Introduction

According to China’s 2023 Civil Affairs Statistical Bulletin, individuals aged 60 and above account for 21.1% of the population, with those aged 65 and older making up 15.4%, underscoring the profound and accelerating trend of population aging. In response, the Communist Party of China and the Chinese government have formulated a comprehensive national strategy to address these challenges. This strategy coordinates efforts and integrates services for older adults across the domains of care, healthcare, social security, participation, and rights protection.

Engaging older adults in volunteering activities is an effective way to enhance their psychological well-being and social integration. Volunteering involves altruistic actions aimed at helping others, communities, or causes without expecting material rewards1. Volunteering can be viewed as a human-created, sustainable resource that nourishes society and is, in turn, replenished by it2. Driven by particular incentives, older adults engage in volunteering, aiming to achieve a sense of fulfillment through continuous social participation3,4. Community volunteering service involves voluntary participation in activities that benefit the local community. It can enhance social, physical, and mental health, promoting a productive and healthy aging process. The scope of community volunteering services is relatively broad, encompassing activities such as community security patrols, environmental sanitation, neighborhood dispute mediation, and more. For older adults, community volunteering, as a convenient way, can strengthen their social connections. The peer effect suggests that these services can encourage broader participation, fostering a positive, active impact on the community5.

Engaging in local community volunteering, such as mutual-aid initiatives, is often more feasible for older adults than joining formal organizations. Such volunteering activities have greater development potential, because they allow for more accessible and practical contributions within their own communities, compared with the structured nature of formal settings6. The majority of positive impacts reported by older volunteers are linked to health benefits and the enhancement of knowledge and skills in helping others7. It is positive to transform older persons from care recipients to service providers and caregivers and to encourage older adults to utilize their knowledge and skills to contribute to and integrate into community life8.

In light of this background, this paper poses two questions: What factors influence older adults’ participation in community volunteering services? How do these diverse factors interact and operate in unison?

Literature review

Current research on factors influencing older adults’ volunteering participation has been diverse, with some studies having focused on individual characteristics like economic status, self-rated health, and education level9. It has also been suggested that volunteering is influenced by various factors related to interpersonal communication, such as serving the community, enhancing well-being, gaining recognition from friends and family, and alleviating loneliness10. The community plays a key role as a platform for member activities and empowerment11. While previous research has categorized influencing factors into individual characteristics, motivations, and social norms, more specific theoretical frameworks are still needed12.

Musick and Wilson’s capital theory provides a foundational framework for understanding older adults’ volunteering participation. They identified three key criteria: Can they participate? Do they want to? Is there a norm influencing their involvement1? Capital theory encompasses three main aspects: human capital (skills and resources invested in work), social capital (social networks as a productive resource), and cultural capital (shared values, attitudes, and behaviors within a group). Higher levels of capital typically increase the likelihood of older adults’ volunteering participation. This paper extends Musick and Wilson’s capital theory by integrating human, social, and cultural capital frameworks while examining the role of media capital in older adults’ volunteering in China.

Human capital: Human capital refers to the skills and knowledge embedded in individuals through investment, capable of generating future returns13. It includes elements such as education, health, and technical expertise. Most human capital indicators have demonstrated predictive effects on volunteering participation14. Previous studies indicate that older adults with better physical health and higher socioeconomic status demonstrate higher volunteer engagement15. Subjective health and economic perceptions also moderately predict participation9. Schultz also highlighted education’s critical role in macro-level human capital analysis16. Higher educational attainment not only creates enabling conditions for older adults to engage in volunteering activities but also cultivates more positive attitudes toward such participation17,18. In summary, this paper operationalizes economic factors, health-related factors, and educational attainment as proxy variables for human capital.

Social capital, originating from sociology, was defined by Bourdieu as resources derived from a network of relationships, both actual and potential19. Lin Nan defined social capital as resources embedded in a social structure that can be drawn upon or mobilized for purposeful action20. Coleman identified five forms of social capital: obligations and expectations, information networks, norms and penalties, authority relations, and social organization21. The social capital held by older adults predominantly facilitates volunteering participation through three key mechanisms: normative expectations that shape behavioral patterns, information networks enabling resource accessibility, and institutional affiliations that formalize engagement opportunities. Social capital indicators, including trust, neighborhood support, and informal social networks, significantly increase the likelihood of participating in volunteering activities22. Building companionship with new friends generally makes older adults feel happier and enhances their life satisfaction23. Information from peers as well as practical support also enhances older people’s participation in volunteering24. Additionally, communities with more resources can provide stronger platforms and organizations, boosting volunteering engagement25. A greater endowment of social capital leads to higher participation in community volunteering services among older adults24. This paper argues that the impact of social capital on older adults’ volunteering participation should be examined from both individual social connections and community-level perspectives.

Cultural capital: Cultural capital was defined by Bourdieu as symbolic resources embedded in cultural practices, which can be measured through their external expressions26. Wilson and Musick further expanded this definition by incorporating moral dimensions27. The use of community services can sustain and enhance the well-being of older adults in a familiar environment, and their actual satisfaction with these services influences participation levels28. Conversely, community deprivation, characterized by scarce resources and weak infrastructure, can erode the cohesion and norms that support participation29. Wilson, however, proposed a competing hypothesis: while thriving neighborhoods foster participation, those in decline might also stimulate collective action due to a "crisis-induced awareness"30. Comparative studies have shown that older adults who receive arts training significantly maintain and enhance their volunteerism compared to those without such training31, which reflects that recreational activities can promote volunteering participation. In addition, the attitude of active aging was also a significant predictor32. This orientation encourages older volunteers to value their services and associated self-worth33. After comparing various factors influencing volunteering, active aging was found to have a significant impact on volunteering participation34,35. Cultural capital accumulation, particularly influenced by personal values, expands volunteering opportunities36. This paper uses older adults’ community satisfaction, recreational activity engagement, and views on active aging as predictors of their participation in community volunteering.

