Examining Donor Preference for Charity Religious Affiliation

Abstract

In the United States, most charitable donations go to religiously affiliated organizations, yet the impact of a charity’s affiliation on donor behavior is currently unclear. To better understand this impact, this article uses a laboratory experiment to explore how a charity’s religious affiliation drives donor behavior. In the experiment, participants select one charity from a list of eight, with each charity varying in religious affiliation. Masked and unmasked sessions differ in the inclusion of religious affiliation from half the charities, with masked sessions omitting religious affiliation of the charities. This article finds that adding religious language decreases donation frequency and average donation amounts for Christian charities competing against other religious charities. This drop is primarily driven by participants that are politically liberal. Participants prefer charity religious affiliation to match their own religious identity; however, participant strength of religiosity is more predictive in charity choice than religious affiliation.

Keywords

religious affiliation charitable giving laboratory experiment religiously affiliated charity

What is the value in a charity being religiously affiliated? On the surface, a religious affiliation seems quite valuable for a charitable organization. Within the United States of America’s giving sector, 29% of charitable giving goes to religious organizations such as churches, mosques, and synagogues (Giving USA, 2021).¹ However, looking at direct religious giving does not encapsulate all religious giving. Expanding the definition to giving to all charities with a religious affiliation, the share of donation dollars jumps to nearly 75% (Giving USA, 2021).

In the United States, there are a plethora of religious charity options from which to choose. Scheitle (2010) reports nine different types of religiously affiliated charity organizations he refers to as “para-church organizations.”² Of all listed “para-church” religiously affiliated charities, 22% do not have a religious keyword identifier on their Form 990, annually filed with the Internal Revenue Service (IRS). This number jumps to 45% for the second-largest category, home to some of the largest charities in the United States, relief and development charities (Scheitle, 2010). While the level of religious involvement within affiliated organizations likely varies, it is also possible that the omission is done intentionally to appeal to a broader donor base while maintaining some religious affiliation to attract religious donor substitutes (Smith et al., 2008).

As such, a natural question arises as to whether religiously affiliated charities have a financial incentive to selectively display their religious affiliation. This extends to how religious belief motivates a donor to select a charity and the donation amount they give to that charity. In addition, it is unclear whether individuals lacking religious belief will identify a secularly affiliated charity as a worthy reason to donate in a manner similar to a religious person identifying with their donation to a religiously affiliated charity. Finally, it is unclear if these results will be dependent upon whether religious charities are religious majorities or minorities in the local culture.

In answering these questions, this article contributes to the broader understanding of organization affiliation appeals to donors through exploratory research. Current knowledge of the impact of charitable organization branding and affiliations is limited (Michel & Rieunier, 2012). A better understanding of religious affiliation impact is of interest to the public economics, public administration, and nonprofit management fields due to the high percentage of donation dollars going to religiously affiliated organizations. Simultaneously, there is a changing religious demographic in the United States, with an increasing share of nonaffiliated individuals and a decreasing share of “moderately religious” individuals (Schnabel & Bock, 2017).

Literature Review

The charitable giving literature does not have many studies focusing on the religious affiliation of charities themselves, as research on the impact of charitable organizational branding and affiliations is currently limited (Michel & Rieunier, 2012). However, there is evidence that religiously affiliated individuals are more likely to donate to religiously affiliated charities than nonreligiously affiliated individuals (Chapman et al., 2018). Furthermore, some evidence exists that organizational religious affiliation, or lack thereof, can inspire pro-social behavior (Chen et al., 2017).

In contrast, there is a vast amount of literature studying how religiously affiliated individuals donate. Bekkers and Wiepking (2011), in a two-part literature review of 550 publications on donor characteristics, consistently find religiosity as a positive indicator for higher charitable donations.³ In addition, the National Center for Charitable Statistics (NCCS, 2017) reports that donations to religious organizations are at least double those of donations to secular organizations, in terms of both average donation and percent of income.⁴ Some of this difference in donation patterns among religious and secular individuals is driven by a social expectation of giving in Protestant churches (Bekkers & Schuyt, 2008).

Due to the spiritual incentives and communal nature of religious giving, it has been argued that religious giving should be examined in a different context than secular giving (Hrung, 2004). Giving toward an organization affiliated specific religious creed that an individual adheres to generally has spiritual incentives within the religion (McCleary, 2007). The community aspect of this creed can create a club good setting for religious participation (Iannaccone, 1992), which can also foster pro-social behaviors such as giving (Warner et al., 2015). The difference in giving rates between religious and secular individuals persists into the Millennial generation; however, the rate of giving difference between secular and religious individuals has decreased to roughly a 2% likelihood of donation (Koczanski & Rosen, 2019).

