The Contingent Effects of Mental Well-being and Education on Volunteering

Abstract

Mental health or well-being provides individuals with an enhanced agentic capacity for formal volunteering. However, this capacity may be realized more effectively through the structural resources for volunteering provided by education. Analyzing white respondents from the 1995-2005 National Survey of Midlife Development Panel Study (N = 1,431), we examine the contingent effects of mental well-being and education on the probability of formal volunteering. In longitudinal models, we find that mental well-being has a role in volunteering only at higher levels of education and that deficits in mental well-being reduce the effect of education. This holds across three representative indicators of mental well-being capturing the absence of negative symptoms as well as the presence of thriving or positive emotion. As a whole, our findings suggest that agency and structure are intertwined in determining who volunteers.

Keywords

distress mental well-being education formal volunteering

Volunteer work is a vital form of civic engagement that contributes to social capital and community well-being more generally (Putnam 2000; Wilson 2012). Because volunteer work helps ensure the continued thriving of communities and nations, scholars across disciplines seek to understand why people volunteer or do not. Some research finds that physical or mental well-being is linked to higher levels of volunteering, suggesting that well-adjusted individuals are more likely to select into and persist at volunteering opportunities (Helliwell and Putnam 2004; Li and Ferraro 2006; Snyder and Omoto 2008; Son and Wilson 2012; Thoits 2003; Thoits and Hewitt 2001; Wilson 2012). Whether measured in terms of positive emotion, the absence of negative emotion or depressive symptoms, or psychological characteristics such as a sense of control or purpose in life, mental well-being contributes to an agentic capacity for productive activity or helping others, making it a widely researched antecedent of volunteer activity (Li and Ferraro 2005; Piliavin and Siegl 2007; Son and Wilson 2012; Thoits and Hewitt 2001).

Rather than being a factor to be explained away or ruled out, agency is quintessentially important to volunteer activity (Thoits and Hewitt 2001). Net of structural or sociodemographic constraints on volunteering, individuals vary markedly in their happiness, depression, satisfaction with life, resilience, motivations to help, empathic concern, and other social-psychological characteristics that indicate differences in agentic capacity (Bekkers 2005; Penner 2002; Snyder and Omoto 2008; Thoits and Hewitt 2001; Wilson 2012). Rarely considered, however, is the confluence of agency and structure in producing voluntary labor. This is a vital oversight given that the choice to volunteer is an exceedingly complex process that is reducible neither to lived individuality nor structural determinism alone (Penner 2002; Thoits and Hewitt 2001). More broadly, social action represents the commingling of agency and structure (Adkins and Vaisey 2009; Sewell 1992).

In this study, we merge agency and structure in the study of voluntary labor by investigating how mental well-being interacts with educational resources. Prior work focused on the agentic potential of mental well-being has overlooked whether capitals or resources may be important to realizing the benefits of mental well-being for civic engagement (Wilson 2012). A volunteer capital perspective has long focused on social, cultural, and human resources as prime determinants of who volunteers and on education in particular as a dominant resource for initiating and sustaining formal volunteering (Bekkers 2010; Musick and Wilson 2008; Smith 1994; Wilson and Musick 1997a). Education produces resources as well as dispositions for volunteering, and highly educated individuals have more social connections and are more likely to be recruited as volunteers (e.g., Brand 2010; Dee 2004; Horowitz 2015; Oesterle, Johnson, and Mortimer 2004). In fact, some scholars have concluded that education is the strongest predictor of who volunteers in Western societies (Putnam 1995; Wilson 2012).

Building on this perspective, we argue here that the enhanced capacity for volunteering provided by mental health or well-being (Thoits 2003; Thoits and Hewitt 2001) may be realized more effectively through educational resources. This important notion remains overlooked by research on volunteering and mental health, which treats education either as an independent form of volunteer capital or merely as a demographic variable to be controlled. Instead, we contend here that mental health produces substantially different rates of volunteering depending on the level of education held by potential volunteers. This occurs because higher education creates structural, dispositional, and social conditions that realize the capacity of well-being to promote volunteering.

Background

Formal volunteer work is unpaid work that is freely chosen, deliberate, extends over time, is “engaged in without expectation of reward or other compensation and often through formal organizations,” and is “performed on behalf of causes or individuals who desire assistance” (Snyder and Omoto 2008:3). Volunteer work conducted through organizations is clearly only one aspect of volunteering (Wilson and Musick 1997a); many individuals also volunteer their time informally in activities such helping a friend move or taking an elderly neighbor to the doctor. Here, we focus on formal volunteering given how it is directly relevant to community life and the public good and because education is less consequential for informal helping or volunteering (Brady, Verba, and Schlozman 1995; Brand 2010; Wilson and Musick 1997a).

According to a volunteer process model (Omoto, Snyder, and Hackett 2010; Snyder and Omoto 2008), formal volunteer activity can be understood in terms of antecedents, experiences, and consequences of volunteering. Traits, motives, and personal characteristics serve as important antecedents of who volunteers (Omoto et al. 2010; Thoits and Hewitt 2001). While research has long recognized that improved mental well-being may be a consequence of volunteering, supporting a social causation perspective on volunteering and mental health (e.g., Piliavin and Siegl 2007; Thoits and Hewitt 2001; Wilson 2012), mental well-being is likely to facilitate volunteering as well, supporting a social selection perspective (Helliwell and Putnam 2004; Li and Ferraro 2005; Thoits 2003; Thoits and Hewitt 2001).

Although prior work has examined differing mental well-being selection and causation processes due to life-course factors such as age or social integration (Kim and Pai 2010; Li and Ferraro 2006; Omoto et al. 2010; Piliavin and Siegl 2007), other resources remain overlooked. Because education is one of the strongest human resources or capitals for volunteering (Wilson 2012), mental health may more effectively lead to volunteering given higher levels of education. Education is a “resource that lowers the costs of doing volunteer work” through multifaceted resource pathways (Dee 2004; Musick and Wilson 2008:124). While mental health provides a broad capacity for daily cognitive and social functioning that enhances agentic capacity (Thoits 2003; Thoits and Hewitt 2001), this foundation may not be optimal for volunteering unless accompanied by the cognitive and civic skills and non-kin social inducements linked to education.

