that gender differences as well as the type of civic activity are important considerations.
In a comparable manner, Ding, Berry, and O'Brien (2015) discovered that while social connectedness and civic engagement were somewhat predictive of mental well-being the following year, political participation was not a significant predictor; in fact, political participation was inversely correlated with mental health. Of note, the authors defined civic engagement as volunteering or group participation, which is different than the current working definition of this research, and they analyzed political participation as a separate construct. Nevertheless, results from Ding et al. (2015) suggest that specific types of engagement may produce varying levels of helpfulness-or harm-to individual health and well-being. In another example, Ziersch & Baum's (2004) cross-sectional study of civil society group participation among Australian adults found an inverse relationship between physical health and civil society group participation. Authors concluded that participation in groups may be beneficial to the community but detrimental to the individual under certain circumstances where the task is particularly mentally or physically draining (2004).
Thus, these data suggest the need to examine differences in health outcomes based on type of civic activity.

Meaning and Self-Efficacy
The literature has also noted other factors that may mediate the relationship between civic engagement and health and well-being. There is some evidence to support the direct relationship between civic engagement and well-being. Among adolescents, organizational membership and community service has been linked to a stronger sense of community, harmony, and feeling supported when compared with disengaged youth (Flanagan, 2015). A promising study using longitudinal data among high-risk youth found that civic engagement during youth was predictive of greater life satisfaction, future optimism and greater educational attainment in emerging adulthood (Chan et al., 2014). However, other researchers have speculated that while civic engagement may certainly be an important determinant of mental and physical health, it may be operationalized more indirectly. For example, in Piliavin & Siegl's (2007) 47-year longitudinal study with older adults, volunteerism predicted well-being above and beyond other social participation predictors of health. Involvement in diverse activities (more than three) and continuous involvement across the life span saw the highest levels of well-being. However, after controlling for "mattering"-the concept that people acknowledge their own importance and impact in the worldvolunteerism was no longer predictive of well-being (2007). This finding is consistent with much of the positive psychology literature, in which meaning in life-or connection to something greater than the self-often results in positive well-being (Peterson, Park, & Seligman, 2005).
Other studies have examined civic engagement's impact on measures of empowerment and self-efficacy as opposed to specific mental and physical health outcomes. Gullan, Power and Leff (2013) explored the benefits of a school-based service-learning program in which 48 inner-city ethnic minority 6th graders spent 45 minutes for 20 weeks learning about community issues, identifying an issue to address, and addressing that issue by organizing and completing a service project.
Authors grounded their approach in Zimmerman's (1995) empowerment theory, which states that sense of intrapersonal control, understanding how social/political systems interact to form power hierarchy, and behavioral efforts to exert control all lead to empowerment. Results showed that student self-reports of empowerment significantly predicted self-efficacy post-intervention, above and beyond pre-intervention self-efficacy rates or program satisfaction. In Brown et al.'s (2017) weight loss study incorporating a civic engagement component, the authors also based their framework on empowerment theory concepts. They posited that civic engagement promotes individual and collective self-efficacy to create change for one's community, thereby increasing feelings of empowerment and overall well-being. Civic engagement has the potential to positively influence well-being, given the body of literature presented herein; but, its effects could be operationalized indirectly, as was the case in some of the aforementioned studies where there appeared to be more immediate predictors of well-being. Constructs of "mattering," or sense of meaning in life, and self-efficacy in one's ability to create change are important mediators to consider when examining civic engagement's relationship with health and well-being.

