Voluntary leadership roles in religious groups and rates of change in functional status during older adulthood

Abstract

Linear growth curve modeling was used to compare rates of change in functional status between three groups of older adults: Individuals holding voluntary lay leadership positions in a church, regular church attenders who were not leaders, and those not regularly attending church. Functional status was tracked longitudinally over a 4-year period in a national sample of 1,152 Black and White older adults whose religious backgrounds were either Christian or unaffiliated. Leaders had significantly slower trajectories of increase in both the number of physical impairments and the severity of those impairments. Although regular church attenders who were not leaders had lower mean levels of impairment on both measures, compared with those not regularly attending church, the two groups of non-leaders did not differ from one another in their rates of impairment increase. Leadership roles may contribute to longer maintenance of physical ability in late life, and opportunities for voluntary leadership may help account for some of the health benefits of religious participation.

Introduction

The timing and rate of decline in functional ability in older adulthood is highly variable between individuals (Christensen et al., 2008), but has important implications for well-being and the need for medical treatment during late life (Manton et al., 2006). As worldwide demographics continue to shift, with the population of older adults expected to continue to increase at more than double the rate of the younger population until at least 2050 (United Nations, 2010), it becomes increasingly important to understand factors that may be related to slower rates of change in functional disability among older adults. Participation in a religious organization has been found to be related to lower overall levels of functional disability in older adulthood (Powell et al., 2003), and previous research suggests that this may be partly explained by social support relationships between members of religious congregations (Hayward & Krause, 2012). But religion is a complex and multi-faceted phenomenon (Fetzer Institute & National Institute on Aging, 1999), and other factors may also play a role in helping older adults to maintain functional status. In particular, numerous studies have shown that giving support to the community can be just as beneficial in terms of well-being as receiving support (Brown et al., 2003; Krause, 2006). Religious groups may provide older adults with the opportunity to play an important role in the community: that of leadership. Although this facet of religious involvement has received relatively little attention from researchers, a few studies suggest that people holding lay leadership positions may derive more benefits in terms of well-being than others who attend religious services (Krause et al., 1998; Pargament et al., 2001). This study examines whether having this type of leadership position in one’s religious group is related to less rapid decline in physical functioning among older adults.

Leadership roles and aging

Aging often brings about disengagement from certain roles of authority in important areas like work and family (Kim & Moen, 2002), but it is also associated with growing needs to make a lasting contribution (Bradley, 1997; Erikson, 1950). Loss of roles during older adulthood may contribute to a deteriorating sense of self efficacy, which may in turn lead to worsening mental and physical health (Adelmann, 1994). A number of studies have found that continuing engagement in multiple social roles is related to better health in late life (Adelmann, 1994), and despite some early theoretical controversy (Achenbaum & Bengtson, 1994), it is generally regarded as an important element in successful aging (Moen et al., 1992). Because older adulthood is frequently marked by social transitions such as retirement and widowhood that entail the loss of some important sources of social roles and identity, participation in voluntary associations becomes especially important as an alternate source of role engagement during this stage of life. Again, numerous studies have demonstrated that volunteering is related to better health outcomes in late life (Musick & Wilson, 2008).

Relatively little research has been directed specifically at voluntary leadership roles, as opposed to volunteering in general, but there are theoretical reasons to expect that acting as a group leader may have a particularly strong impact on self-concept and hence on related health outcomes. Leadership positions often accord a degree of status, responsibility, and potential to achieve tangible accomplishments, all of which may serve to bolster the self-image and general sense of personal control. In addition, the insights of group dynamics research suggests that leaders tend to be perceived by other members as both especially representative of, and valuable to, the group (Hogg et al., 2004), and hence receive more respect and support in comparison to the rank and file (Fielding & Hogg, 1997). Emerging evidence from the perspective of organizational studies may provide some support for this expectation, showing that practicing certain leadership styles contribute to cultivation of a positive sense of self, and consequent well-being enhancement (Gardner et al., 2011).

Religion, leadership, and aging

Numerous studies have examined the connections between religious involvement and health status in older adulthood (Koenig et al., 2012). Fewer have examined religion in relation to functional status and disability, but the evidence suggests that higher levels of religious involvement are generally related to less presence of functional impairment (Powell et al., 2003). Several of these studies further suggest that it is primarily the social elements of religious involvement such as worship service attendance and participation in religious classes and prayer groups, as opposed to elements of private beliefs and practices, that are associated with functional status (Idler & Kasl, 1992, 1997; Idler, 1987; Koenig et al., 2004). Recent research has identified church-based social support as a potential part of this process, with tangible support received from fellow church members being related to slower trajectories of functional decline in late life (Hayward & Krause, 2012).

