Abstract
Objectives
. We examine how changes in social networks influence volunteerism through bridging (diversity) and bonding (spending time) mechanisms. We further investigate whether social network change substitutes or amplifies the effects of education on volunteerism.
Methods
. Data (n = 543) are drawn from a two-wave survey of Social Relations and Health over the Life Course (SRHLC). Zero-inflated negative binomial regressions were conducted to test competing hypotheses about how changes in social network characteristics alone and in conjunction with education level predict likelihood and frequency of volunteering.
Results
. Changes in social networks were associated with volunteerism: as the proportion of family members decreased and the average number of network members living within a one-hour drive increased over time, participants reported higher odds of volunteering. The substitution hypothesis was supported: social networks that exhibited more geographic proximity and greater contact frequency over-time compensated for lower levels of education to predict volunteering more hours.
Discussion
. The dynamic role of social networks and the ways in which they may work through bridging and bonding to influence both likelihood and frequency of volunteering are discussed. The potential benefits of volunteerism in light of longer life expectancies and smaller families are also considered.
Key Words: Education, Longitudinal, Social Networks, Volunteering.
Older persons’ participation in volunteering activities have increased dramatically in the last decades of the 20th century (Ajrouch, Akiyama, & Antonucci, 2007; Chambré, 1993), particularly among the young-old (Windsor, Anstey, & Rodgers, 2008; Wilson, 2000). Volunteering represents an opportunity for older adults to engage their social networks, as they contribute to their communities and society at large. Perhaps, the most compelling aspect of volunteering is the documented positive effects for the person volunteering, including physical and psychological health benefits (Li & Ferraro, 2005; Morrow-Howell, Hinterlong, Rozario, & Tang, 2003; Windsor, Anstey, & Rodgers, 2008). Predicting both the likelihood and frequency of volunteering, therefore, holds special importance for efforts to optimize health and well-being among older people. In addition, while much is known about the role of education in facilitating volunteerism (Brown & Ferris, 2007; Wilson & Musick, 1998), much less is known about the role of social networks.
Social networks are considered critical resources or pathways for volunteering. In the volunteer literature, networks are often measured in terms of community or association linkages, as well as contact frequency (Li & Ferraro, 2005; Morrow-Howell et al., 2003; Wilson, 2000; Windsor et al., 2008). Composition and geographic proximity of network members receive limited attention, yet may represent key aspects of social networks important for understanding volunteerism (Paik & Navarre-Jackson, 2011; Pilkington, Windsor, & Crisp, 2012). Composition captures a more nuanced element of network membership, whereas geographic proximity may promote face-to-face connections. Furthermore, an area still relatively unexplored in the volunteering literature, but which seems especially critical in the lives of older people, is the change in networks over time. Networks may change in that they grow or diminish in size, become more or less diverse, and promote or discourage face-to-face contact over time. A multidimensional approach to the study of social networks taps into the sociability aspect of volunteering activities. We propose that there are at least two potential pathways by which network changes are related to volunteerism: either through “bridging” or “bonding” (Paik & Navarre-Jackson, 2011). This study examines how networks change over time, and whether such changes are related to who volunteers, and the frequency with which they volunteer.
Theoretical Framework
We consider social networks as an important resource that changes across the life course. Therefore, we draw from multiple theoretical perspectives that convey the role social networks play in facilitating access to opportunities and in shaping outcomes. First, we use the Convoy Model of Social Relations (Antonucci, 2001; Antonucci, Ajrouch, & Birditt, 2014; Kahn & Antonucci, 1980), which describes social networks as multidimensional, dynamic and lifelong, changing in some ways, but remaining stable in others, across time and situations. As such, social networks can be considered a form of social capital (Antonucci, Ajrouch, & Park, 2014) one has access to varying amounts across their life. Social capital “is created when the relations among persons change in ways that facilitate action” (Coleman, 1990, p. 304). The ways in which networks change may provide critical insights into potential avenues of sociability, including integration, active engagement, and overall well-being. The systematic study of the multidimensional and dynamic nature of social networks represents a potential avenue for better understanding how older adults access volunteering opportunities.
Little is known about the ways in which networks change over time to influence volunteer behavior. Such change may particularly influence volunteerism in that the multiple dimensions of social networks yield the potential to promote social ties through bridging and bonding concepts drawn on from social network theory. Paik and Navarre-Jackson (2011) describe each. Bridging promotes connections as a result of the diversity of social ties, which includes whether they grow or diminish in size, and whether they change in composition to include those who are in general older or younger, family, or friends. Bonding, on the other hand, promotes connections as a result of spending time with network members. This may be inferred from changes in proximity, that is living closer or further away and increasing contact frequency reports with network members result in bonding between network members. We hypothesize that changes in social networks promote connections with others through bridging and bonding, which may in turn prompt volunteerism either directly, or in conjunction with human capital characteristics, notably education.
We further draw from the convoy model to posit that personal characteristics, such as education operate in tandem with social network characteristics to influence outcomes. Education level, an indicator of human capital, and social network characteristics, an indicator of social capital, are proposed to interact with one another, and have been examined as two competing hypotheses or influences on volunteering activities (Wilson & Musick, 1998). The first hypothesis is one of substitution, suggesting that low human capital is compensated for by access to plentiful social capital. As a result, those with lower education levels, but with advantageous network characteristics (larger, frequently seen, more proximal, and higher proportion of friends), would be more likely to volunteer. In other words, lacking in one resource (education) is substituted, and compensated, for by the strength inherent in the other (networks). Amplification, on the other hand, emphasizes a cumulative advantage position: those who have high levels of education are better able to leverage advantages that come with helpful network characteristics. This effect stems from the likelihood that those with higher education have enhanced cognitive resources and skills (Broese van Groenou & van Tilburg, 2003) as well as a heightened ability to negotiate various social commitments (McNamara & Gonzalez, 2011). They are, therefore, more likely to leverage benefits from networks. Hence, we draw from the traditions of social network theory as well as social gerontology to identify multiple dynamic dimensions of network structure, and education, as key factors that influence the outcome of volunteerism both directly and in conjunction with one another (see Figure 1).
Figure 1.
