Do Mixed Religions Make Families More Generous? An Empirical Analysis Based on a Large-Scale Survey of Chinese Families

Abstract

This study focuses on the relationship between mixed religions and family donations in China as the object of analysis, where individual donations are primarily in the form of family contributions. Although there is considerable research on the relationship between religion and individual donations, the academic community has yet to clarify the connection between mixed religions and family donations. Based on the large-scale survey data from the 2020 China Family Panel Studies, this study employs econometric models such as probit and tobit models to examine the relationship. To mitigate endogeneity and enhance the robustness of the conclusions, this research also conducts instrumental variable analysis and robustness analysis. The study finds that the influence of mixed religions on family donations is greater than that of non-mixed religions, but this influence is more pronounced in families with higher donation levels. Heterogeneity analysis reveals that mixed religions have a greater impact on donations in families headed by individuals aged 41–59, females, those with no religious affiliation, residents of urban areas, and families in the western and northeastern regions. Furthermore, the impact of mixed religions on family donations is higher only when there is a combination of one or three religions. Additional analysis indicates that households with lower educational attainment, non-Party members, individuals suffering from illnesses, and unmarried females are more likely to choose mixed religions.

1. Introduction

Charitable donations, as E.P. Thompson puts it (charity is within the context as this that the function of riot may be disclosed) play a crucial role in narrowing the wealth gap and upholding social fairness and justice (Li and Yan 2023; E. P. Thompson 2015). They are also an important factor reflecting the level of modern civilization. In exploring the role of charitable donations and the mechanisms by which it occurs, Miao (2022) and Yan et al. (2017) indicate that charitable donations have unparalleled advantages in caring for, supplying, and rescuing low-income populations, effectively resolving social conflicts, and promoting the rational flow of wealth. Due to the pivotal role of charitable donations, academia is keen on exploring the driving factors behind them. One perspective suggests that, as an economic behavior, charitable donations are influenced by income and wealth. Another viewpoint argues that, as an altruistic behavior, social capital plays a critical role in the decision-making process of charitable donations (Forbes and Zampelli 2013). However, economic behavior and altruistic behavior are conflicting action systems, with economic behavior placing more value on rational gains and altruistic behavior placing more value on emotional gains. Exploring the motives behind charitable donations from the perspectives of income and wealth, as well as social capital, can effectively supplement related research but may struggle to unveil the complete picture of charitable donations. Religious sociologists provide a unique perspective for the study of charitable donations, considering religion as a cultural factor and an important driver of charitable giving behavior. In recent years, there has been extensive discussion in academia regarding the relationship between religion and charitable donations, largely confirming the close connection between the two. Building on this foundation, scholars have also explored the doctrines, values, types, and functions of religion, attempting to clarify the mechanisms by which religion influences charitable donations (Yasin et al. 2020). However, despite the confirmed affinity between religion and charitable donations in many countries, there is relatively little discussion in the Chinese context. Additionally, current research overlooks the reality of “religious mixing” and fails to comprehensively depict the association between religion and charitable donations. In this context, this study aims to focus on charitable donations in the Chinese context, exploring the complex relationship between mixed religious beliefs and charitable donations. It seeks to enrich and supplement existing research on the relationship between religion and charitable donations, providing decision-making references for relevant government departments in formulating policies to incentivize charitable donations.

In 2016, China enacted its first law, called “Charity Law of the People’s Republic of China”, to regulate the development of philanthropy. China’s philanthropic landscape is poised for a pivotal developmental phase. However, a nuanced analysis reveals persisting challenges hindering the advancement of charitable contributions in the country. Foremost among these challenges is the relatively modest scale of philanthropic giving in China. According to the China Charity Development Report (2022), the aggregate sum of charitable donations in China, after witnessing an increase to CNY 152.6 billion in 2017 from 2009, has subsequently plateaued at approximately CNY 150 billion. Notably, the growth trajectory of China’s charitable donations conspicuously lags behind the corresponding expansion of the GDP. In terms of its contribution to the GDP, philanthropic giving in China constitutes a mere 0.15% (Deng 2021). In a global context, China’s philanthropic contribution ranks 49th out of 119 nations, trailing behind counterparts such as Indonesia, Kenya, the United States, Australia, New Zealand, and Canada (Charity Aid Foundation 2022). Concomitantly, the structural composition of philanthropic donations in China presents inherent deficiencies. There are also major differences between China and the West in terms of the composition of charitable donations. Take the United States as an example. In China, over an extended period, corporate donations have predominated the philanthropic landscape, with individual contributions maintaining a comparatively diminutive share in the overall philanthropic structure. From 2012 to 2020, the percentage of individual donations relative to the total philanthropic contributions has not exceeded 30%. In stark contrast, during the same temporal span, individual contributions in the United States consistently hovered at approximately 70%. This suggests that there are different foundations for philanthropic development in China and the United States. While acknowledging the pivotal role of corporate philanthropy in China, an excessive reliance on such entities runs the risk of obfuscating the demarcation between charitable organizations and profit-driven sectors, potentially diverting philanthropic entities from their predefined objectives and jeopardizing their legitimacy (Li 2020). Consequently, to fully galvanize the dynamism of philanthropic contributions in China, a strategic emphasis on augmenting individual donation levels is imperative.

Indeed, scholars have underscored that the decision-making process concerning individual charitable donations often transpires within the familial context (Burgoyne et al. 2005), representing the culmination of collective decisions within the household (Zhu and Liu 2017). Predominantly, the prevalent model for individual contributions involves joint decision making by spouses. Concurrently, the philanthropic ethos within a family and the effects of intergenerational transmission plays a pivotal role in propelling family-based philanthropy (Yang and Zhang 2017). In the context of Chinese society, the practice of donating as a family unit is more ubiquitous. This proclivity is profoundly shaped by the influence of Confucian cultural norms, wherein the family unit stands as the paramount foundational entity in Chinese societal structure (Fei 1982). Characterized by frequent interpersonal interactions and close familial bonds, this dynamic has engendered a paradigm of individual donations in China primarily centered around familial philanthropy (Fukuyama 2016; Yang et al. 2019). Consequently, this research places particular emphasis on scrutinizing the relationship between eclectically intertwined religious beliefs and familial philanthropic contributions. Given that family donations emanate from both individual contributions and the collective decisions of family members, and considering the relative scarcity of extant research on familial giving, this study initiates an exploration into the intricate dynamics of the relationship between religion and individual donations.

