Introduction

The philanthropic sector is a significant force in the USA—in 2015, there were more than 1.56 million registered nonprofit organizations receiving over $1.98 trillion in annual revenue with approximately 14 percent of this coming from individual, foundation, and corporate giving (McKeever 2019; National Council of Nonprofits 2019). However, many nonprofit organizations led by or serving marginalized groups receive relatively little support from these funders. Studies have found, for example, that only 3.6 percent of foundation dollars go to nonprofit organizations led by people of color (Greenlining Institute 2006). Funders perennially underfund Native Americans; with, for example, only 0.23 percent of philanthropic funds awarded to Native-led nonprofit organizations, “despite the fact that Native Americans represent 2% of the national population and are among communities with greatest need in the US” (Barron et al. 2018, p. 1). Further, only 5 to 7 percent of foundation giving is earmarked specifically for programs and activities benefiting women and girls (Foundation Center and Women’s Funding Network 2009). Among individual donors, 14.6 percent report giving to a particular area that impacts women and girls (Women’s Philanthropy Institute 2016). Finally, according to a report from Funders for LGBTQ Issues, for every $100 awarded by US foundations, only 28 cents specifically supported LGBTQ issues (Kan et al. 2019). Democratizing philanthropy would entail increased giving to these historically marginalized groups to reflect their size and need.

Many giving circles (GCs) have been formed in the USA in response to this funding environment, with about half formed around a particular race/ethnic, gender, or gender-identity, often with the intention of increasing giving to similar identity groups (Bearman et al. 2017). GCs are forms of collaborative philanthropy in which members pool donations and decide together where these donations are given. They frequently include social, educational, and engagement opportunities for members, connecting them to their communities and to one another (Eikenberry 2009). For example, Washington Womenade holds regular volunteer-organized potluck dinners where attendees donate $35 to a fund that provides financial assistance to individuals (primarily women) who need help paying for things like prescriptions, utility bills, and rent. In 2002, a Real Simple magazine story (Korelitz et al. 2002) on Washington Womenade led to the creation of dozens of unaffiliated Womenade groups across the country. This article also inspired Marsha Wallace to start Dining for Women, which is now a national network of more than 400 chapters across the country in which women meet for dinner monthly and pool funds they would have spent eating out, to support internationally based grassroots programs helping women around the world. One of the most-often cited reasons people say they join GCs is to become more engaged in the giving process beyond writing checks—to interact directly with other donors and beneficiaries (Carboni and Eikenberry 2018; Eikenberry 2010; Eikenberry and Breeze 2015).

GCs in the USA are especially popular among groups not typically supported by traditional philanthropy (Carboni and Eikenberry 2018). Eikenberry (2009) found GCs in the USA attract people from diverse backgrounds, including those “new” to philanthropy, from diverse wealth-levels, diverse racial and ethnic identities, and especially women. However, diversity was more apparent across GCs than within. Recently, Bearman et al. (2017) conducted a landscape scan of US GCs and found 1087 independently run and currently active GCs, along with 525 GC chapters that are part of GC networks. This number has more than tripled since the last landscape study (Bearman 2007). Consistent with past research by Bearman (2007) and Rutnik and Bearman (2005), women’s GCs remain the most common type (48.5% of the total GCs identified) and women continue to be the majority of participants in GCs. Bearman et al. (2017) also found a growing number of Asian/Pacific Islander (53), African-American (40), and Latinx (19) groups. There has also been growth in the number of LGBTQ groups. In general, an increasing number of GCs have also been identified in Canada, South Africa, Australia, Romania, Bulgaria, the UK, Ireland, and various countries in Asia (Eikenberry and Breeze 2015).

