Racial and gender disparities among highly successful medical crowdfunding campaigns

Abstract

There has been growing recognition of the popularity of medical crowdfunding and research documenting how crowdfunding arises from, and contributes to, social and health inequities. While many researchers have surmised that racism could well play a role in medical crowdfunding campaign outcomes, research on these dynamics has been limited. No research to date has examined these dynamics among the most successful medical crowdfunding campaigns, focusing instead on average users’ experiences or specific patient subpopulations. This paper analyzes key characteristics and demographics of the 827 most successful medical crowdfunding campaigns captured at a point in time in 2020 on the popular site GoFundMe, creating the first demographic archetype of “viral” or highly successful campaigns. We hypothesized that this sample would skew towards whiter, younger populations, more heavily represent men, and reflect critical illnesses and accidents affecting these populations, in addition to having visually appealing, well-crafted storytelling. Analysis supported these hypotheses, showing significant levels of racial and gender disparities among campaigners. While white men had the greatest representation, Black and Asian users, and black women in particular, were highly underrepresented. Like other studies, we find evidence that racial and gender disparities persist in terms of campaign outcomes as well. Alongside this quantitative analysis, a targeted discourse analysis revealed campaign narratives and comments reinforced racist and sexist tropes of selective deservingness. These findings add to growing calls for more health research into the ways that social media technologies shape health inequities for historically marginalized and disenfranchised populations. In particular, we underscore how successful crowdfunding campaigns, as a both a means of raising funds for health and a broader site of public engagement, may deepen and normalize gendered and racialized inequities. In this way, crowdfunding can be seen as a significant technological amplifier of the fundamental social causes of health disparities.

1. Introduction

Digital platforms have become a major factor in how people find material and emotional support for health care needs (Kneese and Merid, 2018; Zenone, Kenworthy & Maani, 2022). The growing use of charitable crowdfunding platforms, such as the popular site GoFundMe (GFM) reflect this trend. Charitable crowdfunding platforms allow individuals, families, and organizations to create fundraising campaigns that are then easily distributed and accessible via social media. With billions of dollars in transactions processed and an estimated 90% share of the market in the US, GoFundMe is an increasingly prominent hub in the digital health ecosystem (Harris, 2018). Medical crowdfunding makes up more than a third of all fundraising efforts on GoFundMe (GoFundMe, 2020), and has become a common means of paying for medical related costs, especially in places with inadequate health insurance coverage and lack of social safety-net systems, such as the US (Coutrot et al., 2020; Kenworthy and Igra, 2022; Snyder et al., 2020).

Ample research has documented how crowdfunding platforms exacerbate social and health inequities; yet no research to date has specifically studied disparities among highly successful campaigns (Barcelos, 2020; Cohen et al., 2019; Kenworthy et al., 2020; van Duynhoven et al., 2019). This project, examining a subset of the most successful medical crowdfunding campaigns, aimed to answer three questions: first, what is the demographic makeup of highly successful medical campaign users? Second, are there significant differences between the proportions of users by race, gender and age groups found in the sample and the proportion represented in the US population? Third, do campaign outcomes vary by different demographic groups, including users of different races, genders, and ages, or by medical conditions? Finally, we use the answers to these questions to qualitatively assess how highly successful crowdfunding campaigns may be contributing to the further framing of users as “deserving” and “undeserving” of support based on racialized and gendered disparities.

Much of the research on medical crowdfunding to date has focused on representative samples of campaigns or subgroups of campaigns for specific causes or health conditions. These studies have highlighted disparities in crowdfunding outcomes that are often driven by social and structural inequities, including economic inequalities, racism, and sexism (Barcelos, 2020; Cohen et al., 2019; Kenworthy et al., 2020; Rhue and Clark, 2016). In this paper, we contribute to this growing literature using a novel approach: we identify the most successful medical crowdfunding campaigns and analyze these campaigns’ race, gender, medical condition and financial outcome, compare race and gender representation with the US population at large, and evaluate how these characteristics correlate with disparities in campaign outcome. Highly successful campaigns attract the attention and donations of, typically, much larger audiences. This study is the first to examine the characteristics of campaigns that attract this attention and support, and in doing so, have an outsized public visibility as paragons of medical crowdfunding success. As the popularity of medical crowdfunding grows, understanding its implications for racialized and gendered groups will be essential for addressing health inequities fueled by digital media. In addition, because these high-profile campaigns have the potential to shape broader public perceptions of deservingness and health care needs, studying them can uncover new directions for scholarship on the broader political and social impacts of medical crowdfunding.