Durkheim’s theory of social integration posits that religion serves as a source of collective norms, including collectivism, altruism, and principled conduct. These norms are reinforced through communal actions (Regnerus et al. 1998; Reitsma et al. 2006). Individuals, when affiliating with a religious belief system and actively engaging with its doctrines, undergo a process wherein they develop empathetic value systems shaped by religious tenets (Schnable 2015). Subsequently, this engenders acts of generosity. Max Weber, a seminal figure in the sociology of religion, in his seminal work “The Protestant Ethic and the Spirit of Capitalism”, delves into the profound influence of Protestant doctrines on individual work attitudes and behavior. Weber posits that the purpose of diligent work, according to Protestant teachings, is to attain redemption from God. Individuals, under the sway of Protestant doctrines, seek redemption through charitable donations. Nevertheless, substantial heterogeneity exists among different religions, such as Christianity and Buddhism, with notable disparities in their respective emphasis on redemption. Individuals adhering to religions placing a higher premium on redemption demonstrate a greater propensity to donate, often contributing more substantially. Research also reveals disparities, with Protestants in countries like the United States, Canada, and the Netherlands exhibiting greater generosity compared to their Catholic counterparts (Bekkers and Schuyt 2008).

The religious values theory contends that diverse religions embody distinct religious values, positing that individuals affiliated with religions emphasizing higher religious values are predisposed to engage in charitable donations (Thornton and Helms 2013). The religious commitment theory proposes that the magnitude of individual donations correlates with the level of religious devotion. Evidently, higher religious devotion is associated with more substantial contributions to charitable donations. However, religious devotion is intricately linked to religious practices, with greater engagement in activities such as worship, rituals, and incense-burning indicative of heightened religious devotion (Vaidyanathan et al. 2011). Given the potential separation between identity and practice in religious development, certain studies suggest that individual donation levels are more closely aligned with religious practices (Eagle et al. 2017). Scholars assert that individual donations can be dichotomized into religious and secular categories, with religious influence predominantly manifesting in religious donations (Hu 2013). However, some argue that contributions from religious individuals to fellow believers or organizations are often prompted by a sense of duty, with the anticipation of receiving religious rewards (Boechat et al. 2018). Nevertheless, individual charitable donations progressively expand from religious to secular domains, as secular contributions are anticipated actions by religious organizations (Cheung and Kuah 2019). However, research has indicated that the correlation between religious and secular donations may be deceptive, as corroborated by two studies from Australia (Lyons and Passey 2005; Lyons and Smith 2006). In a study on the impact of the scandal of abuse by Catholic clergy in the United States on religious participation, religious beliefs, and pro-social behaviors, a robust cross-sectional correlation between religious participation and charitable donations was found, suggesting an assumed causal relationship (Bottan and Perez-Truglia 2015). Furthermore, certain studies posit a negative association between religion and secular donations, suggesting that individuals with minimal involvement in religious activities exhibit greater philanthropic generosity (Brooks 2005). It is pertinent to underscore that in the contemporary process of modernization, the quest for God’s redemption or afterlife rewards no longer exclusively dictates individual religious belief choices. Rather, motivations encompass a pursuit of earthly rewards, psychological solace, and mental reassurance (Junior and Py 2024). Motivational differentials in religious choices give rise to variations in donation motivations, objectives, and modalities. In the context of secular donations, advanced age, moderate liberal ideologies, and elevated social status are associated with relatively heightened levels of generosity (Çokgezen and Hussen 2021).

Evidently, extant scholarly inquiries into the intersection of religion and individual philanthropy have yielded a rich reservoir of findings, thereby furnishing a robust foundation for an examination of the nuanced dynamics between religion and familial charitable contributions. Nevertheless, the analytical lens ought to be attuned to the discernible impact of “mixed” religious orientations on household donations. Prevailing scholarly consensus perceives Western religions as discrete entities, juxtaposed with the amalgamated nature of Chinese religious practices. In Western religious paradigms, distinct religions manifest heightened exclusivity, whereas their Chinese counterparts reflect a marked proclivity toward inclusivity. Consequently, prevailing investigations into the interplay of religion and familial donations often hinge upon dichotomizing analyses based on the foundational attributes of Western and Chinese religious frameworks. However, recent scholarly revelations indicate a widespread prevalence of “mixed” religious affiliations both in Western and Chinese contexts (Li and Wang 2023). A case in point is Japan, where statistics from 1983 unveil a demographic incongruity wherein approximately 120 million residents coexist with a staggering 220 million believers. Moreover, 72% of the Japanese populace perceives a substantial convergence in the overarching objectives and tenets of various religions, with a considerable segment displaying ambiguity regarding their specific religious affiliations (Laster 2010). Analogously, within Western societies, the phenomenon of syncretic beliefs is pervasive, with individuals often identifying concurrently as adherents of Christianity alongside practitioners of Hinduism, Buddhism, or Judaism (Cornille 2012). China mirrors this trend, where individuals seamlessly intertwine affiliations with Buddhist, Taoist, folkloric, Christian, and Catholic traditions. It is evident that an oversimplified taxonomy of family donations predicated on either singular or amalgamated religious affiliations may fall short in encapsulating the nuanced influence of religion on familial philanthropy, particularly within the Chinese milieu.

Against this backdrop, the methodological orientation of the present study is delineated as follows: Firstly, conduct a reevaluation of the correlation between professed religious beliefs, or lack thereof, and familial philanthropic dispositions within the Chinese context; secondly, in tandem with an exposé of the contemporary landscape of “mixed” religious affiliations in China, elucidate the intricate repercussions of such affiliations on familial philanthropic endeavors; thirdly, employ advanced instrumental variable methodologies to meticulously investigate the causative linkages between “mixed” religious orientations and individual philanthropic propensities, accompanied by rigorous tests for the robustness of findings; fourthly, scrutinize the heterogeneity inherent in the impact of “mixed” religious affiliations on familial philanthropy; finally, undertake an in-depth exploration to unveil the specific contextual contours through which “mixed” religious affiliations exert influence upon familial philanthropic proclivities.

2. Data and Methods

2.1. Data Source and Variables

The dataset under examination in this research originates from the 2020 China Family Panel Studies (CFPS), meticulously administered by the China Social Science Survey Center at Peking University. The CFPS, structured for longitudinal profiling across individual, familial, and community strata, systematically captures and archives the dynamics and attributes of China’s societal, economic, demographic, educational, and health landscapes. The CFPS 2020 initiative encompassed exhaustive visits to approximately 11,000 households, culminating in the collection of 51,000 individual questionnaires. Religious data integral to this study emanates from the “General Restricted Data” enclave within the CFPS 2020 repository. Subsequent to the requisite data acquisition procedures, a judicious curation process was undertaken, entailing the elimination of missing values, aberrant data points, and instances of survey discontinuity. Consequently, a discerningly pruned cohort of 7386 household samples has been assimilated into the analytical purview of this study.