Given the challenges posed by role loss in late life, one important facet of involvement in the religious group may be that it provides opportunities to volunteer for a variety of lay leadership positions. Although group structures vary widely between groups, many religious groups rely on volunteers to fill formal roles ranging from serving on administrative committees, to teaching classes, to acting as spiritual counselors for other members (Stark, 1996). Since it is thought that people often become more involved in religious group participation during older adulthood (Krause, 2008), these roles may be especially likely to be filled by older adults.

Although some studies have examined the health implications of pastoral work for professional religious leaders (Meisenhelder & Chandler, 2001; Proeschold-Bell & LeGrand, 2012), relatively little is known about the impact of occupying a voluntary position of lay leadership within a religious organization. Several studies have been based on data collected by the Presbyterian Church, comparing clergy, lay leaders, and rank and file members. One study found that lay leaders received more social support from other group members, but also had more negative social interactions in the church, and moreover that these factors also had a stronger impact on affect for lay leaders (Krause et al., 1998). Another found that lay leaders engaged in more positive religious coping and less negative religious coping, compared with rank and file church members, but that they also suffered more deleterious consequences for affect and religious satisfaction from engaging in negative coping (Pargament et al., 2001). On the other hand, a number of studies have found that these lay leaders do not exhibit differences from the rank and file in terms of psychological distress and changes in distress over time (Bradshaw et al., 2010; Ellison et al., 2009, 2012). However, these studies are limited in that they focus on a single type of leadership status (being a church elder) within a single denomination; furthermore, the direct implications for functional status have not been examined.

There are at least three reasons why lay leadership might be expected to contribute to slowing the rate of functional decline for older adults. First, the duties of leadership may entail regularly engaging in physical activities that have important health benefits, particularly for those older adults who tend to be otherwise sedentary. For these individuals even mild forms of activity, such as walking from the parking lot to attend a meeting or standing more often during worship services to greet or direct other worshippers, may serve to help maintain and improve physical ability. Previous research indicates that older adults who may be reluctant to participate in physical activities for the sake of health promotion may obtain physical health benefits from participating in community-based volunteer work as a result of these kinds of incidental physical activities (Tan et al., 2006). Second, acting as a leader in the church is likely to offset the effects of age-related role loss in other areas, bolstering the self-image and enhancing the sense of meaning and purpose (Reitzes & Mutran, 1994). Both of these self-related outcomes my in turn contribute to preserving functional status, for example by improving feelings of self-efficacy that encourage positive health behaviors, and by buffering against symptoms of depression that can negatively impact physical functioning (Grembowski et al., 1993). Third, social identity processes within religious groups may work to indirectly influence functional outcomes. Group leaders may be highly visible to other group members, and are likely to be seen as especially prototypical group members, that is, individuals who closely embody the norms and values of the group (Hogg et al., 2004). Several studies have shown that individuals who have the strongest identification with a religious group (Greenfield & Marks, 2007; Keyes & Reitzes, 2007), and who are perceived as most representative of the group prototype (Hayward & Elliott, 2009, 2011), tend to receive stronger benefits in terms of self-reported physical health and general well-being from the group. Although its relationship with physical functioning has received less attention, these factors may have an impact on functional status both because of the established link between mental and physical health in older adulthood (Callahan et al., 2005) and because more highly-identified and prototypical church members tend to receive more social support from other group members (Hayward & Elliott, 2011), which has in turn been linked with better maintenance of physical functioning (Hayward & Krause, 2012).

Hypotheses

The present study uses 5 years of longitudinal data on physical functioning and severity of physical limitations in a sample of older adults. Based on previous research and theory, we predict that older adults who hold lay leadership positions in their religious groups will experience less rapid worsening of both of these outcomes, when compared with both (1) older adults who do not regularly attend religious services, and (2) older adults who regularly attend religious services, but do not hold lay leadership positions.