However, education remains unexamined as a pivotal translator of mental well-being to voluntary action. The precise nature or extent of this potentially vital interplay between agency and structure has yet to be demonstrated. Sociological theory and research have long maintained that agentic capacity or potential depends on structural resources and opportunities in order to realize outcomes (Adkins and Vaisey 2009; Sewell 1992; Thoits 2003). Moreover, while scholars of volunteering and related civic activity maintain that volunteering reflects a confluence of agency and structure (Helliwell and Putnam 2004; Thoits and Hewitt 2001; Wilson 2012), they rarely implement this core insight.

Volunteer Capitals: Education and Formal Volunteering

A focus on mental health selection and causation in studies of formal volunteering has yet to benefit from the core insights afforded by volunteer capital theory. In summarizing a capital-based approach to understanding who volunteers, Musick and Wilson (2008:6) state: “people cannot volunteer if they lack the individual resources needed, they will not volunteer if they are not disposed, or motivated to do so, and they are unlikely to volunteer if nobody asks them to do so.” These precursors to volunteering are distilled from abundant research on volunteering, which illuminates who volunteers in terms of personal and socioeconomic resources, individual motivations, dispositions, values, or identities and also in terms of social capital and the inducements and norms of voluntary labor it promulgates (e.g., Bekkers 2005; Brady et al. 1995; Freeman 1997; Gesthuizen and Scheepers 2012; Thoits and Hewitt 2001; Wilson and Musick 1997a).

With respect to these pillars of volunteering—resources, dispositions, and recruitment through social networks—education figures prominently. Indeed, a volunteer capital perspective has long focused on education and in particular on the numerous and diverse resources for volunteering that education provides, offering a multifaceted explanation for its very strong and robust association with volunteering. With regard to resources, higher education channels individuals into jobs that are more autonomous, higher in status, and more encouraging of occupational self-direction, all of which may enhance the tendency to volunteer (Wilson and Musick 1997b). Furthermore, education contributes to cognitive competence (Dee 2004; Gesthuizen and Scheepers 2012; Hauser 2000) and allows for the development of a number of civic and problem-solving skills, such as effective communication, information gathering, management of organizational funds and resources, and the ability to run meetings (Brand 2010; Hout 2012; Mirowsky and Ross 2003), that facilitate volunteering.¹

Regarding dispositions to volunteer, a primary mission of educational institutions is to develop good citizens (Brand 2010; Dewey 1916). As such, longer exposure to education results in greater orientation toward volunteering through the socialization of norms and values oriented toward helping others (Gesthuizen and Scheepers 2012). This orientation also includes greater awareness of social issues and more cosmopolitan attitudes (Musick and Wilson 2008). Network-based mechanisms involve the strong association between higher education and the number and heterogeneity of non-kin social ties (Fischer 2011; Huang, van den Brink, and Groot 2009; Musick and Wilson 2008) and in turn the strong link between larger and more heterogeneous networks and being asked or expected to volunteer or participate civically (Paik and Navarre-Jackson 2011; Wilson 2012; Wilson and Musick 1997a).

Here, we argue that whether mental well-being promotes volunteering may depend on educational resources. Education provides structural conditions for volunteering by furnishing skill, dispositional, and network-based mechanisms that promote volunteering. In turn, mental well-being enhances the agentic capacity for volunteering. While mental well-being should lower the opportunity costs and heighten the personal rewards of voluntary action in general (Bekkers 2005, 2010), we expect this effect to be manifested to a greater extent among the highly educated because they have more opportunities and inducements to volunteer than those with lower levels of education.²

Method

To investigate whether the link between mental well-being and formal volunteering differs according to educational resources, we use the 1995 and 2005 National Survey of Midlife Development in the United States (MIDUS) panel survey. The panel design allows us to better establish temporal ordering between volunteering and well-being by observing change in voluntary activity across one decade. In addition, it offers multiple indicators of mental well-being capturing distress or negative emotion as well as the presence of thriving or positive emotion, enabling us to examine whether interaction effects between education and well-being are robust across these indicators.

The main component of the 1995-2005 MIDUS panel is a probability sample consisting of English-speaking, noninstitutionalized adults aged 25 to 75 residing in the contiguous United States (referred to herein as the random-digit-dial or RDD sample), with an oversample of individuals aged 65 to 74 and males. In 1995, about 70 percent of initially chosen RDD informants agreed to a phone interview that gathered basic information. Eighty-seven percent of this phone-response sample then also agreed to complete additional questionnaires, for an overall response rate of 61 percent, which is quite similar to other national surveys utilizing similar designs (Braveman and Barclay 2009; Morton, Mustillo, and Ferraro 2014). In 2005, 2,038 of the original respondents in 1995 (69.5 percent) completed a telephone interview, and 83 percent of these individuals also completed additional questionnaires required for the analysis. Once this level of attrition is corrected for rates of mortality typically seen among older individuals, it is comparable to panel response rates observed in other national surveys (see Braveman and Barclay 2009; Schafer, Wilkinson, and Ferraro 2013).

Following prior work on formal volunteering, we restrict our analyses to white respondents. Several key mechanisms of formal volunteering pertaining to social class and religious attendance may differ fundamentally across racial groups (Musick, Wilson, and Bynum 2000), thus potentially complicating the analysis and interpretation of heterogeneous effects of mental well-being or education. In addition, only about 5 percent of respondents in the MIDUS panel elected a non-white racial classification.³

Upon restricting our sample to white respondents who completed the MIDUS panel, N = 1,579 observations are available. While no key covariate had greater than 10 percent nonresponse, some respondents had missing data on one or more key variables for the analysis, resulting in a listwise sample of N = 1,431. Because diligent survey participation arguably is a form of volunteering (Radler and Ryff 2010), we take attrition into account explicitly by utilizing an inverse weighting procedure (Cornwell and Laumann 2015).⁴

Dependent Variable: Any Volunteering in 2005

In 1995 and again in 2005, MIDUS respondents were asked about how often they engage in formal volunteer work: “On average, about how many hours per month do you spend doing formal volunteer work of any of the following types?”: (a) hospital, nursing home, or other health care–oriented work; (b) school or other youth-related work; (c) political organizations or causes; (d) any other organization, cause, or charity. We constructed a binary measure of any volunteering during the past year (1 or more hours for any type of volunteering). In additional models, we also analyzed total monthly hours of volunteering.

Independent Variables

Education

MIDUS surveys education mainly in terms of credential points (i.e., completion of high school, GED, associate’s degree, bachelor’s, master’s, or doctorate), with degree midpoints also present (e.g., 1 or 2 years of college, no degree yet). We recoded this measure to years of education (10 to 20 years). Conclusions do not differ when we measure education as a dichotomous variable indicating a four-year college degree and above versus less education (e.g., Brand 2010).