The interplay of social networks, education, and volunteerism.
Linking Social Networks to Volunteering
To better appreciate the importance of linking social networks to volunteerism, we review two significant areas of scholarship: (a) the multiple dimensions and dynamic nature of network characteristics; and (b) interactions between education and social networks.
Social networks are considered a resource that arises through relationships among individuals (Coleman, 1990). Given that volunteering activities are likely to occur because one is directly asked by a friend, relative, or acquaintance (McNamara & Gonzalez, 2011), we conceptualize networks as a social resource or capital that may provide greater access to volunteering opportunities and/or facilitate taking advantage of opportunities. The focus on the structure of an individual’s personal network emphasizes four main dimensions: size, contact frequency, geographic proximity, and composition (Ajrouch, Antonucci, & Janevic, 2001). Though older age is generally associated with smaller, less proximal, and less diverse social networks (Ajrouch et al., 2001; Cornwell, Laumann, & Schumm, 2008), there is still substantial heterogeneity among older adults on these dimensions (Fiori, Antonucci, & Cortina, 2006; Litwin, 2001; Wenger, 1997). The multiple dimensions and dynamic nature of network characteristics are conceptualized as a resource that may play an important role in facilitating volunteerism.
Social networks appear to influence volunteering, though results are mixed. Bridging aspects of social networks, such as larger size, those whose networks are on average younger, and those who have networks comprised mostly of family are thought to promote volunteer activities (Chambré, 1993; Herzog, Kahn, Morgan, Jackson, & Antonucci, 1989; Li & Ferraro, 2005; Musick, Herzog, & House, 1999; Rotolo, 2000). Bridging aspects are also associated with volunteering more often (Li & Ferraro, 2005; Morrow-Howell et al., 2003; Windsor, Anstey, & Rodgers, 2008). On the other hand, bonding characteristics such as frequent social contact also predict volunteering. For instance, Li and Ferraro (2005) found that the extent to which an older person has contact with friends, neighbors, or relatives influences whether or not they volunteer. We aim to clarify the ways in which social capital influences volunteerism by examining how changes in specific network characteristics have differential effects on volunteering likelihood and frequency. We hypothesize that changes in networks leading to bridging will influence both the likelihood and frequency of volunteering, whereas changes that lead to greater bonding characteristics will influence volunteer likelihood only.
It is necessary to understand the effects of social networks on volunteerism within the context of education level. Education level is widely acknowledged to influence the likelihood that older adults volunteer (Butrica, Johnson, & Zedlewski, 2009; Chambré, 1993; Li & Ferraro, 2005; Tang, 2006; Wilson & Musick, 1997; Wilson, 2000). Given that education signifies a source of human capital, which is “…created by changing persons so as to give them skills and capabilities that make them able to act in new ways” (Coleman, 1990, p. 304), the influence that it has on volunteer activities is considered quite large. Yet, social capital may enhance the effects of education on volunteer activities through amplification, or conversely social capital may substitute, or pave the way for volunteer activities in the absence of such human capital.
Prior research has found support for the amplification hypothesis (Wilson & Musick, 1998). Wilson and Musick reported that among an adult sample aged 25 and older, more highly educated people had a greater probability of volunteering in the presence of social capital, indicated as greater contact frequency, network density, and number of friends. To extend this rationale to volunteer frequency, it may be that those with higher levels of education, and access to greater social capital will report more volunteer frequency as well. Indeed, change in social networks may exert distinctive effects on volunteerism in the context of education level. Social networks that take on bridging characteristics over time, that is that become larger and more diverse, may substitute for lower levels of education to predict a higher likelihood and greater frequency of volunteering. Conversely, it may be that social networks that develop bonding characteristics over time, that is more contact frequency and greater geographic proximity to network members, will amplify the effects of higher education to predict a higher likelihood as well as greater frequency of volunteering. In sum, bridging characteristics of social networks may benefit those with lower human capital (i.e., substitution), whereas bonding characteristics of social networks benefit those with higher human capital (i.e., amplification), each exerting a unique effect depending on level of education.
Study Aims
A focused analysis on how human and social capital operates in tandem may lead to a clearer understanding of how various forms of capital influence the likelihood of volunteering and its frequency. Representing resources at the individual and social levels, we capitalize on unique longitudinal social network data to identify how education and network changes work to influence the likelihood and frequency of volunteering through bridging and bonding mechanisms. First, we propose a main effect hypothesis, predicting that higher education will be associated with both a higher likelihood of volunteering and frequency of volunteering. We further hypothesize that social network change will yield specific effects. Social networks that exhibit bridging over time will be associated with a greater likelihood and frequency of volunteering, whereas social networks that exhibit bonding over time will only be associated with a greater likelihood of volunteering.
Next, we examine how changes in social capital inherent to networks substitutes or amplifies the effects of education on volunteerism. Given that bridging aspects include larger and more diverse networks, factors usually associated with higher education levels (Ajrouch et al., 2005; Broese van Groenou & van Tilburg, 2003), we hypothesize that social network change that represents an increase in bridging characteristics over time will substitute for low education to predict a higher likelihood of volunteering. Conversely, we hypothesize that social networks exhibiting increased bonding characteristics, that is increasing levels of contact frequency and geographic proximity, will amplify the effects of higher education to predict a greater frequency of volunteering.
METHODS
Sample
Data for this study come from a two-wave survey of Social Relations and Health over the Life Course collected by the Survey Research Center at the University of Michigan (Antonucci, PI). The first wave of this regionally representative sample (N = 1,703) was collected in 1992 from those aged 8–93, with a 72% response rate. Wave 2, collected in 2005 consisted of 1,076 of the original respondents. Three hundred twenty had deceased and the remainder refused, were lost to follow-up, or were unable to participate resulting in a response rate of 78%. Respondents aged 50 and older at Wave 2 were selected from the larger study to allow a focus on mid and later life. The sample consisted of 543 adults who participated at both waves, ranging in age from 50 to 100.
Measures
Volunteerism was measured only at Wave 2 using a two-part self-report item assessing both the likelihood (i.e., “Do you do any volunteer work?”) and frequency (i.e., “if yes, about how many hours per…(week, month, or year)”) of volunteering. Frequency responses were converted to a metric of hours per year to create a single continuous variable ranging from zero (no volunteering) to the maximum number of hours (780) reported per year.