The primary inquiry of this study revolves around the impact of mixed religious beliefs on household donation behavior. Through empirical assessment, donation behavior is broadly categorized along two dimensions: the decision to donate or not, and the quantum of the donation. Unlike Western countries, influenced by the “hierarchical pattern” and a “network-based society”, Chinese households typically rely on kinship, geographical, and occupational networks. They exhibit a preference for donating to beneficiaries known to them, as opposed to anonymous or symbolic organizations (Hu and Shen 2013). Therefore, the focal point of this study is distinct from Western philanthropic donations to charitable or societal causes. Drawing inspiration from the work of Deng and Rong (2022), this research explores how mixed religious beliefs influence respondents’ willingness to provide financial assistance to familiar acquaintances. Specifically, the variable of interest is constructed by transforming the CFPS 2020 questionnaire item “providing economic assistance to others (in RMB per year)” into two key dependent variables for this study: (1) Willingness to donate. If a respondent reported providing zero financial assistance to others in the year 2019, it indicates a lack of engagement in household donations, and the variable is coded as 0. Conversely, if financial assistance was reported, indicating participation in household donations, the variable is coded as 1. (2) Donation amount. This variable is measured by the actual amount of financial assistance reported by the respondent in the year 2019.

The pivotal explanatory variable in this study is the household’s mixed religious beliefs. Typically, household heads serve as significant decision-makers (Deng and Rong 2022). Drawing inspiration from the methodologies articulated by Zhou and Fan (2018), the religious beliefs of the household head are leveraged as a proxy variable for the overall household’s religious orientation. Charitable giving is a household-level characteristic, and in single-headed households, the family’s donations are directly attributed to the actions of the household head (Rooney et al. 2005). For the purposes of this study, the financial respondent in the household economic questionnaire is considered the household head. The CFPS 2020 survey employs a comprehensive approach encompassing both deity-centric and sectarian measurements, probing dimensions such as religious affiliation, participation in religious practices, folk beliefs, and deity worship. The deity-centric measurement includes inquiries regarding whether respondents believe in “Buddha or Bodhisattva”, “deities”, “Allah”, “God”, or “Jesus Christ”. The sectarian measurement encompasses queries about whether respondents adhere to “Buddhism”, “Taoism”, “Islam”, “Catholicism”, or “Christianity”. Folk belief measurement involves questioning respondents about their beliefs in “ancestors”, “spirits”, “feng shui”, and “destiny”. Religious practices include assessing the frequency of activities such as “burning incense and praying” and “worship”. Based on the aforementioned items, the following variable designs are implemented: (1) Construction of variables distinguishing Western religious beliefs from Chinese religious beliefs. Any affirmative response in the Western deity, sectarian, or practice options is considered belief or participation in Western religion, while any affirmative response in the Chinese deity, sectarian, or practice options is considered belief or participation in Chinese religion. (2) Construction of the religious belief variable. A value of 1 indicates the presence of religious beliefs (either Western or Chinese), while a value of 0 indicates the absence of religious beliefs (neither Western nor Chinese). (3) Construction of the core explanatory variable for this study: Mixed religious beliefs. A value of 1 signifies that the household head adheres to mixed religious beliefs (believing in both Western and Chinese religions), while a value of 0 indicates that the household head adheres to a non-mixed religious belief (solely believing in Western religion).

In aligning with established research paradigms, this study diligently incorporates an extensive array of control variables at the individual, household, and community strata, aiming to meticulously account for potential influences on household philanthropy. At the individual level, these variables encompass key demographics such as household head age, gender, educational attainment, marital status, physical health status, political affiliation, employment status, as well as subjective evaluations of economic standing and confidence levels (Kamas et al. 2008; Bekkers and Wiepking 2011). Household-level factors include considerations of family size, household income, and residential property valuation. The community-level variables involve distinctions between urban and rural settings, as well as regional categorizations. This judicious integration of control variables serves the dual purpose of discerning and mitigating potential confounding factors, thereby fortifying the study’s methodological framework for a nuanced exploration of the nexus between mixed religious beliefs and household donation behaviors. The inclusion of individual, household, and community-level determinants augments the study’s analytical robustness, fostering a more comprehensive elucidation of the intricate dynamics inherent in household philanthropic activities. The descriptive statistics of each variable are shown in

Table 1. Descriptive statistics of variables.

Variables	Mean	Std. Dev.	Min	Max
Donation willingness	0.15	0.36	0	1
Logarithm of donation amount	3.00	1.74	2.30	12.21
Mixed religious	0.10	0.30	0	1
Age	48.49	14.89	16	95
Gender	0.58	0.49	0	1
Education	2.08	1.29	0	4
Marriage	0.89	0.31	0	1
Health	2.97	1.17	1	5
Political identity	0.13	0.33	0	1
Employment	0.78	0.42	0	1
Subjective confidence	4.13	0.93	1	5
Subjective economic status	3.07	1.05	1	5
Household size	3.55	1.93	1	15
Logarithm of household income	9.91	1.09	2.30	14.51
Logarithm of housing value	3.04	1.36	0.69	8.01
Urban/rural area	0.54	0.50	0	1
Region	2.78	1.15	1	4

.

2.2. Model Specification

2.2.1. Probit Model for the Influence of Mixed Religious Beliefs on Household Donation Intention

Household donation intention is a binary discrete variable, which does not conform to the statistically normal distribution and cannot be estimated using the least squares method. The factors influencing household donation decisions represent a type of unordered choice problem, i.e., whether to choose to donate. When a model is established with such decision results as the dependent variable, it is referred to as a binary choice model. Therefore, following the approach of Yan et al. (2017), and Taniguchi and Marshall (2014), this study employs the Probit model to examine the relationship between mixed religious beliefs and household donation intention. The model is specified as follows:

$$\begin{array}{c}{Donation}_i^{\mathrm{*}}=\alpha +\beta \left({Hybridreligion}_i\right)=\gamma X_i+{\epsilon }_i\\ {Donation}_i=1\left({Donation}_i^{\mathrm{*}}\right)\end{array}$$

(1)

Here, i represents the surveyed households, ${Donation}_i$ represents whether the surveyed households made a donation, with a value of 1 if yes and 0 if no. ${Donation}_i^{\mathrm{*}}$ is the latent variable. When ${Donation}_i^{\mathrm{*}}$ is greater than 0, it takes the value of 1; otherwise, it is 0. $X_i$ represents a series of control variables.

2.2.2. Tobit Model for the Impact of Mixed Religious Belief on Household Donation Amount

Household donations are fundamentally characterized by two situations: one where the household exhibits no donation expenditure, indicating a donation amount of zero; and the other where the household engages in donation activities, reflecting a donation amount greater than zero. It is evident that household donations represent a classic case of censored data. The use of ordinary least squares estimation in such cases may introduce nonlinear disturbances into the error term, potentially yielding inconsistent estimates (Maddala 1985). To circumvent this issue, the Tobit model is selected, drawing inspiration from the work of Brown et al. (2012). This study employs the Tobit model to rigorously examine the association between mixed religious beliefs and the magnitude of household donations. The specific formulation of the model is delineated below:

$$Ln\left({Donation}_i\right)=\alpha +\beta \left({hybridreligion}_i\right)+\gamma X_i+{\epsilon }_i$$

(2)

Here, $Ln\left({Donation}_i\right)$ represents the logarithm of household donation expenditure i1, and the remaining variables are consistent with those in Equation (1). The focal point of this study lies in estimating the marginal effects of mixed religious belief on household donation expenditure. This effect denotes the average marginal impact of mixed religious beliefs on household donations.