1.1. Background

Research on crowdfunding has described its growing use across many different countries and highlighted concerns about how it may be contributing to inequities in financial relief and health disparities, especially where other social supports are thin (Berliner and Kenworthy, 2017; Kenworthy et al., 2020; Lukk et al., 2018; Snyder et al., 2017). Most campaigns do not reach their goals, and new research indicates many are garnering no donations at all (Kenworthy and Igra, 2022). Crowdfunding outcomes have consistently been associated with higher incomes, better education, and greater material and social capital within networks (Cohen et al., 2019; Igra et al., 2021; Igra, 2021; Lee and Lehdonvirta, 2020; Silver et al., 2020; van Duynhoven et al., 2019). While several studies have observed disparities in campaign outcomes by race and gender, these dimensions of inequity have not been explored in depth. In a study of a randomized sample of 637 GoFundMe medical campaigns, Kenworthy et al. (2020) found that non-white users were under-represented and that Black beneficiaries received significantly smaller average donations. In addition, nearly four out of five campaigns started for someone else were organized by women, indicating a largely invisible form of gendered online care labor occurring in this marketplace. In a recent paper, Igra (2021) provided important insights into racial disparities in crowdfunding by showing how these are largely attributable to differential access to financial capital within close networks, which are shaped by conditions of racialized capitalism and structural racism. A cross-sectional study of medical crowdfunding campaigns in Canada, the UK, and the US found that Black recipients had strong disadvantages in fundraising, and that men also raised more than women to a smaller degree (Saleh et al., 2020).

Most crowdfunding campaigns circulate among a relatively limited social network of known contacts, and thus their success is shaped by the forms of capital - material, social, cultural - within those networks (Berliner and Kenworthy, 2017; Giudici et al., 2013; Lukk et al., 2018; Paulus and Roberts, 2018; Silver et al., 2020). As Igra’s (2021) research demonstrates, “donor financial capacity” explains much of the differential outcomes by race among campaigns, but not all of them. It is likely that the remainder may be due to disparities in other forms of networked capital (education, social influence, or technological literacy) as well as perceptions of deservingness that are deeply shaped by social biases, discrimination, and social norms (Berliner and Kenworthy, 2017). Scholarship on “health-related deservingness” as well as deservingness within the context of crowdfunding has highlighted how historical patterns of discrimination, contemporary social mores, and public perceptions of marginalized groups, can deeply impact willingness to allocate resources to those in need, either through social programs or through newer forms of digital technology like crowdfunding (Barcelos, 2020; Berliner and Kenworthy, 2017; Halcomb, 2022; Sargent, 2012; Williams and Mohammed, 2009). For decades, studies have demonstrated that race influences dominant social perceptions of who is deserving of help (Benson et al., 1975; Saucier et al., 2005) even when white respondents believe themselves to hold non-racist views (Nosek et al., 2010). Current research has focused on the complex and ambiguous nature of racial bias where racialized perceptions of who are deserving of help are often concealed in the language of fairness, which others describe as aversive racism (Chen et al., 2021). Further, racism has been identified as a factor influencing attitudes towards who deserves social assistance (Snowden and Graaf, 2019), housing assistance (Tighe, 2012), and charitable giving (Fong and Luttmer, 2011).

The role that digital technology platforms may play in amplifying, altering, or even diminishing the dynamics of racism is complex. But as scholars like Ruha Benjamin (2019) have observed, new technologies frequently reinforce racial codes, and therefore contribute to racial inequities, even as they purport to be neutral and objective. Rather than insulating users from racism, digital technologies often amplify it. Studies of Kickstarter, for example, have documented how donor perceptions of campaigner race, and the prominence of their race in campaign images, strongly influenced campaign outcomes (Rhue and Clark, 2016; Younkin and Kuppuswamy, 2019). More broadly, discrimination, the amplification of hate, and targeted harassment of people of color, women, and trans people online are widely documented by social media scholars (Vogels, 2021). At the same time, the labor, creativity, and culture-making efforts made by marginalized groups have been foundational to the value, success, and entrenchment of digital platforms in everyday life (Brock, 2020; Duffy, 2016).

In this paper we examine highly successful campaigns that, by their very nature, circulate through larger networks of people, often gaining significant social media spread and public exposure. As scholars of “libidinal economies” have noted, social media marketplaces oriented toward growing profits for shareholders have given rise to increasingly competitive environments for our attention (and by extension, money) in crowded media landscapes (Brock, 2020). Yet social media is also a space where subjective desires gain collective expression and amplification: in the case of crowdfunding, it is a marketplace driven not by supply and demand, but by subjective metrics of (socially-biased) deservingness (Barcelos, 2020; Berliner and Kenworthy, 2017; Brock, 2020). Algorithms, platform design choices, and steep social hierarchies among content creators shape what attracts, retains, and produces value from attention. Research has pointed to ways that direct and indirect discrimination produce inequities in how crowdfunding content is picked up, spread, and amplified (Kenworthy et al., 2020; Van Duynhoven et al., 2019).