Mental Well-being

We employ three measures of mental well-being that are representative of measures employed in other longitudinal studies of volunteering (Li and Ferarro 2005; Son and Wilson 2012; Thoits and Hewitt 2001): psychological distress, positive emotion, and psychological well-being. In keeping with prior work, we treat these scales separately, as they are conceptually and empirically distinct manifestations of well-being.⁵

First, MIDUS respondents completed the Kessler K-6 scale of nonspecific psychological distress, which has been used internationally for screening current and prospective mental illness risk (Kessler et al. 2002). Distress has been used widely as a general indicator of mental health as well (e.g., Bierman and Kelty 2014; Caputo and Simon 2013; Kessler 1982). Respondents were asked “During the past 30 days, how much of the time did you feel . . . ” nervous, restless or fidgety, hopeless, worthless, that everything was an effort, and so sad that nothing could cheer you up (5 = all of the time, 4 = most of the time, 3 = some of the time, 2 = a little of the time, 1 = none of the time; alpha >.85). After averaging across all six distress items to obtain a distress score, we took the natural logarithm to help correct for skewed scores (most respondents report low levels of distress).

Next, given how the absence of distress does not equate with the presence of thriving, we focused on positive affect or emotion. MIDUS assesses positive affect during the past month using a six-item scale, as validated in prior work (Diener and Chan 2011; Tellegen, Watson, and Clark 1999). Respondents were asked “During the past 30 days, how much of the time did you feel . . . ” cheerful, in good spirits, extremely happy, calm and peaceful, satisfied, and full of life (5 = all of the time, 4 = most of the time, 3 = some of the time, 2 = a little of the time, 1 = none of the time; alpha = .91). As for distress, we averaged responses to these items.

Finally, we considered a more comprehensive measure of psychological well-being, based on Ryff and Keyes’s (1995) approach developed in the MIDUS data. This measure consists of six subscales: autonomy, environmental mastery, personal growth, positive relations with others, purpose in life, and self-acceptance. Each of the six subscales is based on three items, which are presented in the appendix. Because follow-up work has documented high or extremely high latent correlations among the subscales (Springer and Hauser 2006), we averaged all six subscores to obtain an overall well-being score (subscore alpha = .76).⁶

Sociodemographic Covariates

In all equations, we include basic demographic and social covariates that are commonly controlled in analyses of volunteering (Bekkers 2010; Li and Ferraro 2005; Piliavin and Siegl 2007; Son and Wilson 2012; Wilson and Musick 1997a). Gender is measured as an indicator for male; age is measured in years and specified as age groupings in models to allow for nonlinearities.⁷ Marital status is assessed by whether the respondent was currently married at the time of the interview. Marital involvement carries strong implications for non-kin social relations (Fischer 2011; Rotolo and Wilson 2006). Further, we account for whether the respondent has children, as childrearing is linked to distinctive patterns of social and community involvement (Rotolo and Wilson 2006). We distinguish between children’s age groups (i.e., 0-6 years old vs. 7-17) in order to separate young children from those who are likely to leave the household (e.g., for college) between 1995 and 2005.⁸ Models that instead focused on number of children in the household did not alter the findings presented here.

Labor force status is assessed by whether the respondent works full- or part-time for pay. In addition, we control for work hours per week at the respondent’s main job in order to account for time-related job demands and effort that may restrict one’s availability for voluntary labor. Because a handful of respondents reported an extremely high number of work hours per week (>70 hours), we take the natural logarithm of work hours for our analyses. Next, controlling for self-rated physical health (1 = poor to 5 = excellent) helps reduce any omitted variable bias due to functional limitations that seriously constrain one’s ability to engage in voluntary labor (Helliwell and Putnam 2004).⁹ Occupational prestige is measured using the Duncan Socioeconomic Index (SEI) score; if the respondent is not currently working, previous SEI is used. Household income for the past year (logged) is used to distinguish effects of education from those of financial capital.

Next to education, religious attendance is one of the most potent predictors of volunteering and is among the covariates typically included in analyses of volunteering. MIDUS assessed attendance of religious or spiritual services on a 5-point ordinal scale (1 = once a day or more, 2 = a few times a week, 3 = once a week, 4 = 1-3 times/month, 5 = less than once/month). This ordinal format was rescaled to days per year (logged for analyses).

Parental socioeconomic status (SES) is related both to educational attainment and life-course mental health (Bauldry 2015; Schafer et al. 2013). It also may have persistent effects on civic participation due to primary socialization experiences and ongoing inducements from living parents to participate in community affairs (Brand 2010). We take into account parental SES by way of education in years and occupational SEI (maximum of mother’s and father’s values or the available value when one is missing).

Analytic Strategy

For our volunteering outcome (any volunteering during the past year), we fit a series of logistic regression models.¹⁰ In the models, covariates and the lagged volunteering outcome are observed in 1995. A lagged dependent variable (LDV) helps limit endogeneity due to any improvements in psychological well-being brought about by volunteering or other civic activity (Musick and Wilson 2003; Putnam 2000). The LDV specification treats mental well-being in 1995 as a static variable, which is reasonable given the moderate stability of well-being (Frech and Williams 2007; Musick and Wilson 2003; Thoits and Hewitt 2001).¹¹

Because education and other determinants of voluntary labor are related to survey participation, we weight MIDUS models to help correct for attrition from 1995 to 2005. To calculate attrition weights, we estimated a probit equation of follow-up participation in 2005 on all variables observed in 1995. Based on this equation, we generated probabilities for all observations and took the inverse so that individuals least likely to participate in 2005 receive the greatest weight. While by no means a full correction, this helps adjust parameter estimates to simulate what would have been observed had all respondents been retained (e.g., Cornwell and Laumann 2015). Under weighting, standard errors are robust, and traditional nonlinear model fit statistics such as chi-square or pseudo R² are inapplicable. Instead we present F-ratios with corresponding p values (e.g., Cui et al. 2012). Results do not differ without this attrition correction.