Gaps in the Literature
Many studies have indicated that although positive mood begets giving, giving very strongly influences positive mood and happiness (Ding et al., 2015;Hinterlong et al., 2007;Post, 2017). Yet, the majority of studies conducted have been cross-sectional in nature, limiting the ability to draw conclusions about civic engagement's causal influence on health outcomes. To complicate matters further, much of the previous work looking at civic engagement has chosen to focus on group membership status (how many teams, clubs, and organizations an individual belongs to) as a proxy for civic engagement. While certainly a critical piece of information to consider, group membership is merely one facet of civic engagement that provides a limited understanding of civic engagement's impact. Moreover, many studies compartmentalize civic engagement into its civic, electoral, and sociopolitical voice components, choosing only to examine a single aspect of civic engagement activity.
This approach does provide helpful information in better understanding that specific type of engagement, but a more holistic approach would be helpful to glean a more accurate picture of the construct.
Multiple interventions have already utilized tenets of civic engagement with varying outcomes in efforts to prevent recidivism (Windsor, Jemala, & Benoit, 2014), empower youth (Gullan et al., 2013), re-engage those abusing substances (McCabe, 2014), reduce urban crime (Folgheraiter & Pasini, 2009), supplement weight loss interventions (Brown et al., 2017), and decrease symptoms of social isolation and powerlessness among those with mental illness (Kaplan et al., 2012). The potential is there. However, if civic engagement is to be used as a tool for health promotion among college-aged populations, especially given the nature of this critical period in defining one's civic and social identity, its benefits and burdens must be better clarified.

The Current Study
Data have shown that females do not participate civically as much as males, and that individuals of lower SES are less likely to engage compared to their higher-SES counterparts (Verba et al., 1995). Indeed, several studies have found that SES indicators like parental occupation, educational attainment, and income are predictive of future civic engagement (Brown & Lichter, 2006;Foster-Bey, 2008 what circumstances civic engagement might play a role in bolstering health and wellbeing for emerging adults. To address these questions, an electronic survey was given to a sample of undergraduate students to measure their civic involvement, meaning in life, self-efficacy, perceived social support, sociodemographic factors, and well-being. The current study proposed a model in which SES predicts civic engagement, and type of engagement predicts subjective well-being, mediated by meaning in life and selfefficacy. Based upon the literature, perceived social support was added to the model as a covariate, and the model was assessed across men, women, and men and women combined ( Figure 1). Specifically, the current study proposed the following three hypotheses: 1) A mediational, direct effects, and full model will examine various versions of a predictive model, producing differing results across gender groups, such that civic engagement will not be as strong of a predictor of well-being in women compared to men.
2) Type of engagement-civic activity (e.g., community service), electoral (e.g., canvassing, fundraising), and sociopolitical voice (e.g., marching, social justice acts)-will differentially impact reports of well-being; civic activity and sociopolitical voice forms of engagement will be associated with higher well-being scores than those who choose to engage in electoral forms of participation.
3) The relationship between socioeconomic status and well-being will be mediated by civic engagement and, in turn, by meaning in life and selfefficacy.

Procedures and Sample
Participants for this study included undergraduate students enrolled at the University of Rhode Island (URI). Course instructors and department chairs were asked via email for their assistance in soliciting volunteers to participate in an online survey regarding student health behaviors and wellness. Both in-person and email announcements were made for introductory courses known to enroll a significant undergraduate population. An email template describing the study and its purpose, contact information for the student investigator, and a link to an online Qualtrics survey was sent to instructors to forward to their students. Students responding to this email request for participation were directed to complete an online survey through Qualtrics. The first page in Qualtrics presented the informed consent. Students were asked for their consent to participate, and all students were required to select "yes" at the bottom of the form as acknowledgment of their willingness to participate prior to answering any survey items. Students who selected "no" were thanked for their time and exited from the survey. Of note, considerably more women than men chose to participate in this research study. Due to the nature of planned analyses to compare civic engagement levels across gender groups, an amendment request was made to the IRB to place a quota on female participants and focus recruitment efforts targeting men. Following IRB approval, emails were sent to organizations known to enlist a large proportion of males, including athletic teams, club groups, and Greek-affiliated chapters.
Students who consented to participate were asked to complete a demographic questionnaire, the Civic Engagement Quiz, the Mental Health Continuum-Short Form, the Meaning in Life Questionnaire, the Self-Efficacy Towards Service scale, and the Interpersonal Support Evaluation List-12, described shortly. A total of N=555 undergraduates opened the survey and consented to participate in this study. Students with missing data exceeding 5% and students who did not complete measures of civic engagement and well-being were not retained for analyses, reducing the final sample size to N=438 students. Demographic characteristics of the sample are summarized in Table 1.  (Diemer, Mistry, Wadsworth, Lopez, & Reimers, 2013;Verba et al., 1995).
Participants were asked to report the highest level of education that their mother (or legal guardian 1) and father (or legal guardian 2) received. Then the highest of the two reported levels was coded using the following scale: Less than high school = 1; Some high school (no diploma) = 2; Finished high school (or GED) = 3; Some college credit (no degree) = 4; Trade/technical/vocational training = 5; Associate's degree = 6; Completed Bachelor's degree at a college or university = 7; Master's/Professional/Doctorate degree = 8. Higher levels of education correspond with higher scores, and hence, higher levels of SES.