Education was measured at Wave 2 with a self-report item worded as “What is the highest grade of school or year of college you have completed?” Descriptive statistics revealed 16 respondents with years of education ranging from 1–7 as outliers, and subsequently were recoded to equal 8 years. Education level ranged from 8–17+ years.
Social network change.
A continuous change score was computed for each of the network indicators: size, composition, age, proximity, and contact frequency, with positive values indicating an increase and negative values a decrease in network characteristics. The hierarchical mapping technique (Antonucci, 1986) was used at both waves where respondents were shown a diagram containing three concentric circles and asked to nominate people in their lives based on varying levels of closeness. Respondents were then asked questions about the first 10 people named in their network age 13 or older. Five network indicators were measured at both waves assessing bridging and bonding aspects of networks. Two bridging characteristics were assessed, including: (a) Network Size, which indicates the total number of people the respondent included in his/her network diagram; and (b) Composition, measured as proportion family that is the percentage of the first 10 people named who are family, and average network age calculated by averaging the ages of the first 10 network members. Two bonding characteristics were assessed, including: (a) Network Proximity, which measured the geographic closeness of social network members to the respondent, operationalized as the percentage of the first 10 network members living within 1 hour drive of the respondent; and (b) Contact Frequency, where respondents were asked to rate on a 5-point scale ranging from irregularly = 1 to everyday = 5 how often they are usually in touch with network members. An average was calculated by computing responses of the first 10 people.
Control variables.
Seven variables identified in the literature to be significantly associated with volunteerism and health were used as controls. Age (at Wave 2) calculated from birth date. Gender, coded as male = 1; female = 2. Race/ethnicity was assessed with a self-report item worded as: “Are you white, black, Native American, Asian, Hispanic, Other?” Responses were coded so that white = 1 and non-white = 2. Employment status (at Wave 2): Respondents were asked how many hours they work in an average week. Responses were coded into three categories: not employed = 1 (reference category); work part-time (less than 40 hours per week) = 2; work full-time (40 or more hours per week) = 3. Caregiver Status (at Wave 2) was measured as “do you currently provide any unpaid care to a parent (in-law), spouse, child, or other person?” (no = 1; yes = 2). Health Limitation Change: At both waves respondents were asked if they were limited in any way because of their health (no = 1; yes=2). A change score was computed by subtracting Wave 1 health limitation status from Wave 2, with positive scores an indication of becoming limited over time, and negative values indicating no longer being limited. Depressive Symptoms Change: At both waves, depressive symptomatology was assessed with the 20-item Center for Epidemiological Studies Depression (CESD) scale (Radloff, 1977). Respondents reported the experience of depressive symptoms in the past week on a 4-point scale ranging from 0 (rarely/none of the time) to 3 (most of the time). Item scores were summed to create a total composite score for each wave with higher values indicating greater depression (Waves 1 and 2: α = 0.88). A change score was computed by subtracting Wave 1 depressive symptoms from Wave 2, with positive scores indicting an increase in depressive symptoms between waves, and a negative value indicating a decrease. In the multivariate analysis, the Wave 1 and health change scores were used as control variables.
Analysis strategy.
To examine our main effect hypotheses, we began by examining the effects of the five social network characteristics at Wave 2 on volunteerism, controlling for the variables mentioned above in order to descriptively understand the associations between social networks and volunteering. In the second model, we included social network change scores, controlling for Wave 1 social network characteristics and all the control variables mentioned above. Since volunteering is an on-going activity, we believe this approach illustrates social network change effects in a more informative way than presenting change scores alone.
Specifically, we conducted zero-inflated negative binomial models using Mplus. This technique was chosen due to the count nature of the volunteering measure, the wide range of volunteering hours reported (i.e., overdispersion), and the high proportion of respondents reporting no volunteering (i.e., zero-inflation), as indicated by both a significant (p < .001) dispersion (α) statistic and Vuong test (Zaninotto & Falaschetti, 2011). A zero-inflated negative binomial model conducts two simultaneous regressions to analyze the specified model. One, a logistic regression predicting the odds of no volunteering (i.e., zeros in the model); and the other, a regression predicting the expected count of hours volunteered among those reporting one or more hours. The unstandardized regression coefficients (slope) are reported for each model along with a measure of dispersion, significance of the Vuong test, and the N-adjusted BIC.
Prior to testing the final main effects model, we tested the assumptions of no influential cases (i.e., outliers) and a linear relationship between the predictors and outcome. First, we examined the results for multivariate outliers, looking for Cook’s D values greater than 1.0. We detected 13 possible outliers, upon which we compared model results with and without the outliers included in the data. We found differences in both the significance level and magnitude of effects, which led to the decision to remove the 13 outliers from the data, in addition to 31 cases with missing data, resulting in a final analysis n = 499. Next, we tested for a linear relationship between all predictors and the outcome in both the logistic and count parts of the model. In the logistic regression we performed the Box-Tidwell test for each predictor (Hosmer & Lemeshow, 1989), and detected no nonlinearity. In the count part of the model we examined partial regression plots, comparing variance explained by the linear and quadratic lines of best fit. Only contact frequency at Wave 1 showed substantial improvement when modeled as a quadratic. This was explored by adding a quadratic term for Wave 1 contact frequency in the model, and it was significant (p < .05). We then compared all study results with and without the quadratic term in the model. There were no substantive differences. In favor of parsimony, the decision was made not to include the quadratic term in the final analysis.
To test the substitution/amplification hypotheses, hierarchical regression was used to examine interactions between education and social network change. First, product/interaction terms were created by multiplying mean-centered versions of education and each of the five network change indicators. Next, each of the five interaction terms were added separately to the final main effects model. Only models with a significant interaction (p ≤ .05) are presented. Significant interactions were explored by examining the effects of social network change on volunteering within varying levels of education.