Past research has not examined fully connections between donor identity and who receives support. Thus, the research focus of this article is on understanding how GC member identities (specifically associated with race, gender and gender-identity/sexual orientation) are associated with the identities of funding recipients (particularly People of Color [POC], women and girls, and people who identify as LGBTQ). Theories of philanthropic identity (Schervish 1995; Schervish and Herman 1988; Schervish and Havens 2001), bonding social capital (Putnam 1995, 2000), and subaltern counterpublics (Fraser 1990) suggest that GCs might play a role in reinforcing the connection between member identities and giving to groups with shared identities—what is discussed further below as bonding social capital. However, theories of bridging social capital and social ethics (Addams 1964) suggest GCs may, rather, play a role in expanding the connection between member identities and giving to groups of different identities. Our research questions are: Is GC membership associated with bonding or bridging social capital? That is, are GC members more likely than non-members to give to people who are like them or  to people who are not like them? Whether GCs lead to more bonding or bridging social capital has implications for the democratization of philanthropy by showing to what degree we might expect members to increase giving to historically marginalized groups. Identity is important to consider in relation to democratizing philanthropy because shared identity often drives where or to whom people give (Schervish and Havens 2001).

This research is important because as philanthropic approaches evolve, it is crucial to understand how new philanthropic tools and approaches impact society, in particular nonprofits and beneficiaries, and especially in the current context of social justice-oriented philanthropy. Can some of these new philanthropic tools help to increase and expand giving to groups not historically supported? In addition, community foundations and other philanthropic institutions in the USA, the UK, and elsewhere have devoted staff and resources to start and support GCs with the assumption that these groups will improve giving and its impact on communities. Examining the degree to which GCs might expand which groups are supported will help to understand if this is the case. This research also adds to broader discussions about the relationship between philanthropy and democracy.

The rest of the paper is organized as follows: First, we discuss various theories that help to explain philanthropic identity, social capital, and democracy. Next, we discuss who is supported by traditional philanthropy and what we know about this in relation to GCs. In the following section, we discuss the methodology. We draw on data from a survey of GC members and a comparison control group. The findings show that all respondents, including GC members and those not in GCs, are more likely to give to a shared identity group—related to race, gender, and gender identity/sexual orientation—leading to bonding social capital. However, GC members are also more likely than those not in GCs to give to groups that do not share their identity, when controlling for other variables, suggesting that GCs also encourage bridging social capital. We end with a discussion of the findings. We assert that our findings, considered with other research, indicate GCs may serve as builders of bridging and bonding social capital in communities and may be a path toward more democratized philanthropy by strengthening bonds within marginalized groups as well as creating bridges between privileged and marginalized groups.

Theoretical Framework

GCs have been described as “do-it-yourself” and “democratic” forms of philanthropy. Democratic forms of philanthropy can be understood in both an internal and an external sense. The first is related to the belief that direct and meaningful participation in organization governance is vital for creating a virtuous, engaged, and deliberative citizenry and democratic society. The second is related to who benefits. Direct and meaningful democratic participation involves the degree to which members (and those affected by a decision) have equal opportunities to participate in and learn the skills of agenda setting, deliberation, and decision-making in organizations. Past research suggests GCs enhance internal democracy. Eikenberry (2009) found, for example, that GCs generally provide opportunities for democratic participation within the group—providing opportunities for agenda setting, collaborative decision-making, and face-to-face deliberative discourse among members—and they build the capacities of members through education about voluntary organizations, community issues, and philanthropy.

Past research also suggests GCs have some impact on external democratic outcomes or who benefits. Data from a 2009 US survey of GC members and donors not in GCs found members of GCs were significantly more likely than a control group of donors not in GCs to give to organizations that support women; ethnic and minority groups; and arts, culture, or ethnic awareness (Eikenberry and Bearman 2009). Some of these data may be explained by the fact that GC member respondents were also more likely to be women or from communities of color than the control group respondents; however, when controlling for other variables, as length of time in a GC increased, respondents were significantly more likely to report giving to support women; ethnic or minority groups; and arts, culture, or ethnic awareness areas. Thus, some populations, such as women and girls and those from marginalized racial and ethnic groups, may be benefiting more from GCs than from traditional philanthropy.