Method

This study is based on three waves of data from the Religion, Aging, and Health Survey (Krause, 2002). The sample was randomly drawn from the Centers for Medicare and Medicaid Services beneficiary list in 2001 to be representative of the US population aged 66 and older, excluding residents of Alaska and Hawaii. Screening of the initial sample was used to restrict the final pool of respondents to those reporting their race as Black or White, and whose religious affiliation fell into one of the following categories: currently Christian, formerly Christian but currently practicing no religion, or had never been affiliated with any religion. The rationale for this approach to sample screening is described in detail by Krause (2002). In brief, this method facilitated oversampling of Blacks and comparison with Whites, and allowed the use of a large number of measures of religious belief and practice validated only within Christian samples.

Data collection was conducted by Harris Interactive (New York) via face-to-face interview in respondents’ homes. Functional status was assessed beginning in the second wave of the survey, and hence the present analyses encompass data from Waves 2–4. The baseline survey (Wave 1) was conducted in 2001, and included a total of 1,500 completed interviews, with 752 African American and 748 White respondents; the overall response rate was 62 %. Wave 2 was conducted in 2004, and resulted in 1,024 complete interviews, with a re-interview rate of 80 % after excluding respondents who had died or become ineligible (due to institutionalization) in the interim. Wave 3 followed in 2007 (N = 969, 75 % re-interview rate), and Wave 4 in 2008 (N = 718, 88 % re-interview rate).

Measures

Functional disability

Functional ability was measured using a 15-item index of activities of daily living (ADL) and instrumental activities of daily living (IADL). Sample items include “taking a train or bus by yourself” and “reaching over your head.” Higher scores indicate a greater degree of functional impairment. For each item on the ADL/IADL that a respondent had difficulty with, a second set of questions was used to measure the severity of the difficulty with the activity on a 4-point scale from “a little” to “unable to do.” This index is very similar, in terms of ADL/IADLs assessed and scoring, to measures used in a number of studies of physical function in non-clinical samples of older adults (e.g. Harris et al., 1989; Liang et al., 2003), and which have been found to be reliable over time (Andresen et al., 2005) as well as predictive of other important late life health outcomes (Wolinsky et al., 2005).

Based on these items we calculated two measures of functional impairment. The first is a scale of the number of functional limitations (ranging from 0 to 15), and is computed for all study participants. The second is a measure of limitation severity, calculated as a sum of the severity ratings associated with each activity for which any limitation was reported. This measure is available only for those participants who reported at least one functional limitation. Although these measures are highly correlated (r = .71, p < .001), we analyze both outcomes for conceptual and practical reasons. Conceptually, it is possible to distinguish the breadth of daily activities affected by impairment from the severity of those impairments. From an analytical standpoint, it is desirable to be able to include the maximum available number of observations and participants, but also to evaluate whether different patterns emerge when only the impaired part of the population is considered, for example in case leadership is not related to the types of activities that become impaired, but is related to whether those impairments remain minor or become debilitating.

Church leadership

Participants were asked whether or not they held a variety of leadership positions in the church using a checklist that included seven church offices: deacon (4.8 %), elder (3.7 %), lay minister or pastor (3.1 %), church mother¹ (3.8 %), church committee chair (7.8 %), choir director (1.1 %), and Sunday school teacher (3.6 %). Because it was determined that it would not be appropriate to ask individuals who attended church very infrequently or not at all about their voluntary participation in the church, this item was asked only if the participant reported attending church services at least “several times a year” (i.e., the 4th point on an 8-point scale frequency of church attendance). Thus, participants were divided into three categories on the basis of their church attendance and leadership activity. Those who reported holding one or more of these positions were categorized as leaders. Those who attended church at least several times a year but did not report holding any of these positions were categorized as “attenders”. Finally, those who did not attend church frequently enough to be asked about leadership (i.e., less than several times a year) were categorized as “non-attenders.”

Demographics

Race, gender, age at baseline, and years of education reported at baseline were included as demographic controls. Note that these factors are treated as constant within persons across waves, whereas disability and leadership status are treated as time-varying.

Analyses

Two growth curve models were constructed, with the functional limitation index and functional limitation severity scale measures as outcomes. The sample for the functional limitation index model includes all study participants, whereas the sample for the severity scale includes only those who reported at least one functional limitation. Three points of observation are generally considered to provide adequate information to detect the presence of a linear effect (Curran et al., 2010) particularly when the analyses are based on a large sample of about 1,000 observations or more (Muthén & Curran, 1997) (whereas the present models for functional limitation and severity include 2,640 and 1,800 observations, respectively).