For each mental well-being indicator, we estimate two logistic regressions of volunteering. In the first model, volunteering in 2005 is regressed on the well-being indicator and all covariates in 1995. In the second model, an education × well-being interaction term is added to the equation to test the hypothesis that education and mental well-being interact to determine volunteer activity. Both models use mean-centered values for the education and well-being terms and include the lagged observation of volunteer status from 1995. Simpler models using only a limited set of demographic covariates (male, age, parental SEI, and parental education) revealed larger main effects of education and mental well-being on volunteering probability in 2005, which is consistent with partial mediation by control variables, but they produced the same findings regarding interaction effects involving education and mental well-being (available on request). In visualizing our results, we focus on how the association between volunteering and mental well-being varies according to level of education. This places the analytic focus on whether and to what extent structural resources conveyed by education enhance the agentic capacity of mental well-being.¹²

Results

Descriptive statistics are summarized in Table 1. Forty-five percent of respondents volunteered in 1995 and 48 percent in 2005.¹³ In terms of education, respondents averaged some college (about 14 years). Respondents experienced psychological distress a little more often than none of the time, on average.¹⁴ Levels of positive emotion and psychological well-being were moderate. Demographically, respondents were 46 percent male and about 48 years old on average at baseline. Respondents tended to be married, working, and to attend church two to three times per month.

Table 1.

Descriptive Statistics: 1995-2005 National Survey of Midlife Development in the United States (MIDUS Panel).

Variable	M	SD	Minimum	Maximum
Is volunteer (1995)	0.45	0.50	0	1
Is volunteer (2005)	0.48	0.50	0	1
Education (years)	14.21	2.48	10	20
Psychological distress	1.53	0.59	1	5
Positive affect	3.38	0.71	1	5
Psychological well-being	16.67	2.30	7.67	21
Male	0.46	0.50	0	1
Age	48.05	12.68	25	74
Married	0.69	0.46	0	1
Child: 6 or younger	0.15	0.35	0	1
Child: 7 to 17 years	0.29	0.46	0	1
Working	0.62	0.49	0	1
Work hours/week	31.21	21.37	0	168
Self-rated health	3.58	0.93	1	5
Occupation (SEI)	40.09	14.34	7.13	80.53
Household income	73.03	60.56	0	300
Church attendance (days/year)	29.94	40.36	0	360
Parent education (years)	12.84	2.62	10	20
Parent occupation (SEI)	38.45	13.49	9.56	80.53

Note. Ns = 1,487 to 1,579 (white respondents aged 25 or older). All MIDUS panel variables are observed in 1995 except for follow-up volunteering outcome (2005). SEI = socioeconomic index.

Psychological Distress

The logistic regression results are summarized in Table 2. Because these regressions model any volunteering in 2005 controlling for a lagged (1995) observation of the dependent variable, they estimate volunteering in 2005 holding prior volunteer activity constant. Unstandardized beta coefficients are shown.

Table 2.

Logistic Regression Models of Volunteering.

Variable	1	2	3	4	5	6
Education (years)	0.114**	0.112**	0.113**	0.109**	0.111**	0.108**
Log psychological distress	–0.318	–0.355+
Positive affect (PA)			0.155*	0.157*
Psychological well-being (PWB)					0.068**	0.071**
Education × distress		–0.214*
Education × PA				0.058*
Education × PWB						0.022*
Male	–0.354**	–0.354**	–0.343*	–0.345*	–0.366**	–0.362**
Age: 25 to 34	–0.001	0.002	–0.052	–0.057	–0.057	–0.06
Age: 35 to 44	–0.004	0.001	–0.005	–0.016	–0.03	–0.039
Age: 55 to 64	–0.024	–0.034	–0.034	–0.039	–0.035	–0.043
Age: 65 or higher	–0.132	–0.13	–0.157	–0.126	–0.141	–0.136
Parental education	–0.026	–0.028	–0.027	–0.028	–0.025	–0.027
Parental SEI	0.012+	0.013*	0.012+	0.013*	0.012+	0.012+
Married	0.145	0.145	0.144	0.154	0.142	0.14
Child: 6 or younger	1.052**	1.039**	1.086**	1.083**	1.114**	1.112**
Child: 7 to 17 years	–0.669**	–0.680**	–0.669**	–0.660**	–0.650**	–0.645**
Working	0.267	0.271	0.282	0.282	0.282	0.288
Log work hours	–0.014	–0.011	–0.015	–0.01	–0.012	–0.015
Self-rated health	0.126+	0.139+	0.118	0.128+	0.127+	0.130+
Respondent SEI	0.014*	0.014*	0.013*	0.014*	0.013*	0.013*
Log household income	0.08	0.078	0.081	0.081	0.07	0.071
Log religious attendance	0.122**	0.118**	0.116**	0.117**	0.129**	0.129**
Is volunteer (1995)	1.736**	1.752**	1.765**	1.763**	1.729**	1.732**
Constant	–3.150**	–3.187**	–3.116**	–3.182**	–3.018**	–3.048**
N	1,431	1,431	1,429	1,429	1,431	1,431
F	15.54	14.76	15.74	15.00	15.54	14.99
p	<.0001	<.0001	<.0001	<.0001	<.0001	<.0001

Note. Unstandardized beta coefficients are shown. Standard errors are robust. Education and well-being variables are mean-centered. Observations are weighted by attrition probability. SEI = Socioeconomic Index. +p < .10, *p < .05, **p < .01 (two-tailed significance tests).

In Model 1, education is linked to a higher probability of volunteering in 2005, with each additional year of education predicting a 12 percent increase, all else equal (odds ratio [e ^b] = e ^0.114 = 1.121; p < .01). Meanwhile, psychological distress shows a negative though nonsignificant link to prospective probability of volunteering (logged; OR = 0.728, ns). A number of covariates show significant links to probabilities of volunteering. Men are less likely to volunteer, and respondent SEI shows a significant positive link to volunteering (ps < .05). Parental SEI and self-rated health also positively predict volunteering (p < .10), as does religious attendance (OR = 1.130, p < .01). Children have countervailing effects on volunteering probability, with having a child under age 7 positively predicting volunteering one decade later, while having an older child negatively predicts volunteering.

Model 2 adds an education × distress term to the equation, which is both large and statistically significant (b = –0.214, p < .05). Figure 1 visualizes this interaction effect by showing the contingent effect of psychological distress. Here, volunteering predictions are generated separately by level of education (less than college degree vs. college degree or higher). Distress has no discernible impact on volunteering probability among individuals without a college degree (slope p = .477). However, among college graduates reporting no or mild distress, predicted probabilities of volunteering are significantly higher (0.59 for college graduates reporting no distress compared to 0.44 for non-graduates; 95 percent confidence bands shown in figure). Educational group differences in volunteering are present at levels of distress below the 75th percentile (i.e., about <1.65 on 5-point distress scale; see 95 percent confidence bands in figure). In sum, mental well-being in terms of lack of distress increases the odds of voluntary labor but only given higher education.