However, extant research has not examined fully how identity might play a role in who benefits from philanthropy. That is the focus of this paper. Identity is important to consider in relation to democratizing philanthropy because shared identity, whether conscious or not, often drives where or to whom people give. Giving patterns reflect that people give to whom and to what they know and with which they are familiar, and to causes with which they can identify and are physically or emotionally attached (Ostrander and Schervish 1990, p. 74; Schervish 1995; Schervish and Havens 2001, p. 91; Schervish and Herman 1988). According to Schervish and Havens (2001): “the more closely donors are associated with charitable causes, and the more intensely donors feel the beneficiaries of their giving share a fate with them, the greater is the amount of charitable giving” (p. 91). Schervish and his colleagues have found that philanthropic commitment depends on individuals’ networks of social connection and the type and degree of shared identity they have with the needs of others (Schervish and Herman 1988; Schervish et al. 2001). Based on this literature, we hypothesize:

H1

People are more likely to give to causes that serve people with a shared identity (compared to causes that serve people of a different identity).

This theory of philanthropic identity seems to align with Putnam’s idea of bonding social capital. According to Putnam (1995), social capital includes the “features of social organization such as networks, norms, and social trust that facilitate coordination and cooperation for mutual benefit” (p. 67). He identifies two forms of social capital: bonding and bridging. As Putnam (2000, p. 22) explains, some forms of social capital are inward looking and tend to reinforce exclusive identities and homogeneous groups; this is bonding social capital. Bonding social capital can be important to internal democracy because it enables members to get to know one another and hence make listening and trusting easier (Putnam and Feldstein 2003). Nancy Fraser’s (1990) work on subaltern counterpublics highlights further the potential benefits of bonding social capital, especially for historically marginalized groups. Posed as an alternative to Habermas’ idea of the public sphere—where “a body of ‘private persons’ assemble to discuss [rationally and where inequalities are bracketed] matters of ‘public concern’ or ‘common interest’” (Fraser 1990, p. 58)—Fraser describes counterpublics as groups of all types who are excluded from the (white male) mainstream public sphere and create their own arenas for discourse. Identity-based GCs may serve as just such venues. Fraser advocates for multiple publics as a mechanism to enhance external democracy because these alternative publics or what she calls “subaltern counterpublics” (p. 67) can function as spaces of withdrawal and regroupment and as “bases and training grounds for agitation activities directed toward wider publics” (p. 68) and which can protect against assimilation. They might lead to participatory parity so that subaltern groups can participate equally in creating strong publics that “encompass both opinion-formation and decision-making” (p. 75). That is, subaltern counterpublics provide “protected” spaces where marginalized groups can support one another, create their own discourse or narrative about change, and provide supportive platforms for engaging in majority/mainstream public spaces. In the case of GCs, they may play a role in creating just such an arena among groups with shared identities, particularly for historically marginalized groups and engagement with “mainstream” philanthropy. Based on this literature, we hypothesize:

H2

GC membership will increase bonding social capital; that is, GC members will be more likely (compared to non-GC members) to give to causes serving those that share the same identity.

However, as noted above, past research has shown GCs might also play a role in expanding who benefits and thus may expand identification with others (Eikenberry and Bearman 2009). Putnam notes that some forms of social capital are outward looking and encompass people across diverse social cleavages; this is bridging social capital. Bridging social capital is important for external democracy because of its value in transcending social differences, including identity (race, ethnicity, religious tradition, sexual preference, and national origin) and status (vertical arrangements of power, influence, wealth, and prestige) (Wuthnow 2002, p. 670). The idea of bridging social capital is similar to Jane Addams’ (1964) philosophy of social ethics, which had to do with understanding the value of human life and the interconnected nature of society and emotions. Social ethical action is done through people working together cooperatively, not through individual action. To move from an individual ethics to a social ethics, Addams believed, one must immerse oneself in the direct experience of life as lived by people of all backgrounds. GCs might create bridging due to higher levels of engagement between members and beneficiaries. Based on this literature, we can hypothesize:

H3

GC membership will increase bridging social capital; that is, GC members will be more likely (compared to non-GC members) to give to causes serving those with a different identity than the donor.