Growth curve analyses for this paper were conducted using hierarchical linear modeling (West et al., 2007). Using this method, each participant is treated as a cluster within which measurements at successive survey waves are nested. The purpose of the growth curve model is to estimate the mean slope of within-individual change in the outcome variable over the duration of the study. To do so, a hierarchical linear model is fit with the time index and intercept treated as random effects. The resulting model is based on weighted composites of between and within-person age effects and trajectory information for each participant, including those who were lost due to attrition before the final wave. In this case, time is indexed by age, rather than by time in the study. Examining change as a function of age is generally the most appropriate approach when, as is the case in this study, the initial ages of participants vary widely (Mehta & West, 2000).

One feature of the hierarchical linear modeling method is that it produces estimates of fixed effects coefficients for each model variable. These coefficients can then be applied to any set of hypothetical values for these variables to produce predicted mean scores on the outcome variable, which can be particularly valuable in visualizing the direction and magnitude of interactions between a number of model variables. This approach was used to provide illustrations of the differences between leaders, attenders, and non-attenders in terms of predicted mean trajectories of functional limitation (see Figs. 1, 2). Note that, because age rather than study time is used as the index, the estimated mean growth curve is presented across the full span of ages represented in the sample, rather the 4-year duration of the study. Analyses were conducted in SPSS 20 using the linear mixed modeling procedure, applying a variance components covariance matrix.

Fig. 1.

Mean growth curve for change in functional impairment during older adulthood for lay church leaders, non-leading regular church attenders, and non-attenders. Note: ADL/IADL Activities of Daily Living/Instrumental Activities of Daily Living scale. Predicted values calculated for White woman with high school education. The slope for church leaders differs significantly from attenders (p = .009) and non-attenders (p = .004); slopes for attenders and non-attenders do not differ significantly (p = .463)

Fig. 2.

Mean growth curve for change in impairment severity for lay church leaders, non-leading regular church attenders, and non-attenders. Note: Predicted values calculated for White woman with high school education. The slope for church leaders differs significantly from non-attenders (p = .018). The slope for attenders does not differ significantly from leaders (p = .124) or non-attenders (p = .173)

Model specification

Preliminary analyses showed that, although there was some evidence of a marginally significant quadratic effect in the mean age-related trajectory for functional limitations, this effect was not detectable within any of the three leadership categories, and the corresponding quadratic random effect proved to be redundant. Therefore, age-related change is modeled as a linear effect only for all outcomes. The model can be expressed in hierarchical notation as follows:

• Level 1 (observation):

  $$\begin{array}{l}{\mathit{disability}}_{ti}={\pi }_{0i}+{\pi }_{1i}{\mathit{Age}}_{\mathit{tie}}+{\pi }_{2i}{\mathit{Nonattender}}_{ti}\\ +{\pi }_{3i}({\mathit{Nonattender}}_{ti}\times {\mathit{Age}}_{ti})\\ +{\pi }_{4i}{\mathit{Attender}}_{ti}+{\pi }_{5i}({\mathit{Attender}}_{ij}\times {\mathit{Age}}_{ti})+e_{ti}\end{array}$$
• Level 2 (individual):

  $$\begin{array}{l}{\pi }_{0i}={\beta }_{00}+{\beta }_{01}{\mathit{Race}}_i+{\beta }_{02}{\mathit{Gender}}_i+{\beta }_{03}{\mathit{Education}}_i+r_{0i}\\ {\pi }_{1i}={\beta }_{10}+r_{1i}\end{array}$$

In these formulas, disability_ti is the score on the ADL/IADL measure for individual i at time t. At the observation level (level 1) intercept and age are random effects; π_0i is the within-person intercept, with random variation around the mean intercept β₀₀ given by r_0i and the fixed effects of the time-invariant covariates race, gender, and education (β₀₁Race_i, β₀₂Gender_i, and β₀₃Education_i). The age-related growth curve varies randomly between individuals, as given by the level 2 formula for π_1i. The variables Non-attender_ti and Attender_ti represent dummy variables corresponding to time-varying leadership category membership, with church leaders as the reference category. Consequently, the π_0i and π_1iAge_ti effects represent the mean intercept and age slope of disability change for church leaders, while the terms π_2iNonattender_ti and π_4iAttender_ti represent the difference in means for the other two categories respectively with the mean for leaders, and the corresponding age interaction terms represent inter-group differences in the mean rate of intra-individual change.