Results
We present first a descriptive analysis of the sample, followed by findings that address the study’s hypotheses. Table 1 provides the means and sample distribution of study variables examined from Wave 1 and Wave 2. When variables from both waves are analyzed, we also include a change score. The average age of the study sample at Wave 2 was 66.9 years (SD = 11.6) and ranged from 50 to 100. Approximately 60% of the sample were women, 78% were White, and 60% were married or living with a partner. Just over one-quarter (26%) of the sample reported working full-time at Wave 2, 13% part-time, and 61% were not employed. Among the respondents who were not employed, 73% reported that they were retired. At Wave 2, 18% of the sample identified themselves as a caregiver or currently providing unpaid care to someone. The prevalence of respondents who reported being limited by their health more than doubled from Wave 1 (15%) to Wave 2 (33%). On average, the sample’s mental health significantly (p < .05) improved across waves, as indicated by a reduction in depressive symptoms (M = −0.9; SD = 9.8) from Wave 1 (M = 8.7; SD = 8.9) to Wave 2 (M = 7.8; SD = 8.7).
Table 1.
Descriptive Statistics (n = 499)
| Wave 1 | Wave 2 | Δ Wave 1 to 2 | Sig. diff. | ||||
|---|---|---|---|---|---|---|---|
| M (SD) | Range | M (SD) | Range | M (SD) | Range | ||
| Control variables | |||||||
| Age | 66.9 (11.6) | 50–100 | |||||
| Female (%) | 60.3 | ||||||
| Proportion white (%) | 77.6 | ||||||
| Married/living with partner (%) | 60.3 | ||||||
| Employment status (%) | |||||||
| Work full-time | 25.5 | ||||||
| Work part-time | 13.2 | ||||||
| Not employed / retired | 61.3 | ||||||
| Caregiver (%) | 18.0 | ||||||
| Health limitationa (%) | 14.8 | 32.7 | 0.2 (0.5) | 1.0 to 1.0 | *** | ||
| Depressive symptoms | 8.7 (8.9) | 0–51.0 | 7.8 (8.7) | 0–57.0 | −0.9 (9.8) | −43.0–38.0 | * |
| Human Capital (Years of Education) | 13.4 (2.5) | 8–17 | |||||
| Social Capital (Social Network) | |||||||
| Size of network | 11.0 (5.9) | 1.0–35.0 | 11.5 (7.1) | 1.0–45.0 | 0.5 (7.0) | −23.0–39.0 | NS |
| Proportion family in network | 81.9 (22.8) | 0–100 | 80.5 (23.9) | 0–100 | −1.4 (23.5) | −80.0–85.7 | NS |
| Age of network | 45.7 (9.1) | 21.3–73.3 | 49.9 (10.0) | 17.0–82.0 | 4.2 (10.1) | −24.3–39.7 | *** |
| Proximity of networkb | 75.4 (25.1) | 0–100 | 71.7 (26.0) | 0–100 | −3.7 (25.8) | −90.0–88.9 | ** |
| Frequency of contact with networkc | 3.8 (0.6) | 1.0–5.0 | 3.9 (0.5) | 2.0–5.0 | 0.1 (0.7) | −1.8–3.0 | * |
| Volunteerism | |||||||
| Volunteering (%) | 32.3 | ||||||
| Hours volunteer per year | 67.3 (146.3) | 0–780.0 | |||||
Notes. NS = W1 to 2 difference not significant.
aLimited in any way because of health?. bPercentage of network members within 1 hour drive. cMeasured on 5-point scale (1 = irregularly to 5 = everyday)
*p < .05; **p < .01; ***p < .001.
Respondents reported completing 13.4 years of education on average (SD = 2.5), with a range of 8 years or less to 17 or more years. Some aspects of social networks changed very little over the 12-year period, whereas others were more dynamic. In terms of bridging aspects of social networks, network size minimally increased (M = 0.5; SD = 7.0) with respondents reporting on average 11.0 people in their social networks at Wave 1 (SD = 5.9; range: 1–35) and 11.5 people (SD = 7.1; range: 1–45) at Wave 2. Across the sample, the proportion of family members in networks slightly decreased (M = −1.4; SD = 23.5) from a sample average of 81.9% (SD = 22.8) at Wave 1 to 80.5% (SD = 23.9) at Wave 2. The average age of respondents’ social networks significantly increased by over 4 years (M = 4.2; SD = 10.1) from Wave 1 (M = 45.7; SD = 9.1) to Wave 2 (M = 49.9; SD = 10.0). In terms of bonding characteristics, the average percentage of social networks living within an hour’s drive (proximity) significantly decreased (M = −3.7; SD = 25.8) from an average of 75.4% (SD = 25.1) at Wave 1 to 71.7% (SD = 26.0) at Wave 2. Finally, average contact frequency also significantly increased (M = 0.1; SD = 0.7) over time from 3.8 (SD = 0.6) on a scale ranging from 1 to 5 at Wave 1 to 3.9 (SD = 0.5) at Wave 2, indicating between weekly and monthly contact at both waves.
Approximately one-third (32%) of the respondents reported volunteering. On average, the sample reported volunteering 67.3 hours per year (SD = 146.3) with a range from 0 hours to as many as 780 hours per year.
Table 2 provides the zero-inflated negative binomial regression analyses testing hypotheses about the effect of social capital on the likelihood and frequency of volunteering. Models 1 (cross-sectional/Wave 2 model) and 2 (change model) in Table 2 present the results testing the main effect hypotheses. To address hypotheses about substitution and amplification, we focus on the interaction effects of education and social network change on volunteerism. Only significant interaction models are presented (see Models 1, 2 and 3 in Table 3).
Table 2.