Figure 1.

Contingent effects of psychological distress and education on volunteering in 2005.

Positive Affect

Results for positive affect are presented in Models 3 and 4. In Model 3, education shows a positive association to volunteering in 2005 (OR = 1.120, p < .01), as does baseline positive affect (PA) (OR = 1.167, p < .05). In Model 4, an education × PA term is both large and significant (b = 0.058, p < .05). Figure 2 depicts this interaction term, again by focusing on contingent effects of positive affect. Increasing positive affect produces a modest but nonsignificant gain in predicted probability of volunteering among individuals with less than a college degree (probability gain across entire domain = 0.04; slope p = .292). Among individuals with a college degree or higher, the gain is both much larger and statistically significant (probability gain = 0.18; slope p = .002). Educational group differences emerge beginning at levels of positive affect above the 36th percentile (i.e., about >3.25 on 5-point positive affect scale). As for low distress, positive affect significantly enhances the odds of volunteering given higher levels of educational resources.

Figure 2.

Contingent effects of positive affect and education on volunteering in 2005.

Psychological Well-being

Finally, results for the psychological well-being (PWB) composite measure are presented in Models 5 and 6. Model 5 reveals a main effect of years of education on volunteering in 2005 (OR = 1.118, p < .01) as well as a significant positive effect of the PWB composite (OR = 1.070, p < .01). Model 6 in turn reveals a strong education × PWB interaction (b = 0.022, p < .05). Figure 3 visualizes this contingent effect. Here, psychological well-being is linked to higher probabilities of volunteering among college graduates (probability gain = 0.26; slope p < .001) than among adults with less than a college degree (probability gain = 0.05; slope p = .137). Educational group differences in volunteering are present beginning at levels of psychological well-being just below the median score (i.e., >16 on PWB scale, 44th percentile). This corroborates the other findings, such that mental well-being is linked to greater rates of volunteering but only in the presence of higher education.

Figure 3.

Contingent effects of psychological well-being and education on volunteering in 2005.

Additional Results: Monthly Hours of Volunteering

Because distinctions in frequency of volunteering among those who volunteer at all are also important to understand, in further analyses, we modeled monthly volunteer hours in 2005 as a count outcome using negative binomial regression. We implemented a traditional negative binomial model as well as zero-inflated and zero-truncated models (Ajrouch, Antonucci, and Webster 2016; Son and Wilson 2012) in order to test for robustness to how non-volunteers (i.e., those reporting zero hours in 2005) are treated. For each mental well-being indicator, we used the same covariates and interaction terms as for our logistic regression models and also included monthly hours in 1995, estimating two models for each well-being indicator as described previously. While these results often revealed main effects for education or mental well-being, they did not reveal any education × well-being interaction effects for the count of volunteer hours. Thus, while our aforementioned logistic regression results reveal the importance of interaction effects to the probability of any volunteering, our negative binomial results did not suggest that education and mental well-being work in tandem to determine the precise extent or level of volunteering (available on request).

Discussion

Prominent perspectives on volunteering posit that mental well-being is both an antecedent and a consequence of volunteer activity (Omoto et al. 2010; Thoits and Hewitt 2001). While numerous studies have demonstrated this dual role of mental well-being, few have focused on the interplay of agency and structure as it relates to voluntary labor. Across three representative indicators of well-being, we find that mental well-being is linked to volunteering only at higher levels of education. Individuals who combine education and higher levels of well-being show the highest probabilities of volunteering. Our findings suggest that educational resources provide structural conditions for realizing the agentic capacity for civic work provided by well-being. Moreover, the similarity of our findings across three empirically and conceptually distinct indicators of mental well-being suggests that mental well-being in general—rather than in a specific form such as lack of psychological distress—conveys enhanced value for volunteering when coupled with educational resources.

By showing that mental well-being is an antecedent of volunteering only when coupled with higher levels of education, we provide evidence for a new mechanism by which volunteering occurs. Although selection into volunteering on the basis of mental well-being has been well established, heterogeneous selection on mental well-being due to education represents a new way of understanding antecedent conditions of volunteering and volunteer selection. By merging agency and structure, we provide a more complex view of the volunteer process that integrates prior insights from volunteer capital theory.

Our results also highlight the important distinction between deciding to volunteer at all and deciding exactly how much of one’s time to give. Most Americans do not volunteer formally. According to our results, the interaction between education and mental well-being is more pertinent to understanding the likelihood of volunteering at all rather than the precise level. A compelling reason for this difference could be that social-psychological factors such as sociability, pessimism, and efficacy are strongly related to mental well-being and more relevant to any volunteering rather than volunteering extent (see Forbes and Zampelli 2014).

In addition to shedding new light on mental health selection relevant to volunteering, our study illustrates a novel approach to heterogeneous effects of education. Education provides a broad set of structural conditions that promotes volunteering. Yet, these structural conditions may rely on certain facets of agency in order to translate effectively to voluntary labor. Prior work on heterogeneous effects focuses on conditions prior to educational attainment, under the guiding idea that college completion is related positively to early-life advantages and that those least likely to complete college in the first place may exhibit the greatest returns to education, in terms of health, income, or volunteering, for example (Brand 2010; Brand and Xie 2010; Schafer et al. 2013; but see Bauldry 2015). However, there has been no examination of heterogeneity in civic activity based on factors after attainment is complete. As Musick and Wilson (2008) propose, resources, dispositions, and network-based recruitment all are important pathways by which education enhances volunteering. Mental well-being is likely to impact the effectiveness of all three of these pathways, by shaping problem solving and situational coping (Connor-Smith and Flachsbart 2007; Mirowsky and Ross 2003), one’s felt obligation to volunteer or focus on social concerns (Connor-Smith and Flachsbart 2007; Mor and Winquist 2002), and also the odds of forming ties to potential recruiters or being asked to volunteer by those whom one already knows (Andersson 2012; Steptoe et al. 2008; Turner and Stets 2006).