In this study, we hypothesize that among highly successful medical crowdfunding campaigns, people from historically marginalized racial groups, women, and those with stigmatized health conditions will be underrepresented, likely as a result of technologies that amplify social biases, discrimination, and financial inequities. We also hypothesize that the demographic makeup of highly successful campaigns will be less representative of the true US population at large, and lastly, that campaign outcomes will vary depending on race, gender, age, and or by medical condition. Following the work of Link and Phelan (1995), we posit that crowdfunding exacerbates and amplifies the impacts of social conditions that are “fundamental causes” driving health inequities. Fundamental causes, they write, “[involve] access to resources, resources that help individuals avoid diseases and their negative consequences through a variety of mechanisms” (p. 81). While Link and Phelan focus on the ways that technologies can be differentially relied on to help reduce health risks, depending on one’s social position, in this paper we argue that they can also be differentially relied upon to avoid the negative financial fall-out of disease and attract positive social support, again depending on social position and privilege. This study underscores the importance of digital social media resources, which, while purporting to be free and open to all, often powerfully reinforce and amplify the negative consequences of fundamental causes on health, and undermine efforts to address health disparities. Our results offer important evidence for current public health understandings of health and social support in digital environments, and show how digital spaces can reinforce and exacerbate the harmful inequities that arise from fundamental social causes of health.

2. Methods

To conduct this study, we assessed key characteristics and demographics of the most highly successful medical crowdfunding campaigns on GoFundMe. Because GoFundMe is estimated to control more than 90% of the US, and 80% of the global, social crowdfunding markets, our study focused on this platform (Harris, 2018). The Human Subjects Division at the authors’ university determined that Institutional Review Board approval was not required for this study because only publicly available data was used and did not involve interactions or interventions. Additional efforts were undertaken to protect the data anyways: All data used in this study were stored on password-protected drives and all data coders completed ethics training before accessing data. Below, when discussing individual campaigns, we use only details that are strictly necessary for analysis, as the highly public nature of many of these campaigns makes them particularly hard to anonymize.

2.1. Sampling criteria and exclusions

We first used a targeted sampling process to identify all highly successful or “viral” medical campaigns on the popular crowdfunding site GoFundMe at a single point in time, February 2020, from a database of GoFundMe crowdfunding campaigns (Snyder, n. d.). For the purpose of this research, highly successful campaigns were identified as those that raised more than a baseline of $100,000 in donations; this yielded 858 campaigns or .02% of overall medical crowdfunding campaigns in the sample. GoFundMe only allows users from 17 countries, most in North America and Europe, to create campaigns. Most of the campaigns that reached $100,000 or more in donations were from the US; a small portion was from other countries where the currency exchange was slightly less or more than USD, making for a relatively similar earnings comparison. In Nordic countries (n = 4, 0.5%) where exchange rates were far different, we only included campaigns that raised the equivalent of $100,000 or more at an average 2020 exchange rate. Once the baseline was established, we excluded an additional twelve campaigns that showed evidence of duplicate campaign pages, fake campaign pages, or offline reported donations in very large sums, which made the campaigns appear to have raised far more money than they actually did. To keep our analysis focused on medical fundraising for personal medical expenses, nineteen campaigns raising funds for charities, large groups, or humanitarian initiatives were excluded from the sample. This resulted in 827 campaigns included in the final sample.

2.2. Variable coding and procedures

An automated program was developed and used to collect quantitative and text data from each campaign page in the sample in February 2020; this data was archived along with campaign page screenshots.

Between May and July 2020, we reviewed campaign pages and used coding to create three demographic variables: race, gender, and age. Recipients’ perceived race was coded by three researchers from different racial backgrounds using five categories: white, Black, Asian (including South Asian), other person of color (POC), and unknown. While these categories are imperfect and do not fully capture the vast diversity or complexity of racial identity, experience, or discrimination in the context of the US, research shows that simple racial categories can still map significant social and health inequities (Baciu et al., 2017; Williams and Mohammed, 2009). For this research, the perceived recipient race is likely more relevant in understanding how campaigners are perceived and treated by donors than knowing campaigners’ self-identified race. The perceived gender of the campaign recipient was determined using pronouns, gender descriptors, users’ names (which were compared with the US Census baby names list by gender), and stated relationships (i.e., “mother”, and “daughter”) from the campaign text. Recognizing trans people’s right to self-identification of gender, and realizing that users may not disclose their status as transgender due to personal, privacy, or safety concerns, campaigns were coded according to the participants’ self-described gender as women, men, genderqueer, non-binary, or transgender women/men. If there was disagreement among the sources or insufficient data available to make a conclusive judgment, the gender was then marked as “unknown.” Recipient age was coded as an adult if 18 or older, or a child if under 18, using explicit written information on the campaign page (for example, “My eight-year-old son …“); when information was not clear or available variables were coded as “unknown.” Each campaign was coded for these three variables (gender, race, age) by three different researchers. In campaigns with multiple recipients amongst whom these demographics differed, variables were coded as “multiple people.” Intraclass correlation tests of this coding process using a two-way random effects model in SPSS showed very high correlation: For gender ICC = 0.874; for age ICC = 0.936; and for race ICC = 0.872. We then created a “compiled” variable codebook, which represented the composite results of our group coding efforts for all variables, resolving any divergent variable codes using a standardized process.