3. The Results of Empirical Analysis

3.1. Descriptive Statistics of Key Variables

The mean values of donation willingness and donation amount in the sample are 0.143 and 427, respectively. This is consistent with the findings of Li and Li (2023), who reported mean values of donation willingness (0.20) and donation amount (CNY 570) using CFPS 2018 data. Overall, Chinese households exhibit relatively low willingness and amounts for charitable donations.

Table 2. Overview of variables related to religious beliefs.

Religious Beliefs		Mixed Religious Beliefs
Religious	Not-religious	Mixed religious	Not-mixed religious
			Only Western religion		Only folk religion
5977	1409	595	12		5370
The types of mixed religious beliefs
One type	Two types	Three types	Four types	Five types		Six types
137	215	145	70	1		9
The number of mixed different religions
Chinese religious beliefs		Chinese spiritual beliefs	Chinese folk religions	Incense burning and Buddha worship
0		3	99	493

presents the descriptive statistics of the main explanatory variables related to religious beliefs. Among the total 7386 households, 5977 households (80.92%) profess some form of religious belief. Among these religious households, only 12 follow Western religious beliefs, while a substantial 5370 households adhere exclusively to Chinese religious beliefs. This suggests that current religious beliefs in China are predominantly centered around Chinese religious practices, and even exclusive Western religious beliefs are being transformed by Chinese indigenous culture, illustrating the prevalence of mixed religious beliefs. This aligns with the conclusions drawn by scholars like Li and Wang (2023) regarding the existence of mixed religious practices in China.

Specifically, there are 595 households with mixed religious beliefs, accounting for 9.95% of all religious households. How many households have mixed beliefs in reality? If we use the calculation method based on the document “Policies and Practices of Ensuring Freedom of Religious Belief in China” published by the State Council Information Office of China in 2018, there are approximately 200 million religious citizens in China. According to the data from the seventh national population census of China, the average household size in China is 2.62 persons. Based on this, it can be estimated that there are approximately 66 million religious households in China, with approximately 6.567 million households having mixed religious beliefs. This indicates that mixed religious beliefs in China cannot be ignored. Studying the effect of mixed religious beliefs on donations is meaningful for enriching existing research on religion and donations. Further analysis reveals that the most common type of mixed religious belief involves a combination of two beliefs (215 households), and cases with one to four mixed beliefs are much more prevalent than those with five or more. Among the mixed types, there is no combination of Western religious beliefs with Chinese beliefs, which may be related to the doctrines of the religions. Only three households have a mixedness of Chinese deity beliefs, and the most common mixed religious practice involves burning incense and worshiping Buddha. There are relatively more cases of mixing with Chinese folk beliefs.

3.2. Baseline Regression

Before discussing the relationship between mixed religious beliefs and family donations, this study initially employed Probit and Tobit models to examine the causal relationship between religious beliefs and family donations. Endogeneity and robustness checks were conducted on the results. After controlling for a series of variables, families with religious beliefs showed a higher willingness and greater donation amounts compared to those without religious beliefs. Specifically, political affiliation, employment status, family income, and expenses significantly influenced both willingness and donation amounts. This aligns with existing research findings (Lyons and Smith 2006; Bekkers and Schuyt 2008), confirming the significant positive impact of religious beliefs on family donations.

Table 3. The results of baseline regression.

Dependent Variables	Donation Willingness		Donation Amount
	(1) Probit	(2) LPM	(3) Tobit	(4) OLS
Mixed religious beliefs	0.053 **	0.056 **	0.250 **	0.251 **
	(3.29)	(3.09)	(3.02)	(3.01)
Controlled variables	Controlled	Controlled	Controlled	Controlled
Samples	4740	4740	4740	4740
Pseudo R2	0.059	0.047	0.015	0.057

Note: The table reports marginal effects with t-values in parentheses. Significance levels are denoted by **, indicating significance at the 5%levels, respectively. The marginal estimates are presented for Probit and Tobit models. Numeric variables such as donation amount, household income, and household expenses have undergone logarithmic transformation in all subsequent tables.

reports the baseline results of the impact of mixed religious beliefs on donation willingness and donation amounts. In the Probit regression results in column (1), the marginal effect of mixed religious beliefs is 0.053 and statistically significant at the 5% level. This indicates that mixed religious beliefs significantly increase family donation willingness, with a 5.3 percentage point higher willingness in families with mixed religious beliefs compared to those without. The Tobit regression results in column (3) show that the marginal effect of mixed religious beliefs is 0.250 and statistically significant at the 5% level. This suggests that mixed religious beliefs significantly increase family donation amounts, approximately by 25%. Robustness checks using linear models for donation willingness and amounts showed minimal differences compared to the baseline regression results, affirming the robustness of the conclusions after these adjustments.

To examine whether the baseline regression results of this study are not due to random factors, we followed the approach of Tang et al. (2020) and employed a permutation test as a placebo test. The specific approach involved randomly assigning households in the sample to mixed religious beliefs using Monte Carlo simulation for 1000 iterations to obtain regression coefficients. A test statistic was then constructed to assess the probability, or p-value, of obtaining coefficients in the simulated distribution equal to zero, testing the hypothesis that mixed religious beliefs have no significant impact on donation willingness and donation amount, as observed in the baseline regression. The results of the permutation test consistently rejected the null hypothesis of no significant impact at a 5% confidence level, indicating that the baseline regression results of this study are not driven by other random factors.

To assess the robustness of the baseline regression results in this study against potential spurious influences, we adopted a method inspired by Tang et al. (2020) involving a permutation test as a placebo examination. Specifically, we employed a Monte Carlo simulation with 1000 iterations, randomly assigning households in the sample to embrace mixed religious beliefs. From this simulated dataset, regression coefficients were derived to construct a test statistic. This statistic was then utilized to evaluate the probability (p-value) of observing coefficients in the simulated distribution equivalent to zero. This testing framework aimed to scrutinize the hypothesis asserting that the observed baseline regression coefficients, indicating no significant impact of mixed religious beliefs on donation willingness and amount, could occur randomly. The permutation test consistently rejected the null hypothesis at a 5% confidence level2, affirming that the baseline regression findings in this study are not contingent on fortuitous elements.

3.3. Endogeneity Test

Empirical research primarily faces potential endogeneity issues arising from measurement errors, omitted variables, and reverse causality. In the CFPS 2020 survey, a comprehensive approach was adopted, employing both divine measurement and sectarian measurement methods to extensively investigate dimensions such as residents’ religious affiliation, participation in religious practices, Chinese folk beliefs, and divine beliefs. This approach reduces errors resulting from single-dimensional measurement of religious affiliation, thereby mitigating endogeneity issues attributed to measurement errors. Consequently, this study particularly focuses on the endogenous impacts brought about by omitted variables and reverse causality.