Moreover, our results carry practical community implications, as one-third of the MIDUS panel sample reported some level of psychological distress.¹⁵ While not strictly a measure of depression, psychological distress taps the broad experience of negative emotion and is used routinely as a screening instrument for mental illness risk. The degree to which adult mental health or well-being may moderate the contribution of education to volunteering is particularly timely to assess in light of evidence that psychological distress may have increased during recent decades among American adults (Kessler et al. 2003; Klerman and Weissman 1989).¹⁶ While clearly an important public health issue in its own right, if this increase in distress has dampened the overall effect of education on volunteering and other forms of civic engagement, its societal impact is all the more consequential. If education relies on mental well-being to manifest its influence on volunteering, then increased education in an increasingly distressed population will not lead to greater volunteering or will not lead to the expected increase in volunteering. According to our results, psychological distress is negatively though not significantly associated with a lower subsequent probability of volunteering (similar to longitudinal null associations between depression and volunteering in Li and Ferraro 2005 and Thoits and Hewitt 2001). However, due to an interaction with education, our results concerning distress still clearly demonstrate the possibility that mental illness risk undermines civic participation in the United States.

While MIDUS has a number of strengths for uncovering contingent effects of mental well-being and education, it unfortunately does not permit the observation of mediating processes, nor does it provide an adequate sample for robust analyses of racial differences in contingent effects. Future research will need to convincingly show that educational mechanisms linked to civic participation are in fact enabled by general mental health as indicated by low distress and other well-being indicators. Likewise, although MIDUS provides temporal ordering between well-being and volunteering one decade later, it still does not illuminate the relative timing of changes in mental well-being and changes in volunteering. A variety of longitudinal designs might be implemented to gain greater insight into the issue of relative timing. Meanwhile, research should also draw on existing knowledge about racial differences in educational pathways to volunteering (e.g., Musick et al. 2000).

As an alternative explanation of our results, the observed interactions between education and mental well-being may represent heterogeneous gains in human capital during the schooling process. Although this is a mechanism prior to adulthood, it also would also be broadly consistent with our contingent effects argument in that individuals with higher mental well-being may be better able to derive skills and benefits from academic contexts (e.g., McLeod and Fettes 2007), resulting in better schooling and more robust contributions to civic engagement. Future research that enlists an early life-course panel design would be poised to distinguish differential acquisition of civic resources during attainment from the application of these same civic resources after attainment.

It remains to be seen whether distress or negative affect also dampens other civic returns to education. For instance, voting behavior in response to education may not be similarly moderated by mental health differences. Future research also should examine moderation effects for other civic capital. For instance, religiosity may be a form of civic capital that has multiple facets (e.g., public vs. private religiosity; prayer vs. belief; Lam 2002; Paxton, Reith, and Glanville 2014). Its influence on civic engagement may in turn be influenced by aspects of individual functioning, though not necessarily by mental health in particular. A variety of individual dispositions potentially are relevant (Bekkers 2005, 2010; Wilson 2012).

Another important area for future research is heterogeneous social causation processes relevant to mental health and volunteering. While we develop and test a heterogeneous social selection perspective in this study, prior work on causation has found that volunteering may yield mental health benefits especially for individuals who do not possess other forms of social integration in their lives (Piliavin and Siegl 2007). However, much more work remains to be done in this area. For instance, the mental health benefits of volunteering may also hinge on type of volunteering, personal values, and life-course factors such as age (Bekkers 2005; Piliavin and Siegl 2007). Gender also remains worthy of examination given that gender not only patterns the experience and reporting of mental well-being but also the educational, occupational, and familial commitments that individuals hold in society.

A final, interesting possibility for social causation research on volunteering concerns the stress process (Pearlin and Bierman 2013). It may be the case that individuals who are experiencing personal, familial, or occupational stress in their lives may benefit especially from what volunteering has to offer in terms of finding purpose in life and feeling integrated as part of a larger community. If this is true, then the stress process may further causation and selection research on volunteering. That is, ongoing social stressors may not only determine the well-being rewards of volunteering, which would be a form of heterogeneous social causation, but also may determine the very starting levels of well-being that individuals bring to volunteering for activating their educational resources, making stress relevant to heterogeneous selection as well.

While our results focus on the capacity for volunteering that mental well-being provides, they also speak to the potential of education to produce a civically engaged society. Education, one of the strongest forms of capital contributing to community well-being in America, does not uniformly produce voluntary labor. Another individual factor, mental well-being, seems pivotal to realizing civic returns to education. This raises the possibility that higher education without adequate mental health offers a less robust contribution to the public good.

Footnotes

Appendix: Psychological Well-being Items

Acknowledgements

Thanks to Freda Lynn, Sarah Harkness, Steven Hitlin, Vida Maralani, Brian Powell, and Mark Walker for their useful feedback. Earlier drafts of this article were presented at the University of Wisconsin–Madison Social Psychology Brownbag (SPAM) and at the 2013 Annual Meeting of the American Sociological Association.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was conducted during a Presidential Graduate Fellowship at The University of Iowa and a postdoctoral fellowship at the Center for Research on Inequalities and the Life Course (CIQLE) at Yale University. Neither ICPSR nor the MacArthur Foundation bears any responsibility for any results or interpretations herein.

Notes

References

Adkins

Daniel E.

Vaisey

Stephen

. 2009. “Toward a Unified Stratification Theory: Structure, Genome, and Status across Human Societies.” Sociological Theory 27:99-121.

Ajrouch

Kristine J.

Antonucci

Toni C.

Webster

Noah J

. 2016. “Volunteerism: Social Network Dynamics and Education.” Journals of Gerontology: Psychological Sciences and Social Sciences 71(2):309-19.

Andersson

Matthew A.

2012. “Dispositional Optimism and the Emergence of Social Network Diversity.” Sociological Quarterly 53:92-115.

Bauldry

Shawn

. 2015. “Variation in the Protective Effect of Higher Education against Depression.” Society and Mental Health 5:145-61.

Bekkers

René

. 2005. “Participation in Voluntary Associations: Relations with Resources, Personality, and Political Values.” Political Psychology 26:439-54.

Bekkers

René

. 2006. “Traditional and Health-related Philanthropy: The Role of Resources and Personality.” Social Psychology Quarterly 69:349-66.

Bekkers

René

. 2010. “Who Gives What and When? A Scenario Study of Intentions to Give Time and Money.” Social Science Research 39:369-81.

Bierman

Alex

Kelty

Ryan

. 2014. “The Threat of War and Psychological Distress among Civilians Working in Iraq and Afghanistan.” Social Psychology Quarterly 77:27-53.