Media capital: Media capital, as conceptualized in this study, finds its theoretical roots in the knowledge gap hypothesis. This hypothesis posits that disparities in knowledge across social strata stem from differential capabilities in acquiring and applying information37. Ettema refined the conceptualization in subsequent research, highlighting the dynamic processes of knowledge gaps and arguing that media systems amplify the level of information acquisition and utilization among groups38. This theoretical framework posits that contemporary mediated technologies exacerbate systemic inequalities through differential communication efficiency39 rather than mitigating information polarization. While digital literacy has become essential for social engagement40, older adults face systemic barriers in information access and use, impeding their social integration. Against this background, this paper extends the capital theory framework by incorporating media capital’s role in older adults’ community volunteering participation. Media capital originates from virtual communities where members connect via platforms, texts, and symbols41,42. Media use—including traditional and new media—significantly affects volunteering participation43. Frequent local media engagement (e.g., newspapers, television) significantly enhances social contact44, fostering civic participation and volunteerism45,46. However, significant variations in older adults’ adoption of emerging digital media have generated competing theoretical explanations regarding the social impacts of these disparities47. Competing theoretical explanations exist regarding the impact of internet use on older adults’ volunteering participation. Some studies have found that internet access increases the likelihood of participation in both formal and informal volunteering activities48. Other studies have reported that internet use negatively affects volunteering participation49. Notably, a particularly potent catalyst of internet use is particularly pronounced among older adults50. This dimension encompasses both traditional and new media use. Traditional media mediate collective action within structurally homogeneous groups, while online platforms facilitate cross-sectional information dissemination, enhancing emotional connectivity and communicative efficiency51.

Complex effects: Existing studies mainly analyze the net effects of individual variables. However, the potential interaction effects or complex relationships among explanatory variables require further exploration. Social capital theory highlights the significant impact of education on an individual’s social capital52. Coleman’s foundational work further posits that social capital is instrumental in the formation of human capital53. This suggests that these two forms of capital are not independent but can produce synergistic effects. Cultural capital characteristics, such as ability, intelligence, values, and social norms, may also influence the relationship between education and volunteering service54. Similarly, some studies have found that the use of media capital influences social capital and the social integration of members55.

While regression analysis can incorporate interaction terms or higher-order polynomials to capture certain combined and non-linear effects among variables, it remains limited in its ability to fully characterize the complex interplay of multiple factors56,57. Specifically, interaction terms—though useful—can only capture isolated interactive effects and fail to represent the full complexity of variable relationships. Similarly, introducing polynomial terms may help model non-linearity, but at the cost of increased model complexity and reduced interpretability. Therefore, analytical models based solely on these conventional econometric techniques remain insufficient to fully address the core research questions of this study.

This study examines the mechanisms influencing older adults’ community volunteering participation through four capital dimensions: human, social, cultural, and media capital. To this end, it employs a dual-method approach (as illustrated in Figs. 1 and 2) that integrates logistic regression with fsQCA. This combined analytical strategy allows for an examination of both linear relationships and configurational synergies among variables. Additionally, subgroup analyses by urban–rural residence, gender, and age cohort are performed to identify population-specific impacts and explore differentiated mechanisms across subgroups.

Fig. 1
figure 1

Logistic regression model.

Full size image
Fig. 2
figure 2

FsQCA model.

Full size image

Research design

Research data

This paper utilizes data from the CLASS 2020 study (China Longitudinal Aging Social Survey), a nationwide survey targeting older adults aged 60 and above in China. The CLASS survey encompasses seven major domains: basic demographic information, health and related services, socioeconomic status, retirement planning and social support, psychological perceptions, family relationships, and daily activities/fitness regimens. Employing a stratified multi-stage probability sampling design, the study selected county-level units (including counties, county-level cities, and districts) as primary sampling units, with village/neighborhood committees serving as secondary units. The survey coverage extended to 30 provincial-level administrative divisions in mainland China. The 2020 follow-up survey yielded a total of 11,398 individual samples. After variable selection, processing, and the deletion of samples with missing values, the total effective sample size for this paper is 9,473.

Selection of variables

Dependent Variables

The study selects the following community volunteering activities as explanatory variables to assess older adults’ participation: community policing patrols, caring for older individuals or children (excluding their own), environmental health protection, mediation of neighborhood disputes, professional and technical volunteer services, and educating the next generation (excluding their own children). In terms of the overall participation of older adults in community volunteering, the participation rate is 25%, indicating a relatively low overall level of involvement.

In the regression analysis section, the paper treats the dependent variables as dichotomous, with 0 indicating non-participation and 1 indicating participation. In the fsQCA analysis, the variables are treated as participation intensity data, which are summed to better capture the relationship between the dependent and independent variables.

Independent Variables

Human Capital: The article selects "How do you think of your financial situation when compared to the people around you?" This question is designed to assess the subjective economic status of older adults. The range of values is 1 (low), 2 (about the same), and 3 (better). Meanwhile, in order to better observe the actual economic level of older adults, this paper uses the number of self-owned properties of older adults as the proxy variable. And then this article selects "How has your health changed from last year?" This question item reflects the subjective health status. The range of values is 1 (low), 2 (about the same), and 3 (better). To incorporate measurements of older adults’ actual health status, this study operationalizes their health level using a proxy variable constructed through negative scoring based on diagnosed conditions. Specifically, the presence of nine chronic diseases was summed and inversely coded, resulting in a composite score ranging from −9 to 0. Higher values (closer to 0) indicate better health status, with each unit decrease reflecting additional morbidity burden. The third variable measuring human capital is the level of education, which is 1 for illiterate/2 for middle school and below/3 for high school and above.