Methodology

Our research questions, then, are: Is GC membership associated with bonding or bridging social capital? That is, are GC members more likely than non-members to give to people who are like them or to people who are not like them? To address these questions, we conducted a survey of GC members and a comparison control group in the USA to examine the impact of GCs on philanthropic behaviors. This survey (see questions in the Online Supplement) was administered online using Qualtrics. After testing and piloting the survey before it was finalized, we sent the survey to key contacts drawn from the GC database we created (see Bearman et al. 2017), asking these contacts to share the survey with their members. We also asked philanthropic professionals working with GCs, GC researchers, and other GC network and host contacts to share the survey link with GCs with which they were affiliated. The survey was open from mid-July to mid-August 2017. We sent reminder emails at two points during the time the survey was open. We ended up with 993 GC member respondents from 160 GCs. In addition to surveying GC members, we administered the same survey with a slightly revised title to a comparison group through Amazon Mechanical Turk (MTurk). MTurk enables researchers to pay a small fee for registered and approved users to complete tasks such as surveys as they choose. MTurk has become a popular data collection tool for social science researchers, with many of the most influential social science journals publishing articles using MTurk (Casey et al. 2017). MTurk samples tend to be more politically liberal, younger, less religious, and less racially diverse than the US population (Berinsky et al. 2012; Huff and Tingley 2015); however, research also finds that MTurk is a valid recruitment tool for psychological research on political ideology (Clifford et al. 2015) and MTurk samples produce valid estimates in the context of survey experiments (Berinsky et al. 2012; Buhrmester et al. 2011). We paid participants $.50 per completed survey, limiting the survey to US users. The survey was administered in September 2017. We had 947 respondents in this comparison group (see demographic information about the survey respondents in Table 1). Because of missing data, the sample for analysis is 1467 respondents (725 GC member, 746 non-GC member).

Table 1 Demographic descriptive statistics
Full size table

A limitation of the survey is that members self-select to participate in GCs, so results may reflect people’s predispositions to participate in these groups rather than the direct effect of belonging to the GC or selection bias. It is our position that there is something unique about GC members. This is supported by research cited above. Whether this is a predisposition of individuals to give in certain ways or that GC membership leads to different giving behaviors is outside the scope of this study and impossible to determine with cross-sectional data. Instead, we are examining whether GC members behave differently than their demographically similar non-GC member counterparts in terms of philanthropic behavior. Because of demographic differences in groups, we used propensity score matching on key demographic variables—discussed below—to ensure our models compared giving patterns in similar individuals in GCs and not in GCs.

For this study, we conducted two types of analyses to understand the association between identity and giving. First, we conducted Chi-squared analysis to get a sense of giving to shared or different identity groups among GC members and non-GC members to marginalized groups not traditionally funded by philanthropy. We examined giving specifically by POC and non-POC to POC-associated causes, women and men to women and girls causes, and LGBTQ and non-LGBTQ to LGBTQ causes. For example, for giving to women and girls’ causes, we examined giving differences between women and men to see if women were more likely to give to women and girls’ causes (H1). Then, we looked at differences between women in GCs and not in GCs and their giving to women and girls’ causes (H2) and men in GCs and not in GCs and their giving to women and girls’ causes (H3).

Second, we ran three logistic regression models to understand giving while controlling for demographic and other factors to examine all three hypotheses. In the logistic regression models, we tried to ascertain the effects of identity and giving for causes serving POC, women and girls, and the LGBTQ community, controlling for variables aside from respondent GC affiliation, race/ethnicity, gender, and gender identity. Dependent variables for the models were: giving to POC causes, giving to women and girls causes, or giving to LGTBQ causes. Independent variables were GC member status (GC member = 1), POC1 (POC = 1), (female = 1), or LGBTQ2 (LGBTQ = 1). We used propensity score matching to make the GC member and non-GC member samples comparable as there were some differences among groups. The matching was based on age, gender, education, marital status, and income. We include these terms in the model to ensure we fully account for discrepancies between groups (Rubin and Thomas 1996). We also added a measure of general social capital building to control for civic mindedness. This variable was a count (min = 0, max = 12) of civic and political activities the respondent reported participating in in the prior year. The list was predefined for respondents. We also included interaction terms between being white, male, and non-LGBTQ and GC membership to assess whether GC members gave differently based on belonging to a different cause than the group they gave to.