Brady

Henry E.

Verba

Sidney

Schlozman

Kay Lehman

. 1995. “Beyond SES: A Resource Model of Political Participation.” American Political Science Review 89:271-94.

10.

Brand

Jennie E.

2010. “Civic Returns to Higher Education: A Note on Heterogeneous Effects.” Social Forces 89:417-33.

11.

Brand

Jennie E.

Xie

. 2010. “Who Benefits Most from College?: Evidence for Negative Selection in Heterogeneous Economic Returns to Higher Education.” American Sociological Review 75:273-302.

12.

Braveman

Paula

Barclay

Colleen

. 2009. “Health Disparities Beginning in Childhood: A Life-course Perspective.” Pediatrics 124:S163-75.

13.

Caputo

Jennifer

Simon

Robin W

. 2013. “Physical Limitation and Emotional Well-being: Gender and Marital Status Variations.” Journal of Health and Social Behavior 54:241-57.

14.

Connor-Smith

Jennifer K.

Flachsbart

Celeste

. 2007. “Relations between Personality and Coping: A Meta-analysis.” Journal of Personality and Social Psychology 93:1080-107.

15.

Cornwell

Benjamin

Laumann

Edward O

. 2015. “The Health Benefits of Network Growth: New Evidence from a National Survey of Older Adults.” Social Science and Medicine 125:94-106.

16.

Cui

Ming

Ueno

Koji

Fincham

Frank D.

Donnellan

M. Brent

Wickrama

K.A.S

. 2012. “The Association between Romantic Relationships and Delinquency in Adolescence and Young Adulthood.” Personal Relationships 19:354-66.

17.

Dee

Thomas S.

2004. “Are There Civic Returns to Education?” Journal of Public Economics 88:1697-720.

18.

Dewey

John

. 1916. Democracy and Education. New York: Macmillan.

19.

Diener

Chan

Micaela Y

. 2011. “Happy People Live Longer: Subjective Well-being Contributes to Health and Longevity.” Applied Psychology: Health & Well-Being 3:1-43.

20.

Fischer

Claude S.

2011. Still Connected. New York: Russell Sage Foundation.

21.

Forbes

Kevin F.

Zampelli

Ernest M

. 2014. “Volunteerism: The Influences of Social, Religious, and Human Capital.” Nonprofit and Voluntary Sector Quarterly 43:227-53.

22.

Frech

Adrianne

Williams

Kristi

. 2007. “Depression and the Psychological Benefits of Entering Marriage.” Journal of Health and Social Behavior 48:149-63.

23.

Freeman

Richard Barry

. 1997. “Working for Nothing: The Supply of Volunteer Labor.” Journal of Labor Economics 15:S140-66.

24.

Gesthuizen

Maurice

Scheepers

Peer

. 2012. “Educational Differences in Volunteering in Cross- national Perspective: Individual and Contextual Perspectives.” Nonprofit and Voluntary Sector Quarterly 41:58-81.

25.

Halaby

Charles N.

2004. “Panel Models in Sociological Research: Theory into Practice.” Annual Review of Sociology 30:507-44.

26.

Hauser

Seth M.

2000. “Education, Ability, and Civic Engagement in the Contemporary United States.” Social Science Research 29:556-82.

27.

Helliwell

John F.

Putnam

Robert D

. 2004. “The Social Context of Well-being.” Philosophical Transactions of the Royal Society London B 359:1435-46.

28.

Horowitz

Jonathan

. 2015. “Doing Less with More: Cohorts, Education, and Civic Participation in America.” Social Forces 94:747-74.

29.

Hout

Michael

. 2012. “Social and Economic Returns to College Education in the United States.” Annual Review of Sociology 38:379-400.

30.

Huang

Jian

van den Brink

Henriette Maassen

Groot

Wim

. 2009. “A Meta-analysis of the Effect of Education on Social Capital.” Economics of Education Review 28:454-64.

31.

Hustinx

Lesley

Cnaan

Ram A.

Handy

Femida

. 2010. “Navigating Theories of Volunteering: A Hybrid Map for a Complex Phenomenon.” Journal for the Theory of Social Behaviour 40:410-34.

32.

Kessler

Ronald C.

1982. “A Disaggregation of the Relationship between Socioeconomic Status and Psychological Distress.” American Sociological Review 47:752-64.

33.

Kessler

R. C.

Andrews

Colpe

L. J.

Hiripi

Mroczek

D. K.

Normand

L. T.

Walters

E. E.

Zaslavsky

A. M

. 2002. “Short Screening Scales to Monitor Population Prevalences and Trends in Non-specific Psychological Distress.” Psychological Medicine 32:959-76.

34.

Kessler

Ronald C.

Berglund

Patricia

Demler

Olga

Jin

Robert

Koretz

Doreen

Merikangas

Kathleen R.

Rush

A. John

Walters

Ellen E.

Wang

Philip S

. 2003. “The Epidemiology of Major Depressive Disorder.” Journal of the American Medical Association 289:3095-105.

35.

Keyes

Katherine M.

Nicholson

Ryan

Kinley

Jolene

Raposo

Sarah

Stein

Murray B.

Goldner

Elliot M

. 2014. “Age, Period and Cohort Effects in Psychological Distress in the United States and Canada.” American Journal of Epidemiology 179:1216-27.

36.

Kim

Joongbaeck

Pai

Manacy

. 2010. “Volunteering and Trajectories of Depression.” Journal of Aging and Health 22:84-105.

37.

Klerman

Gerald L.

Weissman

Myrna M

. 1989. “Increasing Rates of Depression.” Journal of the American Medical Association 261:2229-35.

38.

Lam

Pui-Yan

. 2002. “As the Flocks Gather: How Religion Affects Voluntary Association Participation.” Journal for the Scientific Study of Religion 41:405-22.

39.

Yunqing

Ferraro

Kenneth F

. 2005. “Volunteering and Depression in Later Life: Social Benefit or Selection Processes?” Journal of Health and Social Behavior 46:68-84.

40.

Yunqing

Ferraro

Kenneth F

. 2006. “Volunteering in Middle and Later Life: Is Health a Benefit, Barrier or Both?” Social Forces 85:497-519.

41.

McLeod

Jane D.

Fettes

Danielle L

. 2007. “Trajectories of Failure: The Educational Careers of Children with Mental Health Problems.” American Journal of Sociology 113:653-701.