Social Capital: One is “social contact”. The following items are selected to reflect the Social Connection score:1. How many family members/relatives do you see or contact at least once a month? 2.How many family members/relatives are you comfortable talking to about your personal matters? 3.How many family members/relatives are available to help you when you need it? 4. How many friends do you see or talk to at least once a month? 5.How many friends are you comfortable talking to about your personal matters? 6.How many friends are available to help you when you need it? The range of values was 0–5, and then the above 6 questions are averaged. The other is community facilities, serving as an indicator of the activities provided by the community for its older residents. A single point is allotted for each of the community amenities, with the total being calculated at the conclusion of the evaluation process.

Cultural capital: The predict variables are three items. One is that respondents’ satisfaction with the following situations within their community (village)—road conditions/fitness/activity venues/security environment/environmental sanitation/atmosphere of respect for older adults/reflection of the work ability of the neighborhood committee staff. The range of values is 1–5, and the above six items are averaged. Next, variables are selected to reflect participation in recreational activities. The one is "participation in recreational activities". Recreational activities are selected to be reflected in the following questions. The frequency of participation in activities such as singing/playing musical instruments/playing mahjong/chess/cards and square dancing is selected, and the range of values is 0–3. Three items are selected to reflect older people’s perception of active aging.:1. If I had the opportunity, I would be happy to participate in some of the work of the community council.2. I often want to do something more for the community.3. I feel that I am still a useful person to the community. The range of values was 1–5, and the three items were averaged to reflect the level of cultural capital in the affective-attitudinal dimension. The article takes the above 3 question items and averages them.

Media capital: Media capital: the frequency of use of newspapers, magazines, radio, and television is chosen as traditional media use, and the frequency of use of the Internet and customized messaging on cell phones is chosen as new media use. The above two variables reflect the medium vehicle used by older adults during their information interaction with the outside world, 0-unused, 1-low frequency of use, 2-medium frequency of use, and 3-high frequency of use.

Controlled Variables

Existing studies have explored the role of individual characteristics on the voluntary participation of older adults. Considering the study’s specific population, eight controlled variables were included: gender, age, marital status, and whether they live with their children, political affiliation, religious belief, retirement or not, residential areas. Additionally, urban–rural residence, gender, and age cohort were selected as controlled variables to examine potential moderating effects of individual factors on volunteering participation across subgroups (Table 1).

Table 1 Descriptive statistics.
Full size table

Research method

Logistic regression

In the regression analysis part, since the dependent variable is dichotomous, this paper uses logistic regression model for the analysis. If the behavior of older adults participating in community volunteering occurs, Y = 1, and if it does not occur, Y = 0. In logistic regression models, odds ratios (OR) quantify the multiplicative change in the odds of outcome Y per unit increase in predictor X, representing the ratio of favorable to unfavorable odds across exposure levels. The model is set up as follows:

$$P = \left( {Y = 1{|}X} \right) = e^{{\left( {\beta 0 + \beta 1X1 + \beta 2X2 \ldots + \beta nXn} \right)}} /1 + e^{{\left( {\beta 0 + \beta 1X1 + \beta 2X2 \ldots + \beta nXn} \right)}}$$
(1)

Studies on participation determinants rely on linear analysis58, which estimates net variable effects controlling for covariates. Regression models assume symmetric relationships under probabilistic distributions, focusing on average causal effects of independent variables (X) on outcomes (Y). However, regression estimates represent conditional net effects by isolating X’s influence while controlling for other covariates Xi. These symmetric approaches emphasizing individual variable effects cannot capture configurational relationships or combinatorial causation, as outcomes typically arise from interactive processes rather than isolated factors. To address this limitation, this paper integrates logistic regression with fuzzy set qualitative comparative analysis (fsQCA).

FsQCA

Rooted in logic and set theory59, QCA is a methodological approach designed to identify necessary and sufficient conditions for an outcome60. It excels at analyzing causal complexity, characterized by asymmetric relationships (where the conditions leading to an outcome’s presence may differ from those leading to its absence) and equifinality (where multiple, distinct configurations of conditions can lead to the same outcome)61.

QCA encompasses three core approaches: crisp-set QCA (csQCA), multi-value QCA (mvQCA), and fuzzy-set QCA (fsQCA). This study employs fsQCA, which overcomes the categorical limitations of crisp sets through graded membership scores and set-theoretic operations. This makes it particularly suitable for analyzing graded phenomena and the complex causal mechanisms prevalent in social science research. By retaining data richness while quantifying the gradations between full membership and non-membership62, fsQCA therefore provides a precise analytical tool for interpreting real-world asymmetric relationships and equifinality.

$$Consistency\left( 1 \right) = \sum min\left( {Xi,Yi} \right)/\sum Yi$$
(2)
$$Coverage\left( 1 \right) = \sum min\left( {Xi,Yi} \right)/\sum Xi$$
(3)

The formulas for the consistency and coverage of sufficient conditions are as follows:

$$Consistency\left( 2 \right) = \sum min\left( {Xi,Yi} \right)/\sum Xi$$
(4)
$$Coverage\left( 2 \right) = \sum min\left( {Xi,Yi} \right)/\sum Yi$$
(5)

where Xi means the membership of the i th case in the condition set X; Yi means the membership of the i th case in the condition set Y. The value of consistency and coverage is [0,1], and the closer the consistency value is to 1, the stronger the explanatory force, and there is no specific criterion for coverage.