Zero-Inflated Negative Binomial Regression Predicting Volunteerisma,b
| No volunteering | Hours volunteer/year | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Model | 1a | 2a | 1b | 2b | |||||||
| Cross-sectional | Change | Cross-sectional | Change | ||||||||
| Control Variables | b | (SE) | b | (SE) | b | (SE) | b | (SE) | |||
| Age | −0.01 | (0.01) | −0.00 | (0.01) | 0.02 | (0.01)* | 0.01 | (0.01) | |||
| Female | −0.19 | (0.23) | −0.16 | (0.23) | 0.02 | (0.14) | −0.02 | (0.14) | |||
| Non-white | 0.34 | (0.27) | 0.32 | (0.28) | 0.17 | (0.17) | 0.23 | (0.18) | |||
| Married/living with partner | −0.26 | (0.25) | −0.26 | (0.26) | −0.30 | (0.14)* | −0.29 | (0.14)* | |||
| Employed part-time | 0.36 | (0.33) | 0.40 | (0.34) | −0.50 | (0.23)* | −0.46 | (0.25) | |||
| Employed full-time | −0.28 | (0.31) | −0.24 | (0.32) | −0.51 | (0.18)** | −0.61 | (0.18)** | |||
| Caregiver | 0.37 | (0.29) | 0.37 | (0.29) | 0.03 | (0.19) | 0.11 | (0.20) | |||
| Health limitation | 0.29 | (0.26) | 0.20 | (0.27) | −0.34 | (0.14)* | −0.20 | (0.15) | |||
| Depression | 0.03 | (0.02) | 0.03 | (0.02) | −0.03 | (0.01)*** | −0.03 | (0.01)*** | |||
| Years of Education | −0.20 | (0.05)*** | −0.19 | (0.05)*** | −0.05 | (0.03) | −0.05 | (0.03) | |||
| Social Network | |||||||||||
| Network size | −0.04 | (0.02)* | −0.03 | (0.02) | 0.01 | (0.01) | 0.01 | (0.01) | |||
| Proportion family | 0.01 | (0.01)** | 0.02 | (0.01)** | −0.01 | (0.00)* | −0.01 | (0.00) | |||
| Network age | −0.01 | (0.01) | 0.00 | (0.01) | −0.02 | (0.01)** | −0.02 | (0.01)** | |||
| Network proximity | −0.01 | (0.01) | −0.01 | (0.01)* | −0.00 | (0.00) | 0.00 | (0.00) | |||
| Contact frequency | 0.24 | (0.23) | 0.29 | (0.24) | 0.31 | (0.13)* | 0.17 | (0.16) | |||
| Model summary | Cross-sectional (1) |
Change (2) |
|||||||||
| N | 499 | 499 | |||||||||
| α (Dispersion) | 0.70 (SE = 0.08)*** | 0.65 (SE = 0.08)*** | |||||||||
| Vuong Test | *** | *** | |||||||||
| N-Adjusted BIC | 2656.24 | 2677.58 | |||||||||
Notes.
aModels 1a and 1b are cross-sectional examining associations of Wave 2 social network characteristics with both the likelihood and frequency of volunteering at Wave 2. bModels 2a and 2b examine associations between changes in social network characteristics with both the likelihood and frequency of volunteering at Wave 2, controlling for Wave 1 social network characteristics, which are not presented in the table.
*p < .05; **p < .01; ***p < .001.
Table 3.
Interaction Effects of Education and Social Network Change on Volunteerism
| Significant interaction effect models | ||||||||
|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | ||||||
| Education × Proportion family change |
Education x Network proximity change |
Education x Contract frequency change |
||||||
| Main effects | Hours volunteer/year | Hours volunteer/year | Hours volunteer/year | |||||
| Control variables | b (SE) | b (SE) | b (SE) | |||||
| Age | 0.01 | (0.01) | 0.01 | (0.01) | 0.01 | (0.01) | ||
| Female | −0.02 | (0.14) | −0.02 | (0.14) | −0.00 | (0.14) | ||
| Non-white | 0.23 | (0.18) | 0.27 | (0.19) | 0.29 | (0.19) | ||
| Married/living with partner | −0.32 | (0.14)* | −0.32 | (0.14)* | −0.26 | (0.14) | ||
| Employed part-time | −0.46 | (0.24) | −0.55 | (0.24)* | −0.53 | (0.24)* | ||
| Employed full-time | −0.66 | (0.18)*** | −0.66 | (0.18)*** | −0.65 | (0.18)*** | ||
| Caregiver | 0.08 | (0.20) | 0.07 | (0.19) | 0.14 | (0.20) | ||
| Health limitation change | −0.23 | (0.15) | −0.24 | (0.15) | −0.32 | (0.16)* | ||
| Depression change | −0.03 | (0.01)** | −0.03 | (0.01)*** | −0.03 | (0.01)*** | ||
| Years of Education | −0.04 | (0.03) | −0.03 | (0.04) | −0.05 | (0.03) | ||
| Social Network Change | ||||||||
| Network size change | 0.01 | (0.01) | 0.01 | (0.01) | 0.01 | (0.01) | ||
| Proportion family change | −0.00 | (0.00) | −0.01 | (0.00) * | −0.01 | (0.00)* | ||
| Network age change | −0.02 | (0.01)** | −0.02 | (0.01)** | −0.02 | (0.01)* | ||
| Network proximity change | 0.00 | (0.00) | 0.01 | (0.00)* | 0.00 | (0.00) | ||
| Contact frequency change | 0.18 | (0.16) | 0.19 | (0.16) | 0.26 | (0.15) | ||
| Interaction Effects | ||||||||
| Education × Social network change |
−0.00 | (0.00)* | −0.00 | (0.00)** | −0.14 | (0.06)** | ||
| Model Summary | ||||||||
| N-Adjusted BIC | 2680.68 | 2675.52 | 2675.65 | |||||
*p < .05; **p < .01; ***p < .001.
Likelihood of Volunteering
Main effects of education and network change.
Education was found to be significantly and positively related to the likelihood of volunteering. The completion of more years of education was associated with greater odds of volunteering. In the cross-sectional (Wave 2) analysis, we found that larger networks and a lower proportion of family were associated with volunteering. In terms of network change, our hypothesis was partially supported. We found that change in networks resulting in more bridging (less family over time) and more bonding (greater geographic proximity over time) both were related to increased odds of volunteering. Specifically, a decrease in the proportion of family members in social networks and an increase in the proportion of network members living within a one-hour drive of respondents were both associated with increased odds of volunteering.
Interaction of education and social network change
.—None of the education by social network change interactions tested predicted the likelihood of volunteering. Contrary to the hypotheses, changes in the bridging aspects of social networks do not appear to amplify or substitute for the effect of education on the likelihood of volunteering.
Volunteer Frequency
Main effects of education and network change.