On the one hand, there may be a reverse causality between religious beliefs and the likelihood of households engaging in donations (Li and Yan 2023). On the other hand, there could be an issue of omitted variables. For instance, when families face unresolved challenges, they might engage in donations as a means of seeking psychological comfort, with religion being an important avenue for such solace. Therefore, it is crucial to select appropriate instrumental variables to address this endogeneity issue. According to the peer effects theory, peer behavior is a significant determinant of individual behavior (Sampson and Perry 2019). Existing theories also suggest that individual religious activities can be influenced by peers. For example, Pan and Zhong (2016) argue that theoretically, the frequency of an individual’s participation in religious activities can be influenced by the surrounding community, indicating a close correlation between personal religious beliefs and the local atmosphere of religious activities and traditions. Thus, this study proposes that selecting the “proportion of households with mixed religious beliefs in the same village/neighborhood to the total number of households in the village/neighborhood” as an instrumental variable is relatively appropriate.

In terms of the assumption of correlation, Yin and Zhang (2019), in their study on the relationship between religious beliefs and household financial market participation, pointed out that in areas with a stronger religious atmosphere, the likelihood of households having religious beliefs is higher. In places where the “proportion of households with mixed religious beliefs in the same village/neighborhood” is higher, the religious atmosphere in that area is likely to be more pronounced, and the possibility of households having mixed religious beliefs is greater. It can be observed that there is a strong correlation between households with mixed religious beliefs and the proportion of households with mixed religious beliefs in their respective villages/neighborhoods. Moreover, substituting village/neighborhood-level variables for individual-level variables can address endogeneity issues caused by specific individual characteristics, thus meeting the necessary requirement of instrumental variables having a strong correlation with the endogenous variable (Li and Yan 2023). Furthermore, village-level data also exhibit good exogeneity. The mixed religious beliefs at the village level often only influence the religious beliefs of households, and when it comes to significant matters related to household expenditures (such as charitable donations), village-level data are often rationalized to have minimal impact on household donations.

Table 4. The results of re-estimated using the instrumental variables.

The Results of Two-Stage Regression
Dependent Variables	Donation Willingness		Donation Amount
	(1) Iv Probit	(2) 2 SLS	(3) Iv Tobit	(4) 2 SLS
Mixed religious beliefs	0.333 **	0.083 *	0.439 **	0.439 *
	(2.61)	(2.22)	(2.88)	(2.29)
Controlled variables	Controlled	Controlled	Controlled	Controlled
Samples	4752	4752	4752	4752
The result of first-stage
Instrument variables	1.048 ***	1.048 ***	1.048 ***	1.048 ***
	(44.98)	(46.49)	(44.98)	(46.49)
F		2161.49		2161.49
KPF-statistic		2023.23		2023.23

Note: The table reports marginal effects with t-values in parentheses. Significance levels are denoted by ***, **, and *, indicating significance at the 1%, 5%, and 10% levels, respectively. The marginal estimates are presented for Probit and Tobit models. Numeric variables such as donation amount, household income, and household expenses have undergone logarithmic transformation in all subsequent tables.

presents the results after estimating using instrumental variables. From the perspective of the first stage, there is a significant positive relationship between the instrumental variable and mixed religious beliefs. Meanwhile, the first-stage KPF-statistic for 2 SLS is 2023.23, a value substantially exceeding the critical threshold (16.38) for rejecting the weak instrument test at the 10% significance level. The F-statistic is also well above the commonly used threshold of 10. Therefore, we assert that there is no weak instrument problem with the chosen instrumental variable. As this study considers only one instrumental variable, there is no need for overidentification tests to examine the validity of the instrument. According to theoretical assumptions, the “proportion of households with mixed religious beliefs in the same village/neighborhood to the total number of households in the village/neighborhood” is unlikely to be associated with household donations, indicating that the instrumental variable satisfies the exogeneity assumption.

In summary, the regression results obtained through empirical analysis in this study indicate the stability of the relationship between mixed religious beliefs and the willingness to donate as well as the donation amount. Specifically, mixed religious beliefs exhibit a consistently significant positive influence on both the willingness to donate and the amount of donations made by households.

3.4. Robustness Test

Robustness test 1: Outlier removal. In this study, a truncation process was applied to continuous variables at the 10% level, and the results remained significant. See

Table 5. The results of the robustness tests.

Dependent Variables	Robustness Test 1		Robustness Test 2
	Donation Willingness	Donation Amount	Donation Willingness		Donation Amount
Mixed religions	0.060 ***	0.231 ***	(1) nearest neighbor matching	(2) kernel matching	(1) nearest neighbor matching	(2) kernel matching
	(3.66)	(3.58)	0.062 **	0.055 ***	0.314 ***	0.248 ***
			(2.49)	(2.88)	(2.67)	(2.72)
Controlled variables	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
N	4752	4752	4752	4752	4752	4752

Note: The table reports marginal effects with t-values in parentheses. Significance levels are denoted by ***and **, indicating significance at the 1% and 5% levels, respectively. The marginal estimates are presented for Probit and Tobit models. Numeric variables such as donation amount, household income, and household expenses have undergone logarithmic transformation in all subsequent tables.

for the results.

Robustness test 2: Model replacement. The mixing of religious beliefs within families is not the result of random assignment, and there may be a self-selection issue involved. To address this sample selection issue, this study employed the propensity score matching method. Firstly, a Logit model was used to estimate the probability of residents mixing religious beliefs and calculate the propensity scores. Then, based on these scores, the treatment and control groups were matched, and the average differences between the matched groups were estimated. To ensure the reliability of the propensity score matching, this study conducted a balance test, which showed no significant differences (p > 0.5) and standardized biases of less than 10% between the treatment and control groups, indicating a good matching effect. Furthermore, the kernel matching method was found to be more effective in utilizing control group information, and in the case of a large sample, the results were closer to those of the nearest neighbor matching method. To ensure the robustness of the results, both nearest neighbor matching and kernel matching analyses were performed, and the results are presented in Table 5.

3.5. Quantile Regression

The above process primarily examined the “average” impact of mixed religious beliefs on household donations. In order to provide a more comprehensive depiction of the relationship between mixed religious beliefs and household donations, this study utilizes quantile regression to assess the impact of mixed religious beliefs on donations at different quantiles. Figure 1 presents the coefficient plot of quantile regression for the influence of mixed religious beliefs on household donations. The quantile intervals set in this study range from 0.1 to 0.9. Figure 1 indicates that the impact of mixed religious beliefs on household donations is nonlinear. When household donations are below the 0.8 quantile, the effect of mixed religious beliefs on household donations is zero. In the quantile range of 0.8 to 0.85, there is a significant positive impact of mixed religious beliefs on household donations, and this impact increases with the rising quantile point within this range. However, starting from the 0.85 quantile, the impact of mixed religious beliefs on household donations decreases as the quantile point increases. Until the 0.95 quantile, the impact of mixed religious beliefs on household donations returns to zero. This indicates that within the quantile range of 0.8 to 0.95 for household donations, mixed religious beliefs have a significant positive impact, but this effect first increases and then decreases.