The logic for analyzing necessary conditions is presented in Formulas 2 and 3. A condition (or antecedent variable X) is considered necessary for an outcome (Y) when X is a superset of Y63. This relationship is assessed by two key metrics:

Consistency indicates the extent to which instances of high Y are consistently accompanied by high X. In other words, it measures how reliably the outcome occurs only when the condition is present.

Coverage gauges the empirical strength or importance of this necessary condition by showing how much of the outcome (high Y) is explained by the condition (high X). A condition is typically deemed necessary if its consistency score exceeds 0.9 and its coverage is above 0.5.

In contrast, the logic for sufficiency analysis is detailed in Formulas 4 and 5. A configuration of conditions (X) is sufficient for the outcome (Y) when it is a subset of Y. The metrics here are interpreted differently:

Consistency measures the proportion of cases exhibiting a specific configuration of antecedent conditions (X) that also display the outcome (high Y).

Coverage assesses the empirical relevance of this sufficient configuration by indicating how much of the outcome (high Y) is explained or “covered” by it.

For sufficiency analysis, the conventional consistency threshold is 0.7564. During the analytical procedure in the fsQCA software, it is crucial to set a frequency threshold, which should be increased appropriately with larger sample sizes. Additionally, the Proportional Reduction in Inconsistency (PRI) consistency score should be set, typically at a minimum of 0.6, to resolve situations where a causal configuration might simultaneously lead to both the presence and absence of the outcome (high Y and low Y)60. Regarding the coverage threshold for sufficiency analysis, there is no universally defined standard in the existing literature on this topic65.

In the process of sufficiency analysis, different conditional groupings may be formed, which should be explained, and the analytical model is:

$$Volunteering = \left\{ {X1,X2,X3 \ldots Xn} \right\}$$
(6)

Existing literature, however, emphasizes competing methodological interpretations by highlighting QCA’s unique strengths in analyzing complex causation. This study transcends this comparative framework by exploring complementarity: logistic regression tests the probability of outcome Y, while fsQCA identifies diverse condition configurations leading to high-probability Y occurrences.

Much of the existing methodological literature positions QCA in contrast to regression-based methods, rightly highlighting its unique strength in analyzing causal complexity66. This study moves beyond this dichotomy by demonstrating their complementarity: logistic regression estimates the net effect of any single variable on the probability of the outcome, whereas fsQCA uncovers the multiple, distinct configurational paths that lead to the outcome’s (high Y or low Y) occurrence.

Regression analysis

Logistic regression was conducted to analyze the factors influencing older adults’ participation in community volunteering, with the regression results shown in Table 2. The analysis below discusses the impact of each form of capital on older adults’ participation in community volunteering service.

Table 2 Logistic regression.
Full size table

Baseline regression

As shown in Model 1, the logistic regression analysis identifies several factors significantly associated with community volunteering participation among older adults, after controlling for other variables in the model.

Human Capital: Subjective economic status is not a significant predictor. However, property ownership is associated with a 20.8% higher likelihood of participation (95% CI: 1.065–1.370, p < 0.01). A one-unit improvement in subjective health increases the probability of volunteering by 28.8% (95% CI: 1.148–1.445, p < 0.01), while better objective health—indicated by fewer common diseases—is associated with a 12.8% rise in participation likelihood (95% CI: 1.075–1.183, p < 0.01). In terms of education, compared to those with no formal education, older adults with middle school education or below show a 20.7% higher probability of participation (95% CI: 1.056–1.380, p < 0.01), and those with high school education or above show a 21.7% higher probability (95% CI: 1.001–1.481, p < 0.05).

Social Capital: A one-unit increase in social contact raises the likelihood of participation by 10.2% (95% CI: 1.038–1.169, p < 0.01), and greater community facilities completeness increases it by 6.9% (95% CI: 1.031–1.107, p < 0.01).

Cultural Capital: Contrary to expectations, higher community satisfaction is associated with a 34.8% reduction in the likelihood of volunteering (95% CI: 0.598–0.710, p < 0.01). In contrast, participation in recreational activities and stronger active aging awareness are linked to 32.4% (95% CI: 1.227–1.428, p < 0.01) and 23.9% increases in participation probability (95% CI: 1.172–1.309, p < 0.01), respectively.

Media Capital: The use of traditional media and new media is associated with 25.5% (95% CI: 1.141–1.381, p < 0.01) and 7.2% increases in the probability of volunteering (95% CI: 1.008–1.140, p < 0.05), respectively.

Moderating effect regression

In order to examine the differences in factor impacts among different groups, this paper conducts a heterogeneity analysis and discusses in detail the characteristics of the impact factors under the differences between urban and rural areas, gender, and age segments. Model 2-Model 7 shows the results of moderating effect.

Human capital: Subjective economic status is a significant predictor only among rural older adults, with each unit increase associated with an 18.4% higher likelihood of volunteering participation (95% CI: 1.039–1.348, p < 0.05). In contrast, the positive association between objective economic status and volunteering participation is consistently significant in all systematically examined subgroups, with the exception of the male subgroup.

Subjective health status exhibits a strong urban-specific effect: each unit increase raises volunteering participation by 61.5% in urban settings (95% CI: 1.364–1.911, p < 0.01), whereas the association is non-significant in rural areas. Across gender and age groups, improvements in subjective health consistently led to a notable increase in community volunteer participation rates. Objective health status is significantly associated with volunteer participation across both urban–rural and gender subgroups. However, a clear demarcation in age disparities emerges: among young-old adults (aged 60–69), improvements in objective health show no significant association with volunteer participation; while among older-old adults (aged 70 and above), such improvements are strongly associated with a 20.5% increase in the likelihood of participation (95% CI: 1.124–1.292, p < 0.01).