Results

Descriptive statistics for model variables by wave are presented in Table 1. The geographic distribution of the sample included 26.3 % living in the East, 38.4 % in the South, 21.0 % in the Midwest, and 14.3 % in the West. Respondents’ religious affiliations were 18.5 % Catholic, 16.8 % Mainline Protestant, 52.9 % Evangelical Protestant, 7.8 % other religion, and 4.0 % unaffiliated. The relatively high representation of individuals living in the South and affiliated with Evangelical groups is due to the oversampling of African Americans, who tend to belong disproportionately to those categories. Listwise deletion of cases with item nonresponse resulted in a loss of 2.70 % of observations. In total, the functional limitation model was based on 2,640 observations from 1,152 individuals. The limitation severity model, which included only those participants who had at least one ADL/IADL limitation, was based on 1,800 observations from 977 individuals.² Growth curve results for both models are presented in Table 2. Mean extent of disability was higher among African Americans than for Whites (b = 0.70, 95 % CI [0.32, 1.09], p < .001), higher among women than for men (b = −1.03, 95 % CI [−1.40, −0.67], p < .001), and was lower among those with more years of educational attainment (b = −0.08, 95 % CI [−0.14, −0.02], p = .005), all of which are consistent with previous research. Members of both other categories had higher mean levels of disability than leaders (non-attenders: b = 2.28, 95 % CI [1.87, 2.69], p < .001; attenders: b = 0.86, 95 % CI [0.50, 1.22], p < .001). There was a significant trajectory of age-related increase in functional limitations among church leaders (b = 0.07, 95 % CI [0.02, 0.13], p = .013), and interaction effects showed that this trajectory was significantly more rapid among members of both other categories (non-attenders: b = 0.10, 95 % CI [0.03, 0.17], p = .004; attenders: b = 0.08, 95 % CI [0.02, 0.15], p = .009). Supplemental analyses changing the comparison category showed that there was a significant mean difference between non-attenders and attenders (b = −1.42, 95 % CI [−1.73, −1.16], p < .001), but no significant difference in the trajectories of change within these groups (b = −0.02, 95 % CI [−0.07, 0.03], p = .463).

Table 1.

Descriptive statistics

	Wave 2 (N = 1,028)	Wave 3 (N = 969)	Wave 4 (N = 716)
White	543 (52.8 %)	523 (54.0 %)	391 (54.6 %)
Female	649 (63.1 %)	620 (64.0 %)	463 (64.7 %)
Education (years)	11.51 (3.41)	11.59 (3.34)	11.71 (3.24)
Age	77.50 (6.22)	78.89 (5.84)	80.30 (5.28)
Church leadership
Leader	176 (17.1 %)	171 (17.6 %)	137 (19.1 %)
Attender	509 (49.5 %)	462 (47.7 %)	359 (50.1 %)
Non-attender	343 (33.4 %)	336 (34.7 %)	220 (30.7 %)
Health outcomes
Functional limitations (ADL/IADL)	3.23 (3.60)	3.79 (3.81)	3.69 (3.68)
ADL/IADL ≥ 1	674 (65.6 %)	694 (71.6 %)	505 (70.5 %)
Limitation severity	13.08 (11.42)	14.44 (12.09)	14.65 (12.12)
Chronic conditions	2.59 (1.65)	2.79 (1.71)	2.78 (1.63)

ADL/IADL Activities of Daily Living/Instrumental Activities of Daily Living scale. Wave 1 is omitted because ADL/IADL was not included in the survey instrument

Table 2.

Growth curve results for the relationship of religious leadership status with change in disability and chronic medical conditions

	Functional limitationsa		Limitation severityb
	b (SE)	p	b (SE)	p
Fixed effects
Intercept	3.37 (0.43)	<.001	14.09 (1.57)	<.001
Race (Black)	0.70 (0.21)	<.001	1.92 (0.69)	.006
Gender (male)	−1.03 (0.18)	<.001	−2.57 (0.68)	<.001
Education	−0.08 (0.03)	.005	−0.43 (0.13)	<.001
Age slope	0.07 (0.03)	.013	0.17 (0.13)	.177
Non-attender	2.28 (0.21)	<.001	7.43 (0.84)	<.001
Attender	0.86 (0.18)	<.001	2.40 (0.76)	.002
Non-attender × age slope	0.10 (0.04)	.004	0.35 (0.15)	.018
Attender × age slope	0.08 (0.03)	.009	0.22 (0.14)	.124
Random effects variance
Residual	5.29 (0.20)	<.001	49.50 (2.13)	<.001
Intercept	5.95 (0.45)	<.001	44.24 (3.91)	<.001
Age	0.01 (0.008)	.146	0.21 (0.08)	.006