Civic Engagement Quiz (CEQ):
The CEQ is a 19-item measure that assesses for an array of civic engagement experiences within specific periods of time (Andolina et al., 2003). The measure is comprised of three subscales: civic activity, electoral activity, and expressions of political voice (referred to as sociopolitical voice in this study). Participants were asked to select activities in which they have participated "Yes, within the past 12 months," "Yes, but not within the past 12 months," or "No, never." For example, participants were asked "Have you ever worked together with someone or some group to solve a problem in the community where you live?" Given literature on how intensity and frequency of civic behavior influences well-being, and after consulting with the scale's developers, the decision was made to assign scale points to each level of responding (Yes, within the past 12 months=2; Yes, but not within the past 12 months=1; No, never=0). A final subscale score was summed by adding the total number of points on items within a particular subscale, with higher point totals corresponding to higher scores of engagement. Participants were asked to respond to questions like "During the past month, how often did you feel satisfied with life?" on a 6-point Likert scale ranging from never (0) to everyday (5). The scale is broken down into three factors: social well-being, emotional well-being, and psychological well-being. This three-factor structure has been confirmed within college populations, and the scale has demonstrated good internal consistency (>.80) and discriminant validity in the US (Emory University, 2014).

Meaning in Life Questionnaire (MLQ):
The MLQ is a 10-item scale broken into two subscales, presence of meaning and the search for meaning (Steger, Frazier, Oishi, & Kaler, 2006). Participants were asked to respond to questions like "My life has a clear sense of purpose" using a 7-point Likert scale ranging from Absolutely Untrue (1) to Absolutely True (7). Both subscales have demonstrated adequate internal consistency (.86, .87).

Self-Efficacy Towards Service (SETS):
The SETS is a 5-item assessment using a 5-point Likert scale ranging from Strongly Agree to Strongly Disagree (Weber, Weber, Sleeper, & Schneider, 2004). Participants were asked to respond to items like: "I have confidence in my ability to help others." The scale measures how strongly students believe in their ability to contribute time and service to the larger community. The SETS has been validated in an undergraduate sample of business majors (α=0.80).
Interpersonal Support Evaluation List-12: The ISEL-12 is a 12-item scale designed to measure perceived level of social support (Cohen, Mermelstein, Kamarck, & Hoberman, 1985). Participants were asked to respond to questions such as "If I were sick, I could easily find someone to help me with my daily chores" using a 4point Likert scale ranging from definitely false (0) to definitely true (3). Cronbach's alphas compiled from four studies revealed good internal consistency for the overall score (α=0.80-0.90) (Cohen, 2008).