4. Further Analysis

4.1. The Results of Heterogeneity Analysis

4.1.1. Age Heterogeneity

In this study, the sample was divided into three groups based on the age of the household head: below 40 years, 41–59 years, and 60 years and above, as shown in

Table 6. The results of heterogeneity analysis.

Variables	Dimensions	Samples	Donation Willingness	Donation Amount
Age	Below 40	1372	0.031 (0.88)	0.251 (1.37)
	41–59	2089	0.081 ** (3.30)	0.379 * (3.00)
	Above 60	1291	0.041 (1.58)	0.121 (0.92)
Gender	Male	2721	0.052 * (2.25)	0.226 (1.87)
	Female	2031	0.063 ** (2.73)	0.311 ** (2.75)
Religious organization membership	Yes	93	0.007 (0.05)	0.249 (0.43)
	No	4650	0.053 ** (2.94)	0.249 ** (2.73)
Urban/Rural	Urban	2548	0.065 * (2.56)	0.352 ** (2.70)
	Rural	4650	0.053 ** (2.94)	0.249 ** (2.73)
Regions	Northeast	704	0.112 ** (2.96)	0.378 (1.89)
	Central	1374	0.019 (0.61)	0.131 (0.85)
	Western	780	0.099 ** (2.63)	0.623 ** (3.27)
	Eastern	1895	0.046 (1.68)	0.202 (1.44)

Note: The table reports marginal effects with t-values in parentheses. Significance levels are denoted by ** and *, indicating significance at the 5% and 10% levels, respectively. The marginal estimates are presented for Probit and Tobit models. Numeric variables such as donation amount, household income, and household expenses have undergone logarithmic transformation in all subsequent tables.

. Only the group aged 41–59 exhibits a significantly positive impact of mixed religious beliefs on the willingness and amount of household donations. This may be attributed to this age group facing relatively less economic pressure compared to those below 40 and possessing more donation capacity compared to those aged 60 and above, making the positive influence of mixed religious beliefs more pronounced.

4.1.2. Gender Heterogeneity

Mixed religious beliefs have a significantly positive impact on the willingness to donate for both males and females. For females, there is also a significant positive impact on the amount of household donations, whereas this relationship is not significant for males. This might be due to the fact that the willingness to donate is merely a “possibility”, and therefore, both male and female mixers of religious beliefs significantly exhibit a positive inclination. On the other hand, the donation amount is an “actual” measure, and males, being more rational, do not show a significant impact of mixed religious beliefs on the amount of household donations.

4.1.3. Religious Organization Membership Heterogeneity

The sample was divided into two groups based on whether they were members of a religious organization: religious organization members and non-religious organization members, as shown in Table 6. Considering the exclusivity of Western religions, whether someone is a member of a religious organization becomes an interesting topic. From the results, only non-religious organization members with mixed religious beliefs show a significantly positive impact on household donations (both willingness and amount). This is consistent with the descriptive analysis, where the sample size for mixed religious beliefs in Western religions and Chinese religions (Buddhism or Taoism) is zero. It indicates that mixers of religious beliefs do not involve a mix of Western and Chinese religions but rather a combination of Chinese folk beliefs and their religious practices.

4.1.4. Urban/Rural Heterogeneity

The CFPS project team re-evaluated the urban/rural attribute based on the standards released by the National Bureau of Statistics. The sample was divided into urban and rural groups, as shown in Table 6. Urban and rural residents with mixed religious beliefs exhibit a significantly positive impact on both the willingness and amount of household donations. Whether in terms of willingness or amount, urban residents surpass rural residents, possibly due to higher incomes in urban households, leading to higher donation willingness and capacity.

4.1.5. Regional Heterogeneity

In this study, the sample was divided into four groups according to the standards of the National Bureau of Statistics: northeast, central, western, and eastern regions, as shown in Table 6. Only in the western region, individuals with mixed religious beliefs show a significantly positive impact on both the willingness and amount of household donations. In the northeast region, mixers of religious beliefs exhibit a significantly positive impact on the willingness to donate. However, in the central and eastern regions, the impact is not significant. The western region, located in the western part of China, retains a relatively intact traditional culture, and the prevalence of mixed religious beliefs is higher, aligning with Li and Wang (2023) research, which suggests a higher probability of mixed religious beliefs in the western region, leading to a more apparent impact on household donations. The northeast region, with a prevalence of traditional folk beliefs, may result in a significant relationship between mixed beliefs and the willingness to donate.

4.2. Differences between Mixed and Non-Mixed Beliefs

The empirical results above indicate that, compared to households with exclusively non-mixed religious beliefs, households with mixed religious beliefs significantly and positively influence the willingness and amount of household donations. What exactly do non-mixed religious beliefs refer to? And, what are the differences between mixed religious beliefs and these non-mixed beliefs? In this section, based on questionnaire items, we introduce a variable called the “Mixed Religious Beliefs Subdivision”, where 1 represents exclusive Western religious beliefs, 2 represents mixed religious beliefs (simultaneously believing in Western and Chinese traditional religions), 3 represents exclusive Chinese traditional beliefs, and 4 represents no religious beliefs (set as the reference group). We use Probit and Tobit models to examine the impact of the “Mixed Religious Beliefs Subdivision” on household donations, and the results are presented in

Table 7. Regression results for mixed, exclusive Western, and exclusive Chinese religious beliefs.

Types	Donation Willingness	Donation Amount
	(1) Probit	(2) Tobit
Baseline group (non-religious)
Exclusive western	0.031 (0.28)	−0.071 (−0.13)
Mixed religious beliefs	0.113 *** (5.38)	0.491 *** (5.32)
Exclusive Chinese religious beliefs	0.047 *** (4.73)	0.214 *** (3.87)
Controlled variables	Controlled	Controlled
Samples	5962	5962

Note: The table reports marginal effects with t-values in parentheses. Significance levels are denoted by ***, indicating significance at the 1% levels, respectively. The marginal estimates are presented for Probit and Tobit models. Numeric variables such as donation amount, household income, and household expenses have undergone logarithmic transformation in all subsequent tables.

. Compared to households with no religious beliefs, exclusive Western religious beliefs show no significant difference in household donations. However, both exclusive Chinese traditional beliefs and mixed religious beliefs exhibit significantly larger effects. Contrasting the marginal effects of exclusive Chinese traditional beliefs and mixed religious beliefs, the latter has a greater impact on household donations. This suggests that in China, mixed religious beliefs primarily drive the promotion of religious effects on the willingness and amount of household donations.