^aModel includes 2,640 observations from 1,152 individuals

^bModel includes 1,800 observations from 977 individuals with at least one functional limitation

Thus the results support the hypothesis that older adults holding positions of lay leadership undergo significantly less rapid increase in number of functional limitations, compared not only with those who do not regularly attend religious services, but also with fellow regular attenders who do not act as leaders. Figure 1 illustrates these results, depicting the mean within-person age trajectories for functional limitations by leadership category, and demonstrates that lay church leaders and regular church attenders without leadership positions tend to have similarly low levels of physical disability at the beginning of older adulthood, but diverge over time as attenders experience more rapid increase in disability. Those who attend church infrequently or not at all tend to have worse levels of disability throughout older adulthood, and experience mean rates of increase similar to attenders.

Mean ADL severity, among participants with at least one physical limitation, was higher among African Americans than for Whites (b = 1.92, 95 % CI [0.56, 3.28], p = .006), higher among women than for men (b = −2.58, 95 % CI [−3.90, −1.25], p ≤ .001), and lower among those with more years of educational attainment (b = −0.43, 95 % CI [−0.63, −0.23], p < .001). Church leaders showed no significant pattern of increase in limitation severity over time (b = 0.17, 95 % CI [−0.08, 0.43], p = .177). Members of the other two categories had higher mean levels of limitation severity (non-attenders: b = 7.43, 95 % CI [5.80, 9.07], p < .001; attenders: b = 2.40, 95 % CI [0.91, 3.90], p = .002), but only the non-attenders also had more significantly more rapid trajectories of increasing severity (b = 0.35, 95 % CI [0.06, 0.63], p = .018), whereas the attenders did not (b = 0.22, 95 % CI [−0.06, 0.49], p = .124). Supplemental analyses showed that the non-attender and attender categories differed significantly from one another in terms of the overall mean of limitation severity (b = −5.03, 95 % CI [−6.18, −3.86], p < .001), but not in the rate of change of limitation severity over time (b = −0.13, 95 % CI [−0.32, 0.06], p = .173).

Again, these results support the hypothesis that church leaders who suffer from physical limitations tend to experience less rapid increase in the severity of those limitations, not only compared with older adults who do not regularly attend church, but also with regular attenders who do not serve as leaders. Figure 2 illustrates the mean within-person age trajectories for limitation severity by leadership category. Note that although the age slope for attenders does not differ significantly from the non-significant slope for leaders, it does differ significantly from 0.

Discussion

The results of this study appear to indicate that older adults who occupy lay leadership positions in their religious groups experience slower average rates of functional decline, compared both with older adults who do not regularly attend church, and with those who are regular attenders but not leaders. An interesting contrast emerged between regular attendance and leadership and the timing of their association with functional status. Attendance appeared to have an important cross-sectional association, with both leaders and attenders who regularly attended church showing lower mean levels of disability than irregular attenders. But leadership appeared to be related more closely with the rate of within-person change over time; in fact, non-attenders and attenders showed rates of average decline that did not differ significantly from one another, and which were both considerably higher than the average rates of decline for leaders. This basic set of relationships was very similar both for both of the functioning outcomes measured in this study (number of functional limitations, and severity of functional limitations), as Figs. 1 and 2 illustrate.

These findings are consistent with theoretical expectations and the hypotheses derived from them. These results could be interpreted to reflect the contribution of voluntary leadership to successful aging, with continued group engagement providing a sense of purpose and positive self-image (Kim & Moen, 2002; Moen et al., 1992). Similarly, these findings can be seen as evidence that leadership may provide a path to generativity, contributing in turn to a sense of success in late life identity development and meaning (Bradley, 1997; Erikson, 1950). This general view suggests that the sense of personal worth engendered by providing group leadership provides the motivation to engage in positive health behavior in order to remain active in the role, as well as potentially reducing the experience of stress and depression associated with role loss, thus reducing their impact on physical health. Since the present study was not able to address this possibility empirically, more research is needed to test for a relationship between voluntary leadership and the development of a positive sense of self in late life.