Analysis and Findings

Giving and Shared Identity

Our first hypothesis examines wether people are more likely to give to causes that serve people with a shared identity. We used Chi-squared analysis to examine giving by POC and not POC, women and men, and LGBTQ and cishet (Cisgender, heterosexual) people giving to causes serving POC, women and girls, and the LGBTQ community. In all three cases, we found support for H1 that people are more likely to give to causes with a shared identity: POC are significantly more likely than non-POC to give to POC causes, women are significantly more likely than men to give to women and girls causes, and LGBTQ people are statistically significantly more likely than non LGTBQ to give to LGBTQ causes. See Fig. 1.

Fig. 1
figure 1

Giving by identity

Full size image

In the logistic regressions (see Table 2), after controlling for demographic factors, we found POC, GC members, people who engaged in more civic and political activities, and men were significantly more likely to give to POC causes. In addition, women, GC members, and people who engaged in more civic and political activities were more likely to give to women and girls causes. Finally, LGBTQ people, GC members, people who engaged in more civic and political activities, men, and younger people were significantly more likely to give to LGBTQ causes. These findings provide further support for H1.

Table 2 Logistic regression results
Full size table

GCs and Bonding Social Capital

Our second hypothesis examines whether GC membership increases bonding social capital; that is, if GC members are more likely to give to causes serving POC, women and girls, and the LGBTQ community that share their same identity. Using Chi-squared analyses, we found GC members who were POC, women, or LGTBQ were significantly more likely to give to causes serving POC, women and girls and LGBTQ people, than non-GC members. Thus, in all three cases, we found support for H2. See Fig. 2.

Fig. 2
figure 2

Giving by identity and GC membership

Full size image

In the logistic regressions (see Table 2), after controlling for demographic factors, we found related to POC that the interaction term was statistically significant and positive, indicating POC within GCs are significantly more likely to give to POC causes than people not participating in GCs. These findings indicate that POC engage in bonding social capital in the general sample and the effect is stronger within GCs (support for H2). However, regarding gender, the interaction term was not statistically significant. These findings indicate that women engage in bonding social capital regardless of GC membership (not supportive of H2). Similarly, for gender identity, the interaction term was not statistically significant, indicating that LGBTQ people also engage in bonding social capital regardless of GC membership (not supportive of H2). Thus, our regression results provide a somewhat mixed picture, with only partial support of H2 related to giving to POC causes.

GCs and Bridging Social Capital

Our third hypothesis examines whether GC membership increases bridging social capital; that is, if GC members are more likely to give to causes serving POC, women and girls, and LGBTQ people when they have different identities. Using Chi-squared analyses, we found GC members who were white, men, or cishet were significantly more likely to give to causes serving POC, women and girls, and LGBTQ than non-GC members. Thus, in all three cases, we found support for H3 (see Fig. 2).

In the logistic regressions (see Table 2), after controlling for demographic factors, we found white GC members are significantly more likely than white non-GC members to engage in bridging forms of social capital when it comes to giving to POC causes. GC members are also significantly more likely to give to women and girls’ causes regardless of gender when we control for other factors and male GC members were significantly more likely to give to women and girls causes than male non-GC members. This means that GC members who are men are more likely than non-GC members to engage in bridging forms of social capital when it comes to giving to women and girls’ causes. Finally, cishet GC members are not significantly more likely than cishet non-GC members to give to LGBTQ causes though GC members are significantly more likely to give to LGBTQ causes than non-GC members. All this can be taken to mean that GC members are more likely than non-GC members to engage in bridging forms of social capital when it comes to giving to POC, women and girls, and LGBTQ causes, though the findings for LGBTQ indicate it is GC member status that is driving this phenomena rather than the interaction between GC and cishet. These findings provide further support for H3.

Discussion and Conclusion

This research examined findings related to a recent survey on the impact of GCs in the USA, focusing in particular on understanding how identity (specifically associated with race, gender, and gender identity) is associated with funding recipients (particularly POC, women and girls, and people who identify as LGBTQ), considered in the context of theories of philanthropic identity, bridging and bonding social capital, and the democratization of philanthropy. While we focused on specific identities, we believe our findings on identity-based giving could be germane to other identities as well.