42.

Mirowsky

John

Ross

Catherine

. 2003. Education, Social Status, and Health. New York: Aldine de Gruyter.

43.

Mood

Carina

. 2010. “Logistic Regression: Why We Cannot Do What We Think We Can Do, and What We Can Do about It.” European Sociological Review 26:67-82.

44.

Mor

Nilly

Winquist

Jennifer

. 2002. “Self-focused Attention and Negative Affect: A Meta-analysis.” Psychological Bulletin 128:638-62.

45.

Morton

Patricia M.

Mustillo

Sarah A.

Ferraro

Kenneth F

. 2014. “Does Childhood Misfortune Raise the Risk of Acute Myocardial Infarction?” Social Science and Medicine 104:133-41.

46.

Musick

Marc A.

Wilson

John

. 2003. “Volunteering and Depression: The Role of Psychological and Social Resources in Different Age Groups.” Social Science and Medicine 56:259-69.

47.

Musick

Marc A.

Wilson

John

. 2008. Volunteers: A Social Profile. Bloomington: Indiana.

48.

Musick

Marc A.

Wilson

John

Bynum, Jr

William B

. 2000. “Race and Formal Volunteering: The Differential Effects of Class and Religion.” Social Forces 78:1539-71.

49.

Oesterle

Sabrina

Johnson

Monica Kirkpatrick

Mortimer

Jeylan T

. 2004. “Volunteerism during the Transition to Adulthood: A Life Course Perspective.” Social Forces 82:1123-49.

50.

Omoto

Allen M.

Snyder

Mark

Hackett

Justin D

. 2010. “Personality and Motivational Antecedents of Activism and Civic Engagement.” Journal of Personality 78:1703-34.

51.

Paik

Anthony

Navarre-Jackson

Layana

. 2011. “Social Networks, Recruitment and Volunteering: Are Social Capital Effects Conditional on Recruitment?” Nonprofit and Voluntary Sector Quarterly 40:476-96.

52.

Paxton

Pamela

Reith

Nicholas E.

Glanville

Jennifer L

. 2014. “Volunteering and the Dimensions of Religiosity: A Cross-national Analysis.” Review of Religious Research 56:597-625.

53.

Pearlin

Leonard I.

Bierman

Alex

2013. “Current Issues and Future Directions in Research into the Stress Process.” Pp. 325-40 in Handbook of the Sociology of Mental Health, edited by Aneshensel

C. S.

Phelan

J. C

. New York: Springer.

54.

Penner

Louis A.

2002. “Dispositional and Organizational Influences on Sustained Volunteerism: An Interactionist Perspective.” Journal of Social Issues 58:447-67.

55.

Piliavin

Jane Allyn

Siegl

Erica

. 2007. “Health Benefits of Volunteering in the Wisconsin Longitudinal Study.” Journal of Health and Social Behavior 48:450-64.

56.

Putnam

Robert D.

1995. “Tuning in, Tuning out: The Strange Disappearance of Social Capital in America.” Political Science and Politics 28:664-83.

57.

Putnam

Robert D.

2000. Bowling Alone. New York: Simon & Schuster.

58.

Radler

Barry T.

Ryff

Carol D

. 2010. “Who Participates? Longitudinal Retention in the MIDUS National Study of Health and Well-being.” Journal of Aging and Health 22:307-31.

59.

Rotolo

Thomas

Wilson

John

. 2006. “Substitute or Complement? Spousal Influence on Volunteering.” Journal of Marriage and Family 68:305-19.

60.

Ryff

Carol

Keyes

Corey L. M

. 1995. “The Structure of Psychological Well-being Revisited.” Journal of Personality and Social Psychology 69:719-27.

61.

Schafer

Markus H.

Wilkinson

Lindsay R.

Ferraro

Kenneth F

. 2013. “Childhood (Mis)fortune, Educational Attainment, and Adult Health: Contingent Benefits of a College Degree?” Social Forces 91:1007-34.

62.

Sewell

William H.

Jr.

1992. “A Theory of Structure: Duality, Agency, and Transformation.” American Journal of Sociology 98:1-29.

63.

Smith

David Horton

. 1994. “Determinants of Voluntary Association Participation and Volunteering: A Literature Review.” Nonprofit and Voluntary Sector Quarterly 23:243-63.

64.

Snyder

Mark

Omoto

Allen M

. 2008. “Volunteerism: Social Issues Perspectives and Social Policy Implications.” Social Issues and Policy Review 2:1-36.

65.

Son

Joonmo

Wilson

John

. 2012. “Volunteer Work and Hedonic, Eudemonic, and Social Well-being.” Sociological Forum 27:658-81.

66.

Springer

Kristen W.

Hauser

Robert M

. 2006. “An Assessment of the Construct Validity of Ryff’s Scales of Psychological Well-being: Method, Mode, and Measurement Effects.” Social Science Research 35:1080-102.

67.

Steptoe

Andrew

O’Donnell

Katie

Marmot

Michael

Wardle

Jane

. 2008. “Positive Affect and Psychosocial Processes Related to Health.” British Journal of Psychology 99:211-27.

68.

Tellegen

Auke

Watson

David

Clark

Lee Anna

. 1999. “On the Dimensional and Hierarchical Structure of Affect.” Psychological Science 10:297-303.

69.

Thoits

Peggy A.

2003. “Personal Agency in the Accumulation of Multiple Role-Identities.” Pp. 179-94 in Advances in Identity Theory and Research, edited by Burke

P. J.

Owens

T. J.

Serpe

R. T.

Thoits

P. A

. New York: Kluwer.

70.

Thoits

Peggy A.

Hewitt

Lyndi N

. 2001. “Volunteer Work and Well-being.” Journal of Health and Social Behavior 42:115-31.

71.

Turner

Jonathan H.

Stets

Jan E

. 2006. “Sociological Theories of Human Emotions.” Annual Review of Sociology 32:25-52.

72.

Wilson

John

. 2012. “Volunteerism Research: A Review Essay.” Nonprofit and Voluntary Sector Quarterly 41:176-212.

73.

Wilson

John

Musick

Marc

. 1997a. “Who Cares? Toward an Integrated Theory of Volunteer Work.” American Sociological Review 62:694-713.

74.

Wilson

John

Musick

Marc

. 1997b. “Work and Volunteering: The Long Arm of the Job.” Social Forces 76:251-72.