Education was not related to the number of hours respondents reported volunteering per year, contrary to our hypothesis. At Wave 2, lower proportions of family, younger networks, as well as higher levels of contact frequency were associated with more hours of volunteering. In terms of network change, the findings support our hypothesis, in that social network change resulting in bridging did predict volunteer frequency. Specifically, a decrease in the average age of respondents’ networks was associated with volunteering more hours per year.
Interaction of education and social network change.
Contrary to our hypothesis, social networks that changed to yield bonding characteristics appeared to substitute for, not amplify, the effects of education on volunteer frequency. Changes in proportion family, network proximity, and contact frequency were found to significantly interact with education when predicting the number of hours respondents volunteer per year (i.e., frequency).
To explore the significant interactions, we divided the sample into two groups based on the median years of education (15); those with lower (i.e., < 15 years/less than a college degree) and higher (i.e., 16+ years /college degree or more) levels of education. We then examined the effects of proportion family, proximity, and contact frequency change on volunteering frequency separately within these levels of education. First, in terms of proportion family, Figure 2a indicates that among respondents with higher levels of education, a decrease in the proportion of family in one’s network was associated with greater volunteering frequency. In contrast among those with lower levels of education, proportion family was not associated with volunteering frequency.
Figure 2.
Interaction of education and social network (proportion family, proximity, and contact frequency) change on the number of hours volunteer/year.
Second, in terms of network proximity, Figure 2b indicates that among respondents with lower levels of education, an increase in the proximity of their network (i.e., increase in the proportion of their network living within an hour’s drive) was associated with greater volunteer frequency. In contrast, among respondents with higher levels of education, a decrease in network proximity was associated with greater volunteering frequency.
In terms of contact frequency, Figure 2c indicates that among respondents with lower levels of education, an increase in contact frequency with their network members was associated with greater volunteer frequency. In contrast, among respondents with higher levels of education, network contact frequency was not associated with volunteering frequency.
The interactions are in the opposite direction as hypothesized, as bridging (i.e., decrease in the proportion of family) was linked to more volunteering in the context of higher education (amplification), and bonding (i.e., increases in geographic proximity and contact frequency) was related to more volunteering in the context of lower education (substitution).
Discussion
The dynamic nature of social networks suggests the development of both bridging and bonding characteristics over time. Each has unique implications for understanding whether and how context facilitates volunteerism. The likelihood of volunteering, as well as volunteer frequency, was examined in an attempt to explicate the significance of network change among older adults. Bridging and bonding play an important role in shaping directly—and to some extent in conjunction with education—volunteerism in later life. We consider these results in some detail below.
Social Network Change and Volunteerism
Networks change to exhibit both bridging and bonding characteristics as people age. In this study, we examined the ways in which those network changes influence volunteerism. Though family is thought to promote more volunteer activities (Chambré, 1993; Herzog et al., 1989; Li & Ferraro, 2005; Musick et al., 1999; Rotolo, 2000), our findings suggest that as networks change over time to comprise less family, the likelihood of volunteering increases. Moreover, as networks are increasingly comprised of younger members, volunteer frequency increases. Change in composition, not size, holds the most potential to shape volunteer activity in later life. Findings hint at the significance of non-family intergenerational relationships for volunteerism in older adults’ social networks, indicating that over time who is in one’s network matters more than how many are in the network. It may be that volunteerism fills a void created by the absence of family members and the loss of older network members. Additionally, having younger network members may simply prompt more activity among those 50 years and older. In sum, changes in networks that result in bridging (i.e., dimensions that facilitate diverse networks), may operate as a resource to benefit older adults most in this context by facilitating volunteerism.
Of the network characteristics that result in more bonding over time, (i.e., dimensions facilitating spending time together), only increasing geographic proximity was found to promote the likelihood of volunteering. This dimension may serve as a resource in an increasingly mobile society simply because geographic proximity promotes face-to-face interactions, an important aspect of sociability, especially for older adults (Kweon, Sullivan, & Wiley, 1998). This finding extends the ways in which likelihood of volunteering may be seen as an extension of social relationships (Chambré, 1993; Knoke & Thomson, 1977; Rotolo, 2000). As older adults develop networks that are physically proximal, they are more likely to volunteer.
Overall, our findings suggest the dynamic nature of social networks hold special importance for both volunteer likelihood and frequency. We examined distinct dimensions of social networks, and identified how changes in some dimensions were more strongly related to volunteerism than others. In some sense, our hypotheses were supported in that network changes that promoted bridging elements were associated with the likelihood and frequency of volunteering. Moreover, bonding characteristics were more often associated with whether or not one volunteered in the first place. Yet, given the data limitation of having volunteering data only at Wave 2, the directionality of the tested associations should be interpreted with caution. For instance, it could very well be that volunteering at Wave 2 initiated the detected changes in network characteristics. If so, it may be that the act of volunteerism promotes change in social networks that incur both bridging and bonding tendencies rather than the other way around. Regardless, social network change is clearly associated with volunteerism, and we expect that reciprocal causality is most likely, that is that both are causing changes in the other.
The Substitution Effect of Social Capital
Findings supported previous research in that higher education levels were associated with a greater likelihood of volunteering (Brown & Ferris, 2007; Wilson & Musick, 1998). Interestingly, however, higher education levels had no main effect on volunteer frequency. This finding provides a more nuanced understanding of the extent to which education predicts volunteerism. It suggests that education as human capital may directly promote the likelihood of volunteering, but not necessarily the extent to which people volunteer.
Volunteering may not benefit all older people equally (Martinson & Minkler, 2006; Musick, Herzog, & House, 1999), and furthermore, there appears to be a point of diminishing returns. Moderate levels of volunteering have overall positive effects, whereas high levels may produce damaging effects on health and well-being, (Li & Ferraro, 2005; Morrow-Howell, Hinterlong, Rozario, & Tang, 2003; Windsor, Anstey, & Rodgers, 2008). Distinguishing between likelihood and prevalence, therefore, has important well-being implications for aging and older adults. Interestingly, volunteer frequency is influenced by education levels only in the context of networks that exhibit more bonding aspects over time. We consider these findings next.