Observing the significant role of mixed religious beliefs in charitable donations within Chinese households, one of the main questions Li and Wang (2023) aims to address is who is more inclined towards choosing mixed. In this section, using the binary variable of mixed religious beliefs as the dependent variable, the aforementioned control variables are incorporated into a regression model, employing the Probit model for the analysis. The results are presented in

Table 8. Regression results of factors influencing mixed religious beliefs.

Variables	Mixed Religious Beliefs
	Model 1	Model 2	Model 3
Age (Reference group: below 40 years old)
41–59 years old	0.010 (0.91)	0.015 (1.36)	0.020 (1.72)
above 60 years old	0.019 (1.36)	0.026 (1.90)	0.025 (1.70)
Gender (Reference group: female)	−0.016 * (−2.00)	−0.016 (−1.93)	−0.021 * (−2.28)
Education (Reference group: illiteracy)
Elementary school	−0.048 ** (−3.40)	−0.045 ** (−3.25)	−0.045 ** (−2.91)
Middle school	−0.040 ** (−2.80)	−0.034 * (−2.44)	−0.029 (−1.84)
High school	−0.034 * (−2.11)	−0.030 * (−1.87)	−0.023 (−1.28)
College and above	−0.048 ** (−2.60)	−0.035 (−1.83)	−0.034 (−1.65)
Marital status (Reference group: unmarried)	−0.034 * (−2.24)	−0.036 * (−2.30)	−0.035 * (−2.02)
Party membership (Reference group: non-Party membership)	−0.055 *** (−3.83)	−0.053 *** (−3.68)	−0.076 *** (−4.39)
Health status (Reference group: unhealthy)
General	−0.021 (−1.35)	−0.020 (−1.23)	−0.024 (−1.37)
Healthy	−0.025 * (−2.01)	−0.026 * (−2.10)	−0.028 * (−2.09)
Working status (Reference group: Unemployed)	0.002 (0.22)	−0.002 (−0.18)	0.003 (0.22)
Subjective economic status (Reference group: small)
General	0.004 (0.46)	0.003 (0.32)	0.005 (0.43)
Higher	0.015 (1.30)	0.010 (0.89)	0.012 (1.94)
Family size (Reference group: small)	0.000 (0.05)	0.001 (0.09)	0.001 (0.08)
Subjective confidence (Reference group: lower)
General	−0.007 (−0.37)	−0.013 (−0.65)	−0.019 (−0.86)
Higher	0.011 (0.60)	0.009 (0.49)	0.003 (0.15)
Household income	−0.004 (−0.87)	−0.002 (−0.43)	−0.003 (−0.71)
Housing value	−0.001 (−0.40)	−0.000 (−0.49)	−0.000 (−0.14)
Urban/Rural (Reference group: Rural)		−0.020 * (−2.30)	−0.020 * (−2.08)
Regions (Reference group: Northeast region)
Central region			−0.019 (−1.34)
Western region			−0.028 (−1.73)
Eastern region			−0.008 (−0.54)

Note: The table reports marginal effects with t-values in parentheses. Significance levels are denoted by ***, **, and *, indicating significance at the 1%, 5%, and 10% levels, respectively. The marginal estimates are presented for Probit and Tobit models. Numeric variables such as donation amount, household income, and household expenses have undergone logarithmic transformation in all subsequent tables.

. Controlling for other variables, the findings are as follows: (1) Regarding gender differences, the probability of males choosing mixed religious beliefs is smaller compared to females, indicating that females are more likely to be adherents of mixed religious beliefs. (2) In comparison to the illiterate or semi-literate group, only the primary education group shows a significant negative association with mixed religious beliefs, suggesting that those with lower educational attainment are more likely to embrace mixed religious beliefs. (3) Married individuals exhibit a smaller probability of choosing mixed religious beliefs compared to unmarried individuals, signifying that those unmarried individuals are more likely to adopt mixed religious beliefs. (4) Non-Party members are more likely to be adherents of mixed religious beliefs compared to Party members, indicating that those outside the Party are more likely to engage in mixed religious beliefs. (5) The probability of individuals in good health choosing mixed religious beliefs is smaller compared to the unhealthy group, implying that those facing health issues are more likely to embrace mixed religious beliefs. (6) Urban households show a smaller probability of choosing mixed religious beliefs compared to rural households, suggesting that rural families are more likely to be adherents of mixed religious beliefs. In summary, in rural areas, unmarried females with lower educational attainment, facing health issues, and not affiliated with the Party are more likely to adopt mixed religious beliefs. Clearly, the characteristics of the mixed religious group analyzed in our study differ from the conclusions drawn by Li and Wang (2023), primarily due to differences in the study subjects: Li and his team examined the mixing of Christianity with Chinese folk beliefs, while our study explores the blending of Western religions with Chinese deities, folk beliefs, and religious practices.

4.3. The Impact of Mixed Religious Beliefs on Household Donations

Is there a difference in the impact on household donations when mixing different numbers of religious beliefs? To address this question, our study includes a design with six types of mixed religious belief variables. However, due to the limited number of cases with four or more mixed beliefs (a total of 80 households), a new variable for mixed religious beliefs is constructed. Values 1, 2, and 3 represent mixing one, two, and three beliefs, respectively, while value 4 represents mixing four or more beliefs. For ease of comparison, value 5 indicates no mixing. Using Probit and Tobit models, with the reference group being the “no mixing” category, regression analyses were conducted on the sample, and the results are presented in

Table 9. The results of different types of mixed religious beliefs on household donations.

Variables	Donation Willingness	Donation Amount
Reference group: No mixed religious beliefs
Mixing one	0.087 * (2.19)	0.336 * (2.13)
Mixing two	0.035 (1.17)	0.141 (1.06)
Mixing three	0.123 ** (2.98)	0.538 ** (3.39)
Mixing above four	0.042 (0.87)	0.280 (1.30)
Covariates	Controlled	Controlled
N	5962	5962

Note: The table reports marginal effects with t-values in parentheses. Significance levels are denoted by **and *, indicating significance at the 5% and 10% levels, respectively. The marginal estimates are presented for Probit and Tobit models. Numeric variables such as donation amount, household income, and household expenses have undergone logarithmic transformation in all subsequent tables.

. The findings indicate that, after controlling for other variables, households with one and three mixed beliefs show a significantly positive impact on both donation intention and the donated amount compared to households without mixed religious beliefs.