Analysis of education levels reveals a heterogeneous pattern. The positive association is observed predominantly in specific subgroups: it is more evident in rural areas compared to urban settings, and appears stronger among women, while not reaching statistical significance among men. Compared to illiterate peers, rural older adults with a middle school education or below show a 22.7% higher probability of participation (95% CI: 1.037–1.450, p < 0.05). A clear gender-specific pattern is observed: among women, both middle school and below are associated with a 34.5% higher participation probability (95% CI: 1.118–1.619, p < 0.01) and high school and above with a 47.5% higher participation probability (95% CI: 1.102–1.976, p < 0.01), whereas no statistically significant associations are found among men across all education levels.

Social capital: Social capital analysis reveals that stronger social contact increased volunteering participation probabilities by 8.7% for rural older adults (95% CI: 1.002–1.180, p < 0.05), 13.6% for urban counterparts (95% CI: 1.040–1.241, p < 0.01), 17.5% for women (95% CI: 1.074–1.284, p < 0.01), and 17% for young-old adults (60–69) (95% CI: 1.080–1.267, p < 0.01), while showing no significant effects among men and older-old adults (70 +). Community facilities completeness significantly predicts participation only in the rural men, and in the older-old adults subgroup.

Cultural capital: The results for cultural capital indicators—community satisfaction, participation in recreational activities, and active aging attitudes—are broadly consistent with the baseline regression, showing generally stable associations with volunteering participation.

Media Capital: Traditional media use exerts stronger and broader effects across subgroups. For every one-unit increase in the intensity of traditional media use, the probability of participating in volunteering activities increases by 16.7% among rural older adults (95% CI: 1.010–1.347, p < 0.05) and by 39.3% among their urban counterparts (95% CI: 1.218–1.594, p < 0.01). While traditional media shows no significant impact on female groups, it increases male participation by 41% (95% CI: 1.234–1.611, p < 0.01). Among age groups, traditional media use raises participation by 26.2% for young-old adults (60–69) (95% CI: 1.106–1.440, p < 0.01) and 23.3% for older-old adults (70 +) (95% CI: 1.071–1.418, p < 0.01). In contrast, new media use exhibits a more selective pattern, and rural groups, women, and older-old adults play a significant role in boosting volunteering participation.

Qualitative comparative analysis

The above regression analysis examines the net effect of human capital and social capital on the outcome variable. It is challenging for a single factor to provide a comprehensive explanation of the outcome variable. This section is analysed using the fsQCA60. Although certain factors appear insignificant in regression analysis, theory-driven and sample-oriented QCA analysis still provides strong evidence indicating their potential as peripheral or even core conditions67,68.

Calibration

In fsQCA, calibration is the process of assigning set membership scores to cases64. This study adhered to the principles for effective set-theoretic analysis as discussed by Greckhamer et al.69, which emphasize (1) clearly defining the sets representing outcomes and antecedent conditions, (2) using theoretical and substantive knowledge to set the thresholds, and (3) transparently reporting all chosen thresholds—a process described as being both "semi-conceptual and semi-empirical." Guided by these principles, we implemented a hybrid calibration strategy: Due to the presence of a substantial number of zero values in the assignment of membership degrees for both outcome and antecedent variables as fuzzy sets, this study adopted the mean as the crossover point between full non-membership and full membership. For continuous variables, we applied the direct calibration method62, setting the full membership, crossover, and full non-membership anchors at the 95th percentile, mean, and 5th percentile, respectively; for ordinal categorical variables such as education level, we employed direct assignment, converting ordered categories into fuzzy membership scores of 0.33, 0.66, and 162. This approach was designed to align fsQCA’s theory-driven nature with the distinct distributional characteristics of the data (Table 3).

Table 3 The calibration of variables.
Full size table

Necessity analysis

Necessity analysis explores whether a specific condition is always present when the outcome occurs. A condition is considered necessary for the outcome if it demonstrates a consistency level exceeding 0.9 in conjunction with a coverage level that must also be above 0.570. As shown in Table 4, low housing property quantity emerged as a necessary condition for low community volunteering participation. With the exception of this, all other conditional variables influencing older adults’ volunteering participation had consistency levels below 0.9, indicating the causal complexity of factors requires further examination of configurational patterns.

Table 4 Necessity analysis.
Full size table

Truth table construction

Sufficiency analysis is central to conducting a qualitative comparative analysis, aiming to explore the sufficiency of different configurations for the formation of outcome variables71. Due to the large sample size, the study set the frequency threshold at 5, the consistency threshold at 0.95, and the PRI consistency to be greater than 0.6. Based on this analysis, the complex solution, intermediate solution, and parsimonious solution for the sample of community volunteering participants were obtained.

Overall, the configurational analysis demonstrates strong explanatory power, with an overall consistency of 0.913—well above the well-established threshold of 0.75. While there is no universal standard for coverage, the overall solution coverage of 0.262 indicates that the identified configurations account for 26.2% of the cases exhibiting high volunteering participation. This level of coverage meaningfully corresponds to the generally low rate of community volunteering observed among older adults.

This study utilizes the intermediate solution to identify configurational conditions and the parsimonious solution to determine core conditions. As shown in Table 5, analysis of sufficient configurations reveals seven distinct pathways that form two overarching patterns promoting volunteering participation among older adults. Among these, level of education, social contact, and participation in recreational activities consistently served as conditions across all seven configurations. Partial absence of other conditional variables did not preclude the occurrence of high volunteering participation.