In addition, we created a variable called “medical issues motivating campaigns” to provide further description of the medical circumstances leading to GoFundMe campaigns. A two-stage qualitative coding process was used to identify categories for the variable medical issues motivating campaigns. First, descriptions of the motivating medical issue from the campaign page were recorded. Then, these medical issues were grouped into thematic sets, guided by a widely used health issue categorization system which identifies 21 categories of health issues such as strokes and injuries or accidents (UK Clinical Research Collaboration, 2019). When naturally emerging themes in the health data did not align with existing medical categorizations, additional categories were created. For example, transplants are a significant category of medical need on the GoFundMe platform due to the considerable cost involved with the process, and the pressure often applied to patients to fundraise for the cost independently. Transplants were retained as a separate category even though they are typically categorized as an issue related to the organ requiring transplant. Once the categories were developed from the data, each case was re-reviewed and coded according to the new categories.

To answer our third question regarding campaign outcomes, we collected data on 4 measures of campaign spread and success, since there is no single definition of campaign success for crowdfunding. These 4 variables were: 1) amount of money raised; 2) number of individual donations; 3) average campaign donation amount; and 4) social engagement measures such as number of campaign page shares and updates.

2.3. Data analysis

All analysis took place in R Core Team (2022). To analyze our first research question (“what is the demographic makeup of highly successful medical campaign users?“) we conducted frequency and descriptive statistics by medical campaign stratified by race, gender, age and medical condition. To analyze the second question (“are there significant differences between the proportions of users by race, gender and age groups found in the sample and the proportion represented in the US population?“) we ran a two-tailed one sample z test for proportions, with the exception of Black race recipients (n = 26) where a single sample two-tailed t-test was used due to the sample size of Black race recipients being too small for a reliable z test. To analyze our third question (“do campaign outcomes vary by different demographic groups, including users of different races, genders, and ages, or by medical conditions?“) we compared campaign outcomes (total amount raised, number of donations, donation sizes, and campaign shares) between campaigns in these different demographic groups.

Finally, we conducted a targeted qualitative discourse analysis of campaigns to better understand social factors impacting these campaigns’ success compared with other campaigns in the sample. This analysis reviewed all campaigns, but focused on discourses surrounding campaigns for women of color, exploring how these campaigns narrated deservingness and how commenters and donors responded to these narratives. An inductive thematic coding process was used to code for themes of deservingness, blame, care, and racial and gender biases.

3. Results

3.1. Crowdfunding demographics

The results of our analysis for question one revealed the demographics and characteristics of campaigns the sample, which included highly successful US (n = 519) and non-US (n = 308) medical campaigns. As shown in Table 1, we found an unequal gender balance among campaigns, with 55.9% for men (n = 462), 41.0% for women (n = 339), and 3.1% for multiple people (n = 26). The sample did not contain any campaigns that explicitly revealed transgender identity, a phenomenon worth noting given the growing popularity of crowdfunding for transgender health needs that may include gender confirmation treatment/therapies (Barcelos, 2020). White people accounted for 80.3% of overall campaign recipients (n = 665), while Black people represented only 3.1% of recipients (n = 26). Strikingly, only 5 recipients were Black women. Asian (including South Asian) people made up 6.2% of recipients (n = 51), while other people of color represented 6.9% (n = 57). Adults comprised the majority of the campaign recipients (68.1%, n = 564), and children under eighteen years of age represented 29.6% of recipients (n = 245). Age could not be determined for six recipients (0.7%), while campaigns with multiple recipients of different ages represented 1.6% (n = 13) of the sample.

Table 1. Campaign demographics and motivating medical issues.

Includes campaigns from US and non-US countries. For detailed information about how medical categories are defined, see UK Clinical Research Collaboration (2019).

Demographics and Motivating Medical Issues	N	Percent (%)
Recipient Gender	827
Man	462	55.9%
Woman	339	41.0%
Multiple people	26	3.1%
Recipient Race	827
White	665	80.4%
Black	26	3.1%
Asian (incl. South Asian)	50	6.0%
Other person of color	57	6.9%
Unknown	29	3.5%
Recipient Age	827
Adult (18 + )	563	68.1%
Child (under 18)	245	29.6%
Unknown	6	0.7%
Multiple people	13	1.6%
Motivating Medical Issue	827
Cancer	396	47.9%
Accidents, injuries	141	17.0%
Congenital conditions	74	8.9%
Violent crimes	35	4.2%
Other neurological, neurodegenerative diseases	34	4.1%
Infectious diseases, critical illnesses caused by infection	34	4.1%
Cardiovascular, pulmonary issues	27	3.3%
Transplants (except as cancer treatment)	20	2.4%
Stroke, aneurysm	16	1.9%
Unknown/unstated	14	1.7%
Other physical disabilities	13	1.6%
Other inflammatory, immune diseases	8	1.0%
Other physical disabilities	8	1.0%
Behavioral, mental health issues	5	0.6%
Endocrine diseases, issues	1	.1%
Pregnancy and childbirth	1	.1%