5. Discussions and Conclusions

Individual donations serve as the cornerstone of charitable giving (Deng 2007), often emerging as decisions within the context of family interactions (Burgoyne et al. 2005). The motivations behind individual-centered family donations are exceptionally complex, influenced not only by individual physiological traits but also constrained by structural and environmental factors. Factors such as family endowments, socioeconomic status, mobilization methods of organizations, use of networks, community environment, and childhood experiences all impact individual donation levels (Zhou and Wu 2019; Zhu and Liu 2017; Liu 2004; Liu and Zhang 2021; Deng and Rong 2022; Hossain and Lamb 2017; Fang 2023; Cai et al. 2022; Stamas et al. 2020), especially socioeconomic status. Research indicates a U-shaped relationship between income levels and donations, with both low-income and high-income households donating a significantly higher proportion of their income to charitable organizations compared to middle-income households (Duquette and Hargaden 2021). Additionally, studies have found that the level of income inequality also affects donation levels, decreasing as income inequality deepens (Duquette and Hargaden 2021). However, as of now, differences in socioeconomic structures cannot fully explain the challenges faced by the development of donation levels in Chinese families. On the one hand, China’s per capita disposable income has visibly increased, but the proportion of individual and family donations has not seen a synchronous growth. On the other hand, China’s wealth gap is notably smaller than that of some developed Western countries, yet family donation levels have not surpassed those in those Western developed countries.

We believe that to enhance the level of family donations, nurturing and cultivating a distinctive charitable culture is key. Regardless of the country, if families are successfully mobilized and a unique charitable culture is formed throughout society, the level of family donations will continue to grow and be maintained at a relatively stable scale. In this regard, religion is an essential factor in the cultivation of charitable culture and can have a significant impact on individual and family donations (Li et al. 2022). It is important to further clarify the complex relationship between religion and the level of family donations to support the healthy and positive development of family donations. While direct studies on family donations are relatively scarce, academia pays close attention to the intricate relationship between religion and individual donations. Various aspects, such as religious types, values, functions, and donation mechanisms, have been richly explored, providing a crucial foundation for understanding family donations. These explorations have yielded fruitful results in revealing the cultural dynamics driving family donations, yet there is still room for further expansion. A crucial aspect is that, while current research has revealed the widespread phenomenon of “blending” religions in both Eastern and Western societies, there has been no exploration of the association between “blending” religions and family donations. Furthermore, past research has predominantly focused on exploring the correlation between religion and family donations. This study, in investigating the correlation between “blending” religions and family donations, attempts to use the instrumental variable method to explore the causal relationship between “blending” religions and family donations, examining the net impact of “blending” religions on family donations.

This study conducts empirical analysis utilizing data from the CFPS 2020. To address the relationship between religion and family donations, the study initially employs econometric methods to re-examine the relationship between religious beliefs and family donations. Subsequently, Tobit and Probit models are utilized to causally explore the relationship between “blending” religions and family donations. Building on this foundation, the study engages in nonlinear analysis and heterogeneity analysis concerning the relationship between “blending” religions and family donations, attempting to identify specific conditions under which “blending” religions influence family donation levels. Lastly, the study investigates the influencing factors and types of “blending” religions to comprehensively delineate the fundamental characteristics of individuals with blended religious beliefs and seek the specific mechanisms through which “blending” religions impact family donations.

The study reveals a significant causal relationship between religious beliefs and the willingness as well as the amount of family donations in a rationalized society. Individuals with religious beliefs are more inclined to donate or contribute larger amounts compared to those without religious beliefs, confirming the accuracy and reliability of previous research. Further analysis indicates a similarly significant positive impact of “blending” religions on family donations. In contrast to non-“blending” religious families, “blending” religious families exhibit higher willingness and donated amounts. However, the influence of “blending” religions on family donations is relatively limited. The analysis shows that the effect of “blending” religions on family donations is significant only for families with higher donation levels, specifically when the family donation level is above the 80th percentile. Additionally, the impact of “blending” religions on family donations varies with age, gender, religious identity, urban/rural differences, and regional disparities. The analysis suggests that the 41–59 age group, females, those without religious affiliation, urban residents, and families in the western and northeastern regions are more generous among “blending” religious households. Furthermore, “blending” religions form a collective, including various types of blends. Evidently, differences in “blending” types directly impact family donation levels. The study finds that only when blending one or three religions, do “blending” religious families show higher donation levels than non-“blending” individuals. As for the demographic groups more inclined to “blending” religions, the research indicates that households headed by individuals with lower educational attainment, non-Party members, those suffering from illnesses, and unmarried females are more inclined to choose a combination of religions.

The conclusion implies that rational choices remain the primary motivation for individuals to choose blended religious beliefs. In China, this applies not only to Chinese religions but also to Western religions. Many individuals identify as having no religious affiliation but engage in religious practices. They may not seek to join specific religious organizations or possess a formal religious identity. Instead, they hope to enhance their sense of control over future uncertainties, alleviate inner emptiness, and seek psychological comfort through religious practices (Pholphirul 2014). Based on this psychological motivation, Western religions gradually shed the cloak of independent religions in the process of localization in China, becoming more accepting of individuals with diverse religious backgrounds. In China, the pursuit of tangible returns in the present life is the primary reason for individuals to choose religion. Whether opting for a single religion or blending multiple religions, individuals are driven by a desire for practical benefits. However, compared to families choosing non-blended religions, those opting for blended religions have a stronger desire to address real-world issues through religious means. The choices of blended religious families reveal that in rural areas and among lower social classes with less social capital, there is a greater likelihood of choosing blended religions. These individuals tend to be less concerned about boundaries between religions and different sects, selectively believing in doctrines and values that they perceive as useful in improving their real-life situations. Both Western and Chinese religions generally embody rich altruistic principles and values. Adhering to these values is the path individuals choose to gain religious rewards. Therefore, even though members of blended religious families may face difficulties and challenges in their life circumstances, they still exhibit a strong willingness to donate and are willing to contribute more financially. However, our study also found variations in the impact of the number of blended religions on family donations. Families blending one or three religions exhibit a higher willingness to donate and contribute larger amounts. This suggests that blending is not arbitrary; rather, it is a purposeful selection. In ancient Chinese philosophy, the numbers one and three symbolize completeness and perfection, representing auspicious and beautiful meanings. This aligns with the notion that families blending one or three religions seek an ideal and harmonious life. When blending other numbers of religions, it indicates that individuals or families may not carefully choose the number of blended religions but rather make religious choices based on personal preferences, aiming to seek present-day rewards. However, at this point, the donation effects of blended religious families do not significantly differ from those of non-blended religious families.

Certainly, this study has some limitations in the research process. Firstly, it has not delved into the specific mechanisms through which blended religions influence family donations. Secondly, there is a lack of empirical support for the impact mechanisms of different types of blended religious combinations on family donations. Thirdly, there is a deficiency in cross-country comparisons regarding the influence of blended religions on family donations. The primary reason for this is the unavailability of data that adequately supports the extension of subsequent research. In future studies, efforts will be made to collect new data, expand the research scope, enhance mechanism analysis, and conduct cross-country comparisons, providing a comprehensive depiction of the impact of blended religions on family donations.