Data Analysis
Data were examined for normality and completeness. The data violated assumptions of normality for variables of parental education (used as a proxy for SES), self-efficacy, and electoral engagement. Parental education (SES) and selfefficacy were negatively skewed, while electoral engagement was positively skewed.
Following recommendations by Tabachnick and Fidell (2007) and Howell (2007), all three variables were transformed to better approximate a normal distribution. Parental education (SES) was transformed using square root (k-x), where k=9 given education's maximum value of 8; self-efficacy was transformed using square root (k-x), where k=26 given self-efficacy's maximum value of 25 1 ; electoral engagement was transformed using log10 (x+c), where c=1 given electoral engagement's minimum value of 0. All transformations were successful in producing an approximately normal distribution on the three variables. Data were missing across both gender groups, particularly in the measure of civic activity. Data imputation has been debated in the literature for cross-sectional designs, due to inherent limitations of having only one timepoint to estimate missing items. However, given the small sample size of men (N=119) and prior research suggesting that structural equation modeling may require samples above N=100, data imputation was warranted after examining the number of cases missing (Wolf, Harrington, Clark, & Miller, 2013). To account for missing data in both groups, full information maximum likelihood (FIML) estimation was employed to maintain adequate power, which has been recommended in the literature for structural equation modeling (Enders & Bandalos, 2001).
To address all three main research questions, multisample path analysis was attempted to examine the fit of mediational, direct, and full predictive models across male and female gender groups. Given the complexity of the models and the use of FIML to account for missing data, these analyses would not fully execute in the EQS statistical software package. Hence, the mediational, direct, and full models were examined across both men and women, in women separately, and then in men separately.

Descriptive Statistics
Descriptive statistics were calculated on all variables for men and women (Table 2). Of note, only two of 438 respondents identified with a gender other than "man" or "woman." Given this limited sample, these two cases were not included in analyses examining gender differences, although future research would greatly benefit from including larger populations of gender non-binary and non-conforming respondents. About 68% of the overall sample reported having a parent/guardian with a Bachelor's degree or higher. Students were most likely to engage in civic activity, followed by sociopolitical activity and then electoral activity. Importantly, the electoral engagement area encompasses activities like canvassing and fundraising that do not require an individual to be at least 18 years of age in the way that voting does; however, it is quite possible that students participating in this Spring 2018 study had not been of age to vote during the 2016 election, which may have influenced the likelihood of them engaging in other types of electoral activity. On average, students reported levels of well-being that indicate flourishing, high levels of self-efficacy towards service, adequate social support, and moderate levels of meaning in life.

Education is measured on a scale from 1-8, where 1 represents less than a high school education (low SES) and 8 represents graduate-level education (high SES) *significant at the .05 level
Correlations were calculated on all variables to be used in the path analysis models (Table 3). Independent variables differed in their correlational relationships with mediational and dependent variables. There was no evidence of multicollinearity.

Hypothesis 1: Examining Gender Differences
Results from the three predictive models can be found in Table 4

. Full model with direct and indirect effects in women and men
Due to concerns about statistical power limitations of fitting such a large model with a limited sample size among men, a smaller subset of the model was tested to better examine the impact of different types of engagement on well-being for men and women. In this model, SES was added as the first predictor variable, mediated by civic, electoral, and sociopolitical forms of engagement, with well-being as the dependent variable. Multisample analyses across men and women were conducted for mediational and direct models (the full model could not be analyzed due to limited degrees of freedom where df=0). FIML was not employed for these analyses, as the multisample comparison depends upon comparing configural and constrained results and the constrained models could not be conducted with FIML estimation. Table 5 demonstrates findings from configural and constrained models using this smaller subset of the larger model. Results indicate that the mediational model fits the data better than the direct model, evidenced by CFI indices=0.95 and RMSEA indices close to 0.10 ( Figures 5 and 6). In order to assess for statistical differences in the mediational model fit between men and women, degrees of freedom and chi square values from the constrained model (pathways assumed to be invariant) and the configural model (pathways freely estimated) were compared. This chi square difference test demonstrated that there were no statistically significant differences in mediational model fit between men and women. Further, the effect sizes for both men and women were small. Taken together, the larger model seemed to fit men and women well, with a slightly better fit for women. There was no statistically significant difference in model fit between men and women in the smaller subset of the larger path analysis.  In the large, full model, sociopolitical voice seemed to have the strongest path coefficients to meaning in life (.14), self-efficacy (.13), and well-being (-.15) compared to civic and electoral forms of engagement, although all three types of engagement showed relatively small parameter estimates to meaning, self-efficacy, and well-being. However, in the smaller subset mediational model, the path coefficient between civic activity and well-being was strongest [.20(W), .21(M)], perhaps because civic activity comprises elements like group membership that could be construed as more social involvement, and social support was not accounting for any of the variance in this smaller model. Contrary to the original hypothesis, findings show that electoral activity did not demonstrate a negative relationship with well-being; further, in the larger model, civic activity actually had the weakest parameter estimate to wellbeing while sociopolitical activity and well-being were inversely associated.