Another possible interpretation of these findings, which it was also not possible to directly assess within the framework of the present study, may be derived from a social identity perspective. Perceiving oneself as a leader, and being perceived as such by other group members, may serve to enhance feelings of group prototypicality, and hence to enhance both the extent of social support received from the group, as well as one’s own ability to draw internal strength from the group identity (Hogg et al., 2004). This interpretation is consistent with previous research on the relationship between religious group prototypicality and the extent of benefits derived from the group (Hayward & Elliott, 2009, 2011), as well as with findings that certain lay leaders receive more social support (Krause et al., 1998) and engage in more positive forms of religious coping (Pargament et al., 2001) than other members. This convergence of internal and external forms of support might help to explain the dramatic differences observed between leaders and regular members in this study. Conversely, it is also possible that group leaders are simply more visible targets for receiving church-based tangible support from others, which previous research has found to be linked with slower progression of functional decline (Hayward & Krause, 2012). Again, since these factors were beyond the scope of this study, more research is needed to examine the social and psychological mechanisms that may help to account for these findings (such as perceived prototypicality and group cohesiveness).

To the extent that these results imply that lay leadership positions may contribute to health benefits distinct from those experienced by the rank and file, they raise some intriguing questions about the potential for between-group differences. Even within the scope of the Christian groups covered by this study, there is a wide degree of variation in the extent of opportunities that churches may provide for leadership. Groups with very hierarchical structures and relatively many organizational resources, such as the Catholic Church, may have proportionally few lay leaders. While limiting participation, this structure may also enhance the prestige of these leaders to other members. At the other end of the spectrum, small congregations with egalitarian structures may make it easier to become a leader, but provide fewer resources and perhaps have more potential for negative interaction. Other religious traditions are likely to have an even wider array of practices regarding voluntary lay leadership, and thus more research is called for to bridge these groups. Furthermore, these results may suggest that some of the association generally noted between religious participation and health might possibly be explained by making the opportunity to hold leadership roles available to individuals who would otherwise have less access to them. While beyond the scope of this paper, this question provides another avenue for potential future research.

An obvious limitation of this study lies in the question of directionality. Older adults who are experiencing more rapid physical decline may be less likely to take on new leadership roles, and more likely to drop out of existing ones. Mitigating against this to a certain extent is the finding that rates of decline were equally more rapid among both the comparison categories of infrequent attenders and regular (but non-leader) attenders. Older adults with dramatically worsening functional status might be expected to tend to stop attending services regularly as well, so if that was the only factor driving the observed results, we might expect regular attenders to more closely resemble one another, regardless of their leadership status; instead there appears to be a very distinct path related to leadership status, even among those who are healthy enough to continue attending services regularly. Nevertheless, in the absence of a randomized experimental design, the direction of the relationship observed here cannot be definitively established. Other limitations include the availability of a maximum of three observations per participant over a period of 4 years. A larger number of data points covering a longer longitudinal period would improve our ability to test for curvilinear patterns of change over time. Additionally, the present study lacked data on leadership positions that the participants may have held in other institutions outside the church, making it impossible to examine whether there are any differences between religious and secular roles. Related to this, although the general type of church leadership positions held by participants was known, categories such as “Sunday School teacher” may cover a broad range of activities, and combined with the limited sample size of individuals holding each specific type of position, this makes it difficult to identify any potential differences among religious leadership roles.

Functional impairment in older adulthood is a problem with important implications not only for the well-being and longevity of older adults themselves, but also for family members who often act as caregivers and, given that health care costs are estimated to triple when an individual is functionally disabled, for society in general (Freedman et al., 2002). Consequently, identifying existing means that may contribute to slowing the rate at which physical functioning deteriorates in late life may be valuable in suggesting strategies that both increase the quality of life in old age while also reducing the costs of providing care. Since religious groups are already among the organizations with which many older adults are most actively involved, they are a good place to look, both as sources of insight into effective strategies, and as potential sites to target future interventions. We hope that these findings will encourage additional research into the possible impact of lay leadership involvement on health outcomes for older adults, and may be an initial step towards developing new methods of harnessing the power of the group to improve quality of life.