We also examined the medical issues that motivated these highly successful campaigns. Cancer-related treatments and care were a leading issue across the sample at 48% (n = 397), followed by accidents and/or injuries at 16.9% (n = 140). Other conditions that make up a large portion of the global burden of disease – such as cardiovascular conditions, stroke, and pregnancy and childbirth issues – comprise much smaller proportions of the motivating medical issues for these highly successful campaigns. Physical disabilities, endocrine issues such as diabetes, and behavioral and mental health issues – all of which face significant social stigmas – comprise some of the smallest categories. In Supplemental Table A, we providea breakdown of motivating medical issues by race, gender, and age. The most common medical motivation for children under 18, as for adults, was cancer, in 48% of cases. It’s notable that across all race and gender categories the top two motivations for campaigning were related to cancer or personal injury (accidents, injuries, violent crime victims). This may speak to the unexpected nature of such health needs as well as their substantial out-of-pocket costs. Particularly in the US, such unexpected health conditions may come with substantial costs given limited social safety nets and the high costs of care.

3.2. US sample vs. US population analysis (representation)

Our second question asked how campaign demographics compare with demographics at large in the US, to further assess for evidence of demographic disparities in this sample of highly successful crowdfunding campaigns. As shown in Table 2, we compared the demographics of US campaigns from our sample (n = 519) with the demographics of the US population at large. A two-tailed, one sample Z-Test was performed to assess the difference in proportion between the sample mean for US campaigns for men and the true combined states mean for men, according to the US census (U.S. Census Bureau, 2020). US male recipients (n = 320) accounted for 61.7% of the sample, while the true proportion of males in the US is 49.2%; the results of the Z-Test showed significant differences with the raw score z = −17.40, p = .00001. USwomen (n = 177) represented 34.1% of the sample and 50.8% of the true US population; the results also supported statistical significance (z = −24.3, p < .00001).

Table 2. Sample US population versus True US population comparison.

26 campaigns with multiple recipients, 29 with “unknown” recipient race, 6 campaigns with unknown recipient age, and 13 campaigns that had multiple recipients were excluded from the table along with non-US campaigns (n = 308) for the US population vs sample population comparison. Source for US population data: US Census Bureau, American Community Survey (2019). A singl e sample two-tailed T-test for means was conducted comparing the real U.S. Black populations to the sample mean for Black recipients (t (19) = 4.41 p = .00001) rather than z statistic due to sample size limitations.

	N	Percent	US Population at large (%)	Comparison of proportion (z statistic)	p value (α = 0.05)	Df
Recipient Gender
Man	320	61.7%	49.2%	−17.40	.00001	319
Woman	177	34.1%	50.8%	−24.3	.00001	176
Recipient Race
White	405	78%	75.3%	−298.8	.00001	404
Black	20	3.9%	14.0%	4.41	.00001*	19
Asian (incl. South Asian)	36	6.9%	6.6%	−115.7	.00001	35
Other person of color	36	6.9%	7.6%	−150.1	.00001	35
Recipient Age
Adult (18 + )	377	72.6%	77.4	−568.7	.00001	376
Child (under 18)	125	24.1%	22.6%	−481.4	.00001	124

A two-tailed one sample Z-Test was performed to determine if the proportion of US race groups in the sample differed significantly from the combined states US population at large, shown in Table 2. Significant differences in population means were found among recipients identified as white persons, who had significantly greater representation in the sample (z = −298.8, p < .00001, two-tailed) than in the US population. The sample size of Black recipients (n = 20) was too small to reliably use the z-test for proportion comparison to the U.S. population. A single sample two-tailed T-Test for means was conducted comparing the US Black population to the sample mean for US Black recipients (t (19) = 4.41 p = .00001). The result showed Black persons were significantly under-represented in the sample compared to the general population.

Using the Z-tests for comparing age groups, analysis revealed noticeably lower proportion of adults (persons over 18 years) in the sample (z = −568.7, p < .00001, two-tailed) compared to the US. Children (persons under 18 years) had greater representation compared to the US population, (z = −481.4, p < .00001, two-tailed).

When race, gender and age were taken into account simultaneously, disparities in representation became even more clear. As seen in Fig. 1, Adult white male recipients (76.25% of US men in the sample, n = 244) saturated the study sample, while the actual US population of white men is around 49.2% (U.S. Census Bureau, 2020). By contrast, US Black adult women were the least represented demographic in the sample at 1.5% (n = 3), though they make up 12.9% of the US population per US Census.

Fig. 1.

Percentage of White men vs. Black women in sample and US at large.

3.3. Campaign characteristics and outcomes

To answer our third research question about whether campaign outcomes differ between demographic groups, differences in campaign interactions and outcomes were analyzed based on categories of recipient race, gender, age, and medical condition. We used three different variables to measure campaign outcomes: The median amounts campaigns raised, the number of individual donations received, and the average individual donation amount given to campaigns. When it came to campaign interaction in the form of social media shares, the number of times a fundraiser was shared did not vary significantly in the sample across demographic categories, consequently, this data is reported in Supplemental Table B.