Hypothesis Three: Meaning in Life and Self-Efficacy as Mediators
This study had two proposed layers of mediation in the larger model: Types of civic engagement followed by meaning in life and self-efficacy. Results indicated that meaning in life partially mediated the relationship between SES, sociopolitical activity, social support, and well-being. Self-efficacy partially mediated the relationship between SES, sociopolitical and civic activity, social support, and wellbeing. When the smaller path model was used to examine this second mediational model, only civic activity appeared to mediate the relationship between SES and wellbeing, and the effect sizes and path coefficients were small.

This study investigated several factors to elucidate for whom, in what ways,
and under what circumstances civic engagement can be used to bolster health and well-being. Prior literature has shown that while civic engagement is generally associated with positive health outcomes for individuals, those from lowersocioeconomic backgrounds and women may not reap the same benefits (Brown et al., 2017;Ehsan & De Silva, 2015). Further, the literature has shown that perhaps there are mediational factors attenuating the relationship between civic engagement and well-being, or it may be likely that certain typologies of engagement more strongly predict well-being than others (Brown et al., 2017;Gullan et al., 2013;Peterson et al., 2005); Piliavin & Siegl, 2007). The present study assessed the relationship between civic engagement and well-being while accounting for these complex dynamics.
Further, the present study investigated these dynamics through recruitment of a young adult samplea developmental period known to be crucial for social and political identity formation (Amna, 2012;Verba et al., 1995). Using data from a cross-sectional design with undergraduate students, findings indicated that 1) the larger mediational model best fit the current sample slightly better for women than men, although there was no statistical difference in model fit between men and women for the smaller model; 2) meaning in life and self-efficacy were found to mediate relationships between various types of engagement and well-being, while civic activity mediated the relationship between SES and well-being in the smaller model; and 3) electoral activity was not negatively associated with well-being, sociopolitical activity was negatively associated with well-being, and civic activity demonstrated mixed results.
Given prior literature examining how contextual factors like gender and socioeconomic status might influence the relationship between civic engagement and well-being, findings from the current study provide some clarification in addressing who might benefit from civic engagement. On average, women reported experiencing more meaning in life than men and higher civic activity compared to men. However, the tested models did not indicate that these differences resulted in incongruent model fit, and hence, incongruencies in the way in which men's and women's civic engagement may influence their ultimate well-being. If anything, the larger model indicated a slightly better fit in the women's sample compared to the men's sample.
These results run counter to prior literature, which has found that men and women may participate in different types of engagement, and hence, they may experience different levels of well-being as a result (Landstedt et al., 2016). For example, Landstedt et al. (2016) noted that women often report participating in more emotionally burdensome activities, which could lead to poorer mental health outcomes. Given the mediational effects of meaning in life and self-efficacy in the current study, it is plausible that women are still engaging in burdensome activities; yet, if they are able to feel a sense of purpose and agency when performing these acts, a null or positive increase in well-being may result. Further, women reported higher levels of civic activity than men, and civic activity encompasses activities like belonging to community organizations and acts of volunteerism. While certainly not all community organizations and volunteer experiences can be construed as burden-free, perhaps the women in this sample are choosing to engage in less-burdensome activities, resulting in a null or positive change in well-being.
Similarly, Ehsan & De Silva (2015) have highlighted the impact of socioeconomic status on ability to participate among those already feeling overburdened. The current findings are consistent with previous literature regarding individuals from lower SES communities engaging less than those from higher SES communities (Verba et al., 1995). In this sample, path coefficients between SES and civic, electoral, sociopolitical engagement and perceived social support were statistically significant and positive, which suggests that higher-SES individuals were reporting higher levels of engagement and social support while lower-SES individuals reported lower levels. However, for those who do choose to engage, the relationship between engagement and well-being appears to be a positive one when meaning in life and self-efficacy serve as mediators. Taken together, these study findings suggest that women's civic engagement is not associated with lower well-being contrary to some prior literature, nor is civic engagement necessarily ill-advised for those from low-SES communities. However, more research is needed to better understand why someone from these marginalized identities may choose to engage or abstain from involvement, and the mechanisms in which that engagement may or may not impact their mental health.