Table 5 Truth table.
Full size table

Through the identification of core conditions, the seven configuration pathways can be broadly categorized into two potential types. Drawing on relevant scholarly work63,72, this research seeks to provide a visual interpretation of these configurational types. The first type, termed the individual capacity type (Path 1, Path 2, and Path 3), is illustrated in Fig. 3. This configuration pattern is primarily characterized by the core presence of human capital conditions. A common feature observed in these pathways is a relatively high level of resource endowment, which may enable older adults to contribute based on their capacities. Through volunteering, they potentially engage in community building and realize personal value in later life.

Fig. 3
figure 3

FsQCA’s individual ability type. Red represents the core condition, orange stands for the peripheral condition, blue indicates the missing core condition, light blue denotes the missing peripheral condition, and colorless refers to the condition not included in the configuration.

Full size image

As shown in Fig. 4, a second potential type can be characterized as social integration (Path 4, Path 5, Path 6, and Path 7), which is distinguished by the joint presence of social, cultural, and media capital. This pattern appears to capture older adults who, while potentially experiencing a gradual withdrawal from formal productive roles, maintain a desire for social connection. These individuals may seek information, expand their social networks, and strengthen ties in both physical and digital spaces, with volunteering emerging as a potential pathway for fulfilling these integration needs.

Fig. 4
figure 4

FsQCA’s social integration type. Red represents the core condition, orange stands for the peripheral condition, blue indicates the missing core condition, light blue denotes the missing peripheral condition, and colorless refers to the condition not included in the configuration.

Full size image

Following Xu’s approach73, this study attempts to further explore potential configurational differences of fsQCA across urban–rural, gender, and age subgroups. The results of the grouped truth table are presented in Table 6. Key findings reveal the following differential configurations:

Urban–rural comparisons indicate differing core conditions associated with high volunteering participation: social, cultural, and media capital appear central among rural older adults, whereas human capital tends to function as the core condition in urban contexts.

Gender-based configurational comparisons reveal that high participation among older women is linked to more stringent combinations of multiple forms of capital. In contrast, configurations for older men appear more flexible, in some cases accommodating the absence of certain conditions—such as health—while still associating with high participation. This aligns with regression results, which also indicated fewer consistent predictors for male participants.

An interesting pattern also emerges across age groups. Adults aged 60–69 are represented by a limited number of configurations, primarily characterized by high individual capital. By comparison, those in the older cohort show greater diversity, with four distinct configuration pathways identified. While two of these largely overlap with those of the younger group, the other two suggest that high participation may occur even in the absence of some typical enabling conditions.

Discussions

Based on the above review and configurational analysis results, it is evident that the two methods yield partially consistent conclusions. Human capital exerted stronger effects in both overall and subgroup regression analyses (with high OR values in the overall model), demonstrating a stronger predictive effect on older adults’ volunteering participation. In configurational analysis, human capital served as a core driver in configurations 1–3, while partial human capital variables also functioned as core conditions in configurations 4–7.

The complementarity between the two methods is further evidenced in the regression analysis: although the odds ratios (ORs) for some dimensions were not statistically significant, their effect sizes were non-negligible. This suggests that the focus should extend beyond estimating their average net effect to investigating the outcomes produced by their configurations. fsQCA confirmed that these very conditions play peripheral or even core roles in different causal recipes. This finding suggests two key inferences: First, subjective economic evaluations may operate differently in predicting the occurrence of volunteer behavior versus high—intensity volunteering participation. Second, fsQCA demonstrates significant advantages in identifying complex conditional pathways. Second, community satisfaction — which exhibited a negative association with volunteering participation in the logistic regression — appears in most fsQCA configurations as a contributing or even core condition. Additionally, the configurational analysis suggests that the absence of certain conditions may be compensated for by the presence of others in forming pathways associated with high participation. This divergence aligns with the established methodological distinction between the two approaches: while regression analysis estimates average net effects, fsQCA reveals how different configurations of conditions may lead to the same outcome, thereby capturing causal heterogeneity across cases73.

The discussion of logistic regression

The findings of the regression analysis are further discussed as follows.

Human capital: Objective economic factors significantly impact older adults’ community volunteering in the full sample regression analysis, though income self-assessment shows no significant effect. This aligns with previous studies, which suggest that economic factors do not directly influence volunteering participation when considering the cognitive dimension of older adults74. In the rural group, income self-assessment proves significant, suggesting that rural older adults make decisions based on their subjective economic perceptions. Economic disparities between urban and rural areas mean older adults in rural regions still face constraints from basic needs. Additionally, self-assessed health does not fully align with objective health indicators18. Notably, older adults with poorer objective health tend to withdraw from social activities. This suggests that while subjective health perception matters, the actual objective health level plays a more significant role in influencing social engagement. The level of education also influences older adults’ participation in community volunteering. This effect of education arises from the outcomes of a specific socialization process75. Older adults may be withdrawing from society, yet their social experiences remain rich, and their skills are valuable. Emphasizing the preservation and cultivation of resources in later life is essential.

Social capital: Stronger social ties increase the likelihood of older adults engaging in community volunteering, while well-equipped community facilities further encourage their participation. Establishing meaningful relationships with new acquaintances often leads to greater happiness and enhanced life satisfaction among older adults23. Typically, when volunteers receive encouragement and suggestions from members of their social circles, they are more likely to develop an interest in volunteering activities76. A better community environment and well-equipped facilities provide older adults with a platform to build friendships, engage in meaningful interactions, and participate socially, helping prevent disengagement and protect their well-being.