Our findings show that all respondent groups are likely to give to their shared identity group (bonding), including GC members and those not in GCs (supporting H1 and largely H2). Regarding H2, Chi-squared results showed clear support but we found somewhat mixed results with the logistic regression results where POC within GCs are significantly more likely to give to POC causes than people not participating in GCs, but regarding gender and gender identity, the interaction term was positive but not statistically significant. Finally, we found that GC members are also more likely to give to groups that do not share their identity, when controlling for other variables, particularly related to race and gender, suggesting that GCs also encourage bridging social capital (supporting H3). These findings suggest GCs can make a difference in growing philanthropy among people of different genders and other identities historically not well funded by philanthropists, largely supporting both bonding social capital and bridging social capital (Putnam 2000; Wuthnow 2002). While these estimates are not causal because of the cross-sectional nature of the data, they do suggest a strong link between identity and giving. Other work has also shown that GC members give locally (Bearman et al. 2017) and that they give more strategically and are likely to be engaged in community beyond financial gifts (Carboni and Eikenberry 2018). Our findings considered with other studies indicate that GCs may serve as builders of bridging and bonding social capital in communities beyond financial donations and may be a path toward more democratized philanthropy by strengthening bonds within marginalized groups as well as creating bridges between privileged and marginalized groups.

Specifically, findings suggest when donors give on their own, they are likely, as Schervish and colleagues suggest, to give to those with which they identify, which could play a role in reinforcing inequalities since those with the most resources to give are wealthy, white, and male (Badger 2017; Traflet and Wright 2019). However, if donors give as part of a GC, they may be likely to still give to people like them but also expand their giving to give to people who are not like them. This aligns with previous findings showing that GCs cause members to both increase and expand their giving (Carboni and Eikenberry 2018; Eikenberry 2009; Eikenberry and Bearman 2009).

This is good news for the possibility of democratizing philanthropy. In the current context of philanthropic and wider societal attention to empowering marginalized groups, supporting GCs presents philanthropy with a way to support and expand social justice and equity in philanthropy. Of particular note is that GCs are relatively easy to set up and more nimble than mainstream philanthropic foundations. Supporting and promoting GCs, then, may be a means to expand and shift giving to historically underserved groups in philanthropy, such as to women and girls and POC. As GCs based around marginalized identities continue to grow, we might also expect that giving to their shared identity groups will also grow. Enhancing bonding across groups enables members to get to know one another and hence make listening and trusting easier (Putnam 2000; Putnam and Feldstein 2003), while such subaltern counterpublics create alternative spaces for marginalized groups to support one another and create their own arenas for non-assimilated discourse that can perhaps lead to participatory parity in mainstream public spheres (Fraser 1990).

Simultaneously, GCs seem to enhance bridging social capital (Putnam 2000; Wuthnow 2002) and a social ethics (Addams 1964). This is good because of its value in transcending social differences, including identity (race, ethnicity, religious tradition, sexual preference, and national origin) differences. This may mean that non-marginalized GC members may be more likely to support marginalized groups not typically funded by mainstream philanthropy. This also suggests that philanthropic identity can be expanded through mechanisms such as GCs. Future research should examine whether particular activities in GCs do more or less to build bonds or bridges. It might also increase research on the role philanthropic institutions, such as community foundations, or local governments, might play in promoting and expanding philanthropy to direct funding to marginalized groups. Additional research on whether bridging philanthropy reduces prejudice against different identities and whether these findings hold for other identity groups is also worth pursuing.

As noted in our methodology section, a limitation of the survey is that members self-select to participate in GCs, so results may reflect people’s predispositions to participate in these groups rather than the direct effect of belonging to the GC or selection bias. It is our position that there is something unique about GC members. We used propensity score matching and controls for demographic variables to account for this in our models, but more research is necessary to understand substantive differences between GC members and non-GC members.

Notes

  1. 1.

    Respondents were allowed to list multiple races/ethnicities. This variable is constructed from all people who did not identify as solely white or did not decline to comment.

  2. 2.

    Respondents asked to self-identify as LGBTQ. It is possible the sample is biased.