This study explored some of the ways and circumstances in which civic engagement may influence well-being, specifically by examining meaning in life and self-efficacy as mediators. In this sample, the full effects model explained the data particularly well in women only, men only, and in women and men combined.
Interestingly, meaning in life only mediated the relationship between sociopolitical activity and well-being and social support and well-being, such that greater sociopolitical activity and social support resulted in greater meaning in life, which then related to stronger well-being. Self-efficacy was also a partial mediator, where civic activity, sociopolitical activity, and social support showed positive parameter estimates with self-efficacy, which in turn resulted in a significantly positive parameter estimate to well-being. Thus, both meaning in life and self-efficacy seem to play an important role in mediating the relationship between engagement and wellbeing.
Of note, sociopolitical voice demonstrated a significantly negative parameter estimate with well-being in the larger model, although pathways to meaning in life and self-efficacy were significantly positive. It could very well be that individuals who make concerted efforts to engage socio-politically may become disillusioned if they are unable to see the fruits of their labor. Oftentimes, sociopolitical activities like marching and calling congress people require significant emotional energy and time.
By comparison, civic activity like volunteering at a local soup kitchen or fundraising money during a bike-a-thon may produce instantaneous good will, as was the direct relationship between civic activity and well-being in the smaller path model. Thus, it is important to consider what drives people to engage in the first placeare they participating out of internal conflict or desire to change policy (as is likely with sociopolitical engagers), to reap potential material or social benefits that may coincide with civic forms of participation, or perhaps some combination of both (Verba et al., 1995). If the motivational drive is due to conflict, perhaps meaning in life and self-efficacy become that much more central to attenuating the relationship between civic engagement and well-being. For college students still in the process of developing their social and political identities, interventions might prompt young adults to reflect There are several limitations of the current study that should be noted. As the study employed a cross-sectional design, the relationships shown in the models are correlational rather than causal in nature; hence, it is plausible that those who experience higher levels of well-being may also civically engage more frequently. The proposed predictive models were based on a priori information from previous literature on civic engagement and health factors, but the relationships are complex and not easily interpretable. Longitudinal designs could help to bolster some of the preliminary findings noted above. Further, this research was conducted in a sample of predominantly White, female undergraduates. While the demographic breakdown of this sample does not significantly differ from the URI student body, students who have reached the point of a college education already experience certain privileges and perhaps have certain qualities that may make them more or less likely to civically engage. As an example, the mean and mode for self-efficacy toward service was a score of 20 out of a possible 25. These were individuals who may have already felt capable of enacting change in their surrounding environment, regardless of whether they actually decided to engage in that change. For these reasons, these findings may not be generalizable to other young adults. So it is with caution that these results be used to suggest to women and those from lower SES communities that civic engagement would be to their benefit-indeed, it might be, but further information is required before making such a claim.
It should also be noted that sample size, missing data, and normality were problematic in the current study.  (2) 3A. Do you belong to or donate money to any groups or associations, either locally or nationally such as charities, labor unions, professional associations, political or social groups, sports or youth groups, and so forth?
o o Active member of at least one of them (2) Member, but not active in at least one of them (1) Given money only (1) No (0) 3B  (2) Yes, But not within the last 12 months (1) No, Never (0) 10. Have you given money to a candidate, political party, or organization that supported candidates?

o o o
Have you done any of the following to express your views?
Yes, Within the last 12 months (2) Yes, But not within the last 12 months (1) No, Never (0) 11. Contacted or visited a public official-at any level of government-to express your opinion? Please use the following scale to respond to each item. Circle the letter or letters that best describe the extent to which you agree or disagree with the statement.  (Lamers, Westerhof, Bohlmeijer, ten Klooster, & Keyes, 2010) Place a check mark in the box that best represents your experiences and feelings.