Two of the three significant interaction effects illustrate that substitution best describes the multiplicative influence of education (human capital) and social network change (social capital) on the frequency of volunteering among older adults. Wilson and Musick (1998) found support for the amplification hypothesis in their sample aged 25+. The present study, with a focus on middle-aged and older adults (50+), found support for the substitution hypothesis. Taken together these findings suggest that both hypotheses are valid, but that the likelihood of volunteering is differentially influenced at different points of the life course. In particular, the present findings indicate that bonding characteristics appear particularly influential in the context of lower education levels concerning volunteer frequency. As networks become more geographically proximal and older adults have more contact frequency with their networks, they are more likely to volunteer at higher levels. Of course it may be that networks develop that way because of volunteering frequency. Given the one-time only limitation of our volunteer measure, we cannot discern the directionality of this association. Nevertheless, this finding points to the significance of changes in bonding characteristics among those with lower levels of education. In other words, lacking in one resource (education) is substituted for, and compensated by the strength inherent in the other (social networks). From this perspective, it can be argued that social capital helps reduce disparities in volunteer engagement that result from differential access to human capital. Such findings make visible the potential impact of a social resource. However, it is important to note that we do not see a complete substitution effect in that human capital does not influence volunteer frequency directly.
Future Directions
Societal level transitions in population aging and improved health status among older adults make the present time an ideal opportunity to seriously consider how social capital may provide a potential link to enhanced role opportunities in society. Though we must be cautious about assuming all older adults are healthy and should or will want to volunteer (Martinson & Minkler, 2006), it is widely acknowledged that societal structures are lagging behind in providing roles and opportunities for the newly emerging healthier, more functionally able older adult (Riley & Riley, 1994; Wilmoth, 2010).
This study characterized social networks in terms of bridging and bonding aspects that can change over time. This approach positions networks as an important resource for integration, embeddedness, and ultimately continuous and long lasting engagement in various social roles. The data reported here suggest that network changes do indeed happen, and the ways in which they occur have implications for the roles older adults play in society. Limitations, however, must be recognized. The volunteering measure was collected at only one point in time; therefore, future research will need to disentangle whether these changes in network structure precipitated the act of volunteering, whether the volunteering act itself produced changes in social networks, or if there are reciprocal effects. Moreover, detailed information on the type of volunteering activities, including formal and informal aspects, would yield a better understanding of the ways in which social network change predicts such activities. Despite these limitations, the central point here concerns the fact that likelihood and frequency of volunteering are uniquely associated with variations in those network dynamics. Such expressions of social capital may provide interesting pathways which can maximize health and well-being through the creation of roles and opportunities from which the newly emerging elder can benefit.
Our focus on network change provides an empirical test of social network dynamics. Furthermore, by introducing bridging and bonding concepts, we elaborate mechanisms through which various network dimensions have potential to impact outcomes. These findings suggest that policy makers and program planners should not view pathways to volunteerism simplistically, but instead include both individual- and social-level resources. Considering which type of change in social capital substitutes or amplifies the effects of education on likelihood and frequency of volunteering yields an important understanding about the ways in which social networks benefit an aging society.
Funding
This work was supported by the National Institute of Mental Health (RO1MH066876 to T.C.A.); and National Institutes of Health (RO1 AG027021 to T.C.A.)
Conflict of Interest
K.J. Ajrouch and T.C. Antonucci planned the study, conceptualized the theoretical framework, supervised the data collection, and analysis. N.J. Webster developed the analytic strategy and condcuted the statistical analyses. All authors jointly wrote and edited the manuscript.
Acknowledgments
Previous versions of this paper were presented at the Gerontological Society of America, the American Sociological Association, and the Society for the Study of Human Development. The authors would like to thank the Life Course Development Program at the Institute for Social Research for comments made on earlier versions.
References
- Ajrouch K. J., Akiyama H., Antonucci T. C. (2007). Cohort differences in social relations among the elderly. In Wahl H.-W., Tesch-Romer C., Hoff A. (Eds.) Emergence of New Person-Environment Dynamics in Old Age: A Multidisciplinary Exploration, (pp. 43–64) Amityville, NY: Baywood Publishing; dx.doi.org/10.1080/03601270701700714 [Google Scholar]
- Ajrouch K. J., Antonucci T. C., Janevic M. R. (2001). Social networks among blacks and whites: The interaction between race and age. Journal of Gerontology: Series B: Pscyhological Sciences and Social Sciences, 56B, S112–S118. doi: 10.1093/geronb/56.2.S112 [DOI] [PubMed] [Google Scholar]
- Ajrouch K. J., Blandon A., Antonucci T. C. (2005). Social networks among men and women: The effects of socioeconomic status and age. Journal of Gerontology: Series B: Pscyhological Sciences and Social Sciences, 60, S311–S317. doi:10.1093/geronb/60.6.S311 [DOI] [PubMed] [Google Scholar]
- Antonucci T. C. (1986). Social support networks: A hierarchical mapping technique. Generations, X, 10–12. [Google Scholar]
- Antonucci T. C. (2001). Social relations: An examination of social networks, social support and sense of control. In Birren J. E., Schaie K. W. (Eds.). Handbook of the Psychology of Aging (5th ed, pp. 427–453). New York, NY: Academic Press. [Google Scholar]