Cultural capital: Interestingly, community satisfaction shows a significant negative association with volunteering participation. This may reflect an altruistic mindset among older adults, who appear more motivated to engage when they perceive a need for better community services. This pattern resonates with Wilson’s competing hypotheses—particularly the notion that neighborhood decline can foster a “crisis consciousness”, thereby stimulating collective action30. In the Chinese context, when community services are perceived as inadequate, older adults tend to participate more in mutual-help volunteering as a means to improve public goods and enhance local service provision. Additionally, frequent engagement in cultural and recreational activities is positively associated with community volunteering. Older adults involved in these activities are more likely to sustain and enhance their spirit of volunteerism31. A positive outlook on aging has a broad impact, significantly influencing older adults’ participation in volunteering in both the full sample analysis and subgroup regressions.

Media capital: Traditional media’s one-way communication fosters social bonds and encourages older adults’ participation in volunteering through public service messaging. As society evolves, the internet expands social connections, facilitates integration, and provides access to valuable information, increasing the likelihood of their involvement in volunteering activities. Hence, the implications of this research suggest that promoting interaction between older adults and virtual social capital can help prevent social isolation and enhance the level of social interaction among older adults77.

Moderating effect: This section reflects, to some extent, the imbalance in individual capital endowments across different groups.

In the urban–rural subgroup analysis, volunteering participation among urban older adults was not significantly associated with subjective economic status, educational attainment, community facilities, or new media use. Instead, participation in this group continued to be linked to non-material factors such as health status, social contact, and positive aging attitudes. In contrast, a wider range of variables remained significantly associated with volunteering participation among rural older adults. This pattern suggests meaningful urban–rural disparities in the resources and mechanisms influencing social participation.

Table 6 Grouped truth table.
Full size table

Striking differences emerge in gender subgroups: female volunteering participation is influenced by education, social contact, and new media use, while male participation is more affected by community facilities and traditional media use. These distinctions may reflect differing motivational orientations—leaning toward relational and communicative engagement for women, and spatially embedded or traditionally informed participation for men. Such patterns are consistent with the influence of socially constructed roles, wherein women may expand weaker ties through emotionally oriented interactions, while men tend to maintain stronger social networks through physical community spaces.

Compared to older-old adults (70 +), young-old adults (60–69) are better able to transcend the influence of their actual health status and external environments (e.g., community facilities completeness). However, the aging process is inevitable: as they grow older, older adults’ participation in community volunteerism and broader social activities will face increasing constraints from both internal and external factors.

The subgroup differences described above indicate significant moderating effects, with these effects rooted in the disadvantaged positions of vulnerable older adults’ groups in terms of resources, information, and other factors. How to ensure that older adults can participate in social life without concerns during the aging process warrants further research.

The discussion of fsQCA

No single condition forms a necessary condition for high community volunteering participation among older adults. Instead, the configurational analysis suggests that even in the absence of certain individual capital factors, compensatory effects from other conditions may still be associated with a high level of volunteering participation. Analyzing the grouping of various capitals, this paper identifies two pathways for older adults to participate in community volunteering: individual ability and social integration. The characteristics of participation in community volunteering should be fully captured for different types of configurations.

The configurational analysis suggests two potential types of older adult volunteers, as shown in Table 7. The first, tentatively labeled as the individual ability type, may primarily leverage their available time, energy, and accumulated social experience to engage in community service. A representative example of this pattern is the "Silver-Haired Volunteers Action" in China, which often involves retired professionals such as teachers, engineers, and doctors. Their activities typically emphasize knowledge-based contributions—such as technical support, cultural preservation, and medical assistance—and appear to be motivated partly by goals of self-realization and social contribution.

Table 7 Practice comparison.
Full size table

The second type, termed the social integration type, seems to participate largely as a means to maintain social connectedness and prevent isolation. Their engagement is often oriented toward expanding social networks in both physical and virtual contexts, with an emphasis on mutual well-being and community building. This type is exemplified by initiatives such as the "Time Bank," which establishes an intergenerational mutual-aid network. It mobilizes a wide social network—specifically the "young-old" and healthier older adults—to provide services directed at older adults in need. This mutual-aid system operates on time credits that are earned for contributions and are redeemable or transferable in the future. It thereby breaks through the limits of traditional volunteerism and enables a precise matching of service supply with demand.

Subgroup fsQCA analyses further echoed the results of subgroup regression analyses, revealing significant conformational differences between groups. The pathways to high volunteering participation among rural older adults involve a more diverse set of conditions, which may reflect the structural differences in urban–rural resource endowments. While urban participants often rely on personal capital such as education, their rural counterparts appear to depend more on community networks and cultural traditions to sustain engagement.

The configurational patterns suggest that high volunteering participation among women tends to rely on a comprehensive combination of capital resources, whereas men appear to achieve similar outcomes through more flexible configurations, sometimes even in the absence of conditions such as robust health. This observed discrepancy may be linked to socially shaped roles and activity types: women’s volunteering is often oriented toward affective services (e.g., companionship, cultural activities), which may require broader resource support, while men’s participation is frequently channeled into instrumental tasks (e.g., community patrols), which might be less dependent on multidimensional capitals.

The analysis indicates that young-old adults (60–69 years) appear to rely predominantly on a high individual capital pathway to achieve volunteering participation. In contrast, the pathways observed among older-old adults are more diverse, with four distinct configurations identified—two of which remain effective even with the absence of certain conditions. This pattern suggests the increasing importance of compensatory mechanisms in later life: as individual capacities decline with age, social support networks (such as family assistance) and environmental adaptations (e.g., improved community facilities) may help to sustain meaningful participation.