As shown in Table 3, recipients identified as white men had the highest median amount raised at $186,180; this was $30,000-$50,000 more than the other race and gender groups. White men also received the largest median individual donation amounts at $131 per donation, but had relatively low numbers of donations. By comparison, white women had lower median amounts raised ($134,514) and a lower median donation amount ($109), but higher median donations per campaign (1,230). Men identified as other persons of color had the lowest median amount raised at $132,388, and Black women had the lowest median number of donations. Campaign recipients identified as other persons of color or Asian (including South Asian) received the largest median number of donations made but still raised less than other race groups, in large part because median donation amounts were smaller. Though campaigns across the sample were shared roughly similar amounts, demographic groups raised substantially different substantially amounts, in part because they received different size donations and donations in different volumes.

Table 3. Outcomes by gender and race.

47 campaigns excluded due to having either multiple recipients of different genders, or unknown race.

Recipient race and gender	Campaigns n (%)	Amount raised ($)			# of Donations			Average donation size ($)
		Median	Mean (sd)	Range	Median	Mean (sd)	Range	Median	Mean (sd)	Range
White	664 (80.3%)	$134,261	$180,336 (177,780)	$100,020–2,078,890	1162	2123 (4448)	19-84,434	$120	$233 (540)	$16–6845
Men	364	$186,180	$182,831 (186,180)	$100,020–1,981,990	1006	2181 (5246)	19-84,434	$131	$221 (448)	$16–5543
Women	284	$134,514	$178,883 (171,701)	$100,170–2,078,890	1229.5	2055 (3330)	26-37,171	$109	$255 (651)	$24–6845
Black	26 (3.1%)	$139,504	$154,457 (72,595)	$100,876–454,923	1003	2240 (2357)	194- 8841	$121	$147 (146)	$27–691
Men	21	$137,492	$156,540 (79,586)	$100,876–454,923	1056	2550 (2541)	194- 8841	$101	$142 (160)	$27–691
Women	5	$148,318	$145,710 (33,904)	$101,640–182,441	823	1004 (512)	643-1881	$124	$167 (74)	$97–232
Asian (incl. South Asian)	50 (6.2%)	$150,834	$241,447 (391,734)	$101,878–2,696,160	1371	2285 (2814)	29-18,011	$87	$1304 (5658)	$35–30,638
Men	31	$150,293	$168,930 (107,450)	$101,878–682,508	1530	2615 (3321)	217-18,011	$78	$95 (95)	$38–470
Women	18	$147989	$365,752 (629,271)	$113074–2,696,160	1181	1545 (1422)	29-5556	$122	$3418 (9213)	$35–30,638
Other person of color	57 (6.9%)	$132,388	$195,638 (172,280)	$102,682–932,087	1636	3161 (5131)	86-26,307	$96	$191 (323)	$13–2055
Men	38	$131,460	$195,331 (173,141)	$102,682–932,087	1522	3507 (5864)	97-26,307	$97	$151 (191)	$13–1160
Women	19	$153,010	$196,254 (175,265)	$105,382–864,668	1662	2468 (3237)	86-14,550	$95	$271 (485)	$30–2055
Total	780

3.4. Discourse analysis on communicating deservingness

Given the substantial disparities observed in representation and campaign outcomes by race and gender, we conducted a targeted qualitative discourse analysis of campaigns for marginalized recipients to explore the extent to which racial and gender biases seemed to be shaping campaign narratives and audience engagement. In general, campaigns for recipients of color, and Black recipients in particular, typically used longer narratives, gave more frequent updates regarding the recipient’s health, and more extensively engaged with donors via comments. All campaigns for Black recipients contained language which heavily signaled to reasons why the recipient was deserving and worthy of assistance; while such language was present in some campaigns for white recipients, it was hardly universal, nor as extensively deployed. For example, in a campaign for a Black woman seeking funds to pay for cancer treatments, the recipient was repeatedly described as “lovely, sweet, and giving.” While the campaign was for an ill Black woman, the messages and updates on her health status repeatedly centered the needs, efforts, and sacrifices of her white husband. Many of the messages on the campaign expressed empathy for the white husband and the hardships he was taking on, with some explicitly stating they were donating because of the husband, while the wife’s illness appeared secondary to the husband’s deservingness of aid. Similarly, a campaign for a Black man with cancer frequently centered the needs of his white wife and their children.

Of the 26 campaigns for Black persons, 3 were for individuals with some level of notoriety or fame and included donations from other celebrities or notably affluent individuals. Other campaigns for Black recipients demonstrated their deservingness by describing professional and educational accomplishments and accolades, often extensively documenting exceptional characteristics and recognitions in text and photographs. For example, a campaign to provide care for an elderly Black man offered lengthy and repeated descriptions of him as a war hero and the “oldest living US veteran.” Slickly produced videos, photos of him receiving various honors, news coverage, and even support from GoFundMe helped spread the news of this campaign and repeatedly pointed to the recipient’s worth as due to his exceptional longevity and history as a veteran.