- Antonucci T. C., Ajrouch K. J., Birditt K. S. (2014). The convoy model: explaining social relations from a multidisciplinary perspective. The Gerontologist, 54, 82–92. 10.1093/geront/gnt118 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Antonucci T. C., Ajrouch K. J., Park S. (2014). Social capital and gender: Critical influences on living arrangements and caregiving in old age. In Pachana N. A., Laidlaw K. (Eds.), Oxford Handbook of Clinical Geropsychology: International perspectives, Oxford, UK: Oxford University Press. [Google Scholar]
- Broece van Groenou M. I., van Tilburg T. (2013). Network size and support in old age: differentials by socio-economic status in childhood and adulthood. Ageing and Society, 23, 625–645. doi: 10.1017/S0144686X0300134X [Google Scholar]
- Brown E., Ferris J. M. (2007). Social capital and philanthropy: An analysis of the impact of social capital on individual giving and volunteering. Nonprofit and Voluntary Sector Quarterly, 36, 85–99. doi: 10.1177/0899764006293178 [Google Scholar]
- Butrica B. A., Johnson R.W., Zedlewski S.R. (2009). Volunteer dynamics of older Americans. Journal of Gerontology: Series B: Psychological Sciences and Social Sciences, 64B, 644–655, doi:10.1093/geronb/gbn042 [DOI] [PubMed] [Google Scholar]
- Chambré S. M. (1993). Volunteerism by elders: Past trends and future prospects. TheGerontologist, 33, 221–228. doi: 10.1093/geront/ 33.2.221 [DOI] [PubMed] [Google Scholar]
- Coleman J. S. (1990). Foundations of Social Theory. Cambridge, MA: Harvard University Press. [Google Scholar]
- Cornwell B., Laumann E. O., Schumm P. L. (2008). The social connectedness of older adults: a national profile. American Sociological Review, 73, 185–203. doi:10.1177/000312240807300201 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fiori K. L., Antonucci T. C., Cortina K. S. (2006). Social network typologies and mentalhealth among older adults. The Journals of Gerontology, 61, 25–32. doi:10.1093/geronb/61.1.P25 [DOI] [PubMed] [Google Scholar]
- Herzog A. R., Kahn R. L., Morgan J. N., Jackson J. S., Antonucci T. C. (1989). Age differences in productive activities. The Journal of Gerontology, 44, S129–S138. doi:10.1093/geronj/44.4.S129 [DOI] [PubMed] [Google Scholar]
- Kahn R. L., Antonucci T. C. (1980). Convoys over the life course: Attachment, roles, and social support. In Baltes P. B., Brim O. G. (Eds.), Life-span development and behavior (pp.253–286) New York, NY: Academic Press. [Google Scholar]
- Knoke D., Thomson R. (1977). Voluntary association membership trends and the family life cycle. Social Forces, 56, 48–65. doi: 10.2307/2577412 [Google Scholar]
- Kweon B. S., Sullivan W. C., Wiley A. R. (1998). Green common spaces and the social integration of inner-city older adults. Environment and Behavior, 30, 832–858. doi: 10.1177/001391659803000605 [Google Scholar]
- Li Y., Ferraro K. F. (2005). Volunteering and depression in later life: Social benefit or selection processes? Journal of Health and Social Behavior, 46, 68–84. doi: 10.1177/002214650504600106 [DOI] [PubMed] [Google Scholar]
- Litwin H. (2001). Social network type and morale in old age. The Gerontologist, 41, 516––524. doi: 10.1093/geront/41.4.516 [DOI] [PubMed] [Google Scholar]
- Martinson M., Minkler M. (2006). Civic engagement and older adults: A critical perspective. The Gerontologist, 46, 318–324. doi: 10.1093/geront/46.3.318 [DOI] [PubMed] [Google Scholar]
- McNamara T. K., Gonzales E. (2011). Volunteer transitions among older adults: the role of human, social, and cultural capital in later life. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 66, 490–501. doi:10.1093/geronb/ gbr055 [DOI] [PubMed] [Google Scholar]
- Morrow-Howell N., Hinterlong J., Rozario P. A., Tang F. (2003). Effects of volunteering on the well being of older adults. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 58B, S137–S145. doi: 10.1093/geronb/58.3.S137 [DOI] [PubMed] [Google Scholar]
- Musick M. A., Herzog A. R., House J. S. (1999). Volunteering and mortality among older adults: Findings from a national sample. The Journals of Gerontology: Series B: Psychological and Social Sciences, 54B S173–S180. doi: 10.1093/geronb/54B.3.S173 [DOI] [PubMed] [Google Scholar]
- Paik A., Navarre-Jackson L. (2011). Social networks, recruitment, and volunteering: are social capital effects conditional on recruitment? Nonprofit and Voluntary Sector Quarterly, 40, 476–496. doi: 10.1177/0899764009354647 [Google Scholar]
- Pilkington P. D., Windsor T. D., Crisp D. A. (2012). Volunteering and subjective well-being in midlife and older adults: the role of supportive social networks. The Journals of Gerontology: Series B, Psychological Sciences and Social Sciences, 67, 249–260. 10.1093/geronb/gbr154 [DOI] [PubMed] [Google Scholar]
- Riley M. W., Rile J. W. (1994). Age integration and the lives of older people. The Gerontologist, 34, 110–115. doi: 10.1093/geront/34.1.110 [DOI] [PubMed] [Google Scholar]
- Rotolo T. (2000). A time to join, a time to quit: The influence of the life cycle transitions on voluntary association membership. Social Forces, 78, 1133–1161. doi: 10.2307/3005944 [Google Scholar]
- Tang F. (2006). What resources are needed for volunteerism? A life course perspective. Journal of Applied Gerontology, 25, 375–390. doi: 10.1177/0733464806292858 [Google Scholar]
- Wenger C. G. (1997). Social networks and the prediction of elderly people at risk. Aging and Mental Health, 1, 311–320. doi:10.1080/ 13607869757001 [Google Scholar]
- Wilmoth J. (2010). Aging policy and structural lag. In Hudson R. B. (Ed.), The new Politics of Old Age Policy (pp. 42–63). Baltimore, MD: John Hopkins University Press. [Google Scholar]
- Wilson J. (2000). Volunteering. Annual Review of Sociology, 26, 215–240. doi:10.1146/annurev.soc.26.1.215 [Google Scholar]
- Wilson J., Musick M. (1998). The contribution of social resources to volunteering. Social Science Quarterly, 79, 799–814. [Google Scholar]
- Windsor T. D., Anstey K. J., Rodgers B. (2008). Volunteering and psychological wellbeing among young-old adults: How much is too much? The Gerontologist, 48, 59–70. doi: 10.1093/geront/48.1.59 [DOI] [PubMed] [Google Scholar]
- Zaninotto P., Falaschetti E. (2011). Comparison of methods for modelling a count outcome with excess zeros: application to Activities of Daily Living (ADL-s). Journal of Epidemiology and Community Health, 65, 205–210. 10.1136/jech.2008.079640 [DOI] [PubMed] [Google Scholar]