Conclusions

Research contributions

From the perspective of the research content, this paper makes three marginal contributions: First, it uses Chinese national data to verify the factors affecting older adults’ participation in community volunteer services, exploring their participation patterns, motivations, and the governing social norms. Second, the paper integrates regression analysis with fsQCA, progressing from an analysis of the net effects of influencing factors to an exploration of their configuration effects, thus overcoming the limitations of previous studies that solely examined the individual effects of volunteer service participation. Third, the study contributes an integrative approach for large-sample factor identification by combining logistic regression and fuzzy-set Qualitative Comparative Analysis (fsQCA). It uses logistic regression to explore the determinants of volunteer behavior among older adults while employing fsQCA to identify configurations leading to high volunteering participation.

From a methodological standpoint, this study makes several contributions to the application and integration of fsQCA in social gerontology research. First, this study confirms the applicability of fsQCA to large-sample research60,78. It also highlights a critical caveat: while larger samples allow for the inclusion of more antecedent conditions, they simultaneously demand stricter analytical standards, particularly in the form of higher consistency and frequency thresholds. As fsQCA is applied to larger samples, it commonly results in solutions with relatively lower coverage79,80,81. This pattern, also noted in earlier large-sample QCA studies, does not invalidate the method but reflects its more deductive character in such settings65. Second, regression and QCA are shown to be complementary rather than mutually exclusive. Regression models estimate net effects and identify variables with strong average predictive power, such as odds ratios of independent factors on volunteering participation. In contrast, fsQCA captures sufficiency-based configurational explanations, revealing critical supplementary conditions—even those not significant in regression—that jointly enable the outcome. Third, compared to traditional causal inference techniques, QCA still has room for development. One promising direction lies in refining variable granularity. A promising direction for future research lies in refining variable granularity. It has been cautioned that oversimplifying complex social phenomena risks obscuring critical causal insights62. Excessively coarse dichotomization may lead to information loss and configurational explosion. Therefore, more nuanced calibration strategies and case selection criteria should be developed to enhance the rigor and interpretability of large-sample QCA.

Policy implications

Enhancing older adults’ human capital necessitates strengthening economic and health support systems. Targeted resource allocation should prioritize rural older adults to alleviate financial barriers to social participation. Moreover, establishing comprehensive community health service systems can mitigate health-related barriers to social engagement. Prioritizing lifelong learning opportunities for older adults enhances their information access and knowledge acquisition capabilities.

Cultivating integrative environments for older adults involves organizing culturally diverse activities to foster intrinsic motivation for community participation. Such initiatives expand social networks and enhance community emotional cohesion. Cultural literacy programs should be multi-channeled to enhance older adults’ spiritual and cultural well-being. Branded community volunteering programs aligned with older adults’ interests and skills should be developed. Expanding online-offline activity integration enriches spiritual-cultural lives, promotes social integration, and enhances mental well-being.

Urban–rural, gender, and age gaps necessitate the establishment of compensatory mechanisms to foster equitable and inclusive environments. First, rural areas should receive policy priority through fiscal subsidies for upgrading community infrastructure (e.g., senior centers, medical stations), developing heritage-focused rural cultural volunteering programs with mutual-aid care components, and leveraging traditional community networks. Second, older women should be empowered to mobilize community cultural resources, strengthen emotional support systems, and assume roles as primary providers of care services (e.g., community care, childcare). Third, intergenerational compensation mechanisms should prioritize mutual assistance between youth and older adults, and among older adult age cohorts. Meaningful volunteering policies should include older-adult-led community classrooms and "Silver-Haired Think Tank" platforms to leverage older adults’ knowledge and skills.

Limitations

However, this study has several limitations that warrant consideration. First, the measurement framework’s reliance on self-reported indicators introduces potential biases, including social desirability bias and other response distortions, which future research could mitigate through refined operationalization with increased measurement granularity. Second, the cross-sectional design partly constrains the ability to establish causal relationships between variables, necessitating longitudinal investigations to verify temporal dynamics. Third, while the current analytical focus centers on micro-level individual characteristics and meso-level social connectivity, the macro-level policy implications remain underexplored, suggesting the need for multilevel analyses incorporating institutional and regulatory dimensions in subsequent studies.

Prospects

The present cross-sectional study supports the viability of integrating econometric regression models and fsQCA as complementary analytical techniques in large-sample research. To advance scholarship on aging governance and social inclusion/exclusion of older adults, we propose two key directions for future research:

Longitudinally, the dynamic fsQCA technique offers a methodological foundation for investigating evolving mechanisms82. We recommend employing balanced panel data from two or more waves to trace the evolutionary pathways of social participation patterns among older adults and examine temporal changes in causal configurations.

From a cross-sectional perspective, despite the use of national data in this study, significant cross-national variations persist due to differing socio-cultural contexts. It is noteworthy that the participation rate in community volunteering among Chinese older adults rose significantly from 18% in the baseline survey (2014) to 25% in 2020. Despite this rapid growth, the overall participation level remains substantially lower than that in the United States and Japan (both exceeding 40%) and Nordic countries (approximately 30%)83,84,85. Therefore, promoting international comparative research on volunteering is crucial, as it not only enhances the external validity of findings from single-country studies but also elucidates how factors such as aging intensity, religious culture, scale of social organizations, and welfare regimes shape patterns of volunteering participation among older adults. Such comparative work will ultimately contribute to developing more nuanced policy insights for promoting active aging globally.