Finally, several other campaigns for Black recipients were started and run by white organizers, which may have contributed to their success, but also reflected and reinforced tropes of white saviorism. For example, a campaign run by a well-known white woman for a young Liberian boy emphasized the tragic circumstances of his situation and appealed to the sentiments of white American donors: “one look at this child ….and we all want to help. Americans are like that.” The campaign organizer’s narrative also reinforced racist tropes of child neglect and abandonment, writing, “His mother has never been in the picture and his father died 2 years ago from Ebola.”

4. Discussion

This study is the first to examine a targeted sample of highly successful medical crowdfunding campaigns. We find extensive evidence of racial, gender, and age disparities among these campaigns. Highly successful medical crowdfunding is most often represented by young, white men coping with unexpected medical crises like cancer or injuries. This indicates that vastly disproportionate financial support and donor attention goes to campaigns for whiter and younger recipients who are men. Black recipients, and black women in particular, were highly underrepresented, as were self-identified trans recipients, for whom no campaigns appeared in the sample. While other studies have documented racialized and gendered under-representation among medical crowdfunding campaigns, this study shows that among highly successful campaigns these dynamics are far more acute (Kenworthy et al., 2020; Saleh et al., 2020). For example, Kenworthy et al. (2020) found that about 3.4% of recipients in a randomized sample of medical crowdfunding campaigns were Black women; here we find that only 0.6% (n = 5) of highly successful campaigns are for Black women. Since success here is measured in terms of overall amounts raised by campaigns, we posit that two intertwining factors likely contribute to these disparities. First, campaigns for whiter, younger men seem to accrue disproportionate attention and social media spread, contributing to their success. And second, as Igra (2021) helps document, it appears that networked wealth inequalities, which occur along lines of race, class, and gender, contribute to inequitable campaign outcomes even when campaigns tap into roughly equivalent sized networks for support.

We also find evidence of racial and gender disparities in the levels of success that campaigns in this group achieve. Notably, though all campaigns had similar median measures for social media sharing, recipients from demographic groups such as white men raised the most money, but from larger donations in smaller numbers. By contrast, Asian and other POC men have far more donations but in substantially smaller amounts, indicating that they must work harder to achieve similar levels of success because donations to their campaigns are generally smaller. We hypothesize, as above, that the reasons for this reflect both interpersonal racism and biases as well as differences in networked wealth, which arise from structural racism and inequities. Campaign donor pools are often made up of family, friends, and other life acquaintances. In well-resourced networks with relatively affluent donors, campaigns are more likely to raise higher amounts quickly and easily. Networks that have more affluent donors also have access to highly privileged skills and knowledge that can contribute to highly successful campaigns, such as PR expertise, law, and finance experience in addition to other rich and influential friends as possible donors (Kenworthy, 2021). More research is needed to tease out the contributions of these individual factors towards campaign outcomes and success, though we note the significant contribution of Igra (2021) to one side of this discussion. Our research also points to the importance of intersectional analyses that explore multi-dimensional disparities between crowdfunders of different races, genders, and other dimensions of marginality.

As others have highlighted, it is important to pay attention to crowdfunding not just as individual appeals for assistance, but as broader sites of public engagement, visibility, and attention, such as the case for campaigns related to the Black Lives Matter (BLM) movement, which have drawn high levels of public engagement and donation (Kneese, 2018; Saleh et al., 2020). The data here reflects evidence of racial biases in dynamics contributing to crowdfunding engagement behaviors and outcomes among highly successful campaigns. These findings are further consistent with research on the role of implicit bias and generosity (Stepanikova et al., 2011). Interpersonal racism influences how people interact on digital platforms, and the kinds of scrutiny people face while online (Noble and Tynes, 2015; Vogels, 2021). We observed racism and sexism influencing campaign narratives and engagement in the form of racist and sexist tropes of deservingness and trustworthiness among recipients found in a targeted qualitative analysis of the sample. These campaign narratives and interactions can also fuel further racism and reinforce racist tropes as they spread across social media, demonstrating how crowdfunding campaigns can reinforce or amplify racist perceptions of deservingness. High-profile campaigns such as the “Gorilla Glue girl” campaign for Tessica Brown, a Black woman who went “viral” for cementing her hair with Gorilla Glue, gained rapid popularity and donations but also faced substantial racism from online audiences. Her accounts were frozen and the campaign was investigated for fraud after criticism and judgments from internet netizens over the campaign’s validity.

The saga was an example of how Black women and femmes can struggle to be believed and taken seriously, but also of how crowdfunding can subject users to additional harm and provide a platform for amplifying messages about racist tropes and differential deservingness. Media scholars have repeatedly documented how marginalized people, and black women in particular, face additional scrutiny online; and research on charity and social support has shown how these tend to reify racially-exclusive categories of individual deservingness that distract from systemic injustices (Benjamin, 2019; Noble and Tynes, 2015; Saleh et al., 2020; Snowden and Graaf, 2019). Further research is needed to more extensively study dynamics of bias and interpersonal racism in crowdfunding campaigns. There is also an acute need for research which more closely examines the impacts that such high-profile crowdfunding campaigns may have on public opinions, perceptions of deservingness, and racial discrimination.

Lastly, campaigns are impacted by corporate decisions and platform dynamics that can perpetuate structural racism and racial biases. Many algorithms—human creations with documented histories of racism due to how they are designed and the data they learn from—disproportionately harm communities of color, women, trans people, and other gender non-conforming people by restricting access and opportunity and increasing surveillance and law enforcement (Benjamin, 2019; Chung, 2022). Racism has also been well documented in the design, moderation choices, and experiences of social media platforms that crowdfunding relies on for spread, such as Facebook, Instagram, and Twitter (Dwoskin et al., 2021; Hao, 2020; Noble and Tynes, 2015). In the context of crowdfunding platforms, biases can manifest in the form of content moderation and directing website traffic to campaigns that are highly successful, as well as how GoFundMe leverages its own resources (public relations, social media posts, and employee expertise) to generate traffic for certain campaigns. Information on how these corporate decisions are made and how tools are designed is extremely difficult to access, and more research (and public access to corporate data) is needed to better understand these potential impacts.

Taken together, these factors demonstrate that medical crowdfunding is a digital technology that exacerbates the effects of fundamental social causes on health in several ways: First, in the way it amplifies and extends social biases; second, in the way it exacerbates inequities in access to material resources such as donations through networks; third, in the way that it influences broader public engagement around questions of deservingness and social marginalization; and fourth, in the way that technological decisions and features may both reinforce these dynamics and conceal them from public view. Crowdfunding, then, has the potential to not only influence what Link and Phelan (1995, p. 81) identified as the crucial dimension of “access to resources that help individuals” to avoid or weather ill health, but also impacts our collective perceptions of what resources individuals should, and do, have access to, and whose appeals for resources come to matter and be seen. We posit that these point to an important way that digital technologies relate to fundamental causes, by reinforcing, exacerbating, and normalizing the impacts of fundamental social causes on how people find social and financial support to navigate health crises.

4.1. Limitations

Much work remains to be done before a full understanding of the extent of racial and gender disparities in crowdfunding is established. Although the present results clearly demonstrate that race and gender are associated with significant disparities in outcomes among highly successful medical crowdfunding campaigns, this study has several limitations. While it was not the objective of this paper to gather a nationally-representative sample, further research using representative samples and experimental studies would complement the findings in this study. Additionally, race was determined by the research team’s perception of recipients’ race based on photographs used in the campaigns, using five categories to represent racial groups: white, Black, Asian (including South Asian), other person of color (POC), and unknown. We acknowledge that these categories are limited and may not be reflective of the ways actual individuals may identify racially. Campaign “success” was measured by several limited quantitative measures of campaign performance, which may not adequately measure all dimensions of success or virality. The study also lacks data on how individual donors make decisions regarding who to donate to, and the amount to donate; rather we see aggregated decisions as reflected in what is highly successful. Certain limitations of this study could be addressed in future research, such as experimental studies like those conducted on equity crowdfunding to demonstrate how crowdfunder race impacts donor behavior (Younkin and Kuppuswamy, 2018). Despite these limitations, the present study has enhanced our understanding of the relationship between crowdfunding outcomes and race and gender among a rarified group not previously studied.

5. Conclusion

Medical crowdfunding has impacted how Americans secure financial support while navigating acute and chronic health crises, yet the industry is still largely unregulated and there has yet to be a robust analysis of campaign outcomes and impacts within population groups. There have been calls for strengthening the oversight of medical crowdfunding platforms such as GoFundMe, in order to maximize the benefits and equity in crowdfunding for medical costs (Snyder et al., 2017; Young and Scheinberg, 2017). Despite steadily growing attention to racial and gender disparities in health, there has been limited attention towards racial/gender health disparities in digital media environments, or to the ways racial dynamics, as experienced through social media, impact population health. Medical crowdfunding platforms can function as a social conduit for reproducing race, gender, and class inequalities that have the potential to result in poorer health outcomes. Highly successful campaigns can also convey information about health and deservingness to large public audiences; there is an urgent need for policymakers and public health practitioners to consider how successful crowdfunding campaigns convey messages about health, deservingness, and access to support, and what impacts this has for health and social disparities. Through this paper we demonstrate how crowdfunding’s unequal, highly competitive digital environment creates severe inequities in representation, in access to resources for health, and in the visibility of different types of patients and their health needs. These findings contribute novel evidence about highly successful campaigns and their impacts to a growing body of scholarship illuminating disparities in medical crowdfunding.

Data availability

Data will be made available on request.