The observed difference between religious and secular giving in field data is not generally observed in experimental settings where religious and secular individuals give at the same frequency and amount. Eckel and Grossman (2004), using a variety of national charities with different causes, found no statistical difference in giving when responding to a subsidy. In addition, religious affiliation has been found to have no effect on contribution levels in a public goods game (Anderson & Mellor, 2009) or in a bilateral trust game (Anderson et al., 2010).

The difference in laboratory and field data may stem from differing levels of religious salience in the donors in a laboratory setting versus a field setting. Brenner (2011a, 2011b, 2011c) details the importance of religious salience and its impact on religious Americans’ behavior in survey questions where survey respondents overreport their religiously moral behaviors to convey the importance of their religious identity. Benjamin et al. (2016) provide some evidence for this hypothesis, as they found Protestants to give more in public goods games after a religious prime, increasing the salience of the participant’s religious identity.

However, Benjamin et al. (2016) also find that religious primes have no effect on giving in the context of a dictator game. Combined, these lab findings imply that differences in religious and secular giving may not stem from differences in beliefs or signaling morality but instead may be due to mechanisms such as a difference in opportunities to give. This theory was echoed by Bottan and Perez-Truglia (2015) when they saw donations drop from formally Catholic individuals who left the church and by Bekkers and Schuyt (2008) studying religious giving in the Netherlands. This explanation is also consistent with Wang and Graddy’s (2008) findings that donors with a more bridging social network and civic engagement donate more to religious and secular causes.

Experiment Design

If a financial incentive exists to selectively display religious affiliation, charities with a religious affiliation may find it in their interest to distance themselves from their religious affiliation. For example, in 2009, the Christian Child Fund distanced themselves from their religious affiliation by changing their name to Child Fund International (Banks, 2009). Due to the nature of changing religious affiliation for an entire organization, it is not likely that many exogenous changes from the charity perspective exist in the field. Therefore, this article utilizes a laboratory experiment.

An outline of the overall experimental procedure is included in Figure 1. The experiment is conducted as follows: students are recruited via the Florida State University XS/FS online system ORSEE, and all sessions take place in the Florida State University XS/FS lab (Greiner, 2004) using the Z-tree computer software (Fischbacher, 2007). Each participant is compensated with a US$7.00 show-up fee, with the potential to earn more in Phase 2. The experiment then goes into the three-phased interactive program as indicated by Figure 1.

Figure 1.

Experimental design.

As in Brown et al. (2017), Phase 1 consists of a charity selection stage. Participants are asked to select one charity for a potential donation from a list of eight.⁵ Participants are told that they do not have to donate anything to the charity; however, they must select one, and only one. Each charity is focused on international poverty aid or disaster relief. This type of organization was selected for two reasons. The second-largest group of religiously affiliated charities is relief and development charities. Furthermore, relief and development charities also happen to have the highest rate of not reporting a religious affiliation. All charities used are based in locations outside of Florida to avoid any location effects (Eckel et al., 2018). All charities have vague descriptions of international disaster or poverty relief efforts. These descriptions are constructed from materials that the organizations have authored. Participants were informed that all charity ratings are similar based on Better Business Bureau (or the U.K. government equivalent in one case) and Charity Navigator ratings when applicable.⁶ The order of placement for charities is randomized for each session.

Based on the experiment’s location in the United States, the experiment uses Christian charities, secular charities, and Islamic charities. Specifically, Islamic charities are included for several reasons. First, the effects of adding religious language to a charity description may be different for majority and minority religious groups. Second, Islamic-affiliated charities provided a religious “out-group” based on running the experiment on a college campus in the southeastern United States. Finally, the experiment design required a religion with a sufficient number of international aid charities operating within the United States.

The charities available to participants, as well as their descriptions, vary depending upon the experimental treatment. As the main goal of the experiment is to derive the value of a charity’s religious affiliation, one experimental treatment used is the addition of religious affiliation information into a mission statement. However, a study on how donors react to changes to a religious affiliation requires nuance in terms of potential competition among charities for the same donation dollars. The most obvious realm of competition is between religious charities and secular charities for a specific pool of donors, such as those individuals who give to international relief charities. The article defines this competition as an extrareligious competition:

Definition 1: Extrareligious Competition: A competition for donors between religious charities and secular charities.

The effect of looking at religious charities as a whole is an incomplete analysis as it assumes that the value of a religious affiliation in a charitable organization is homogeneous across religions. This is likely not the case. The value of a religious affiliation for the majority religion in a region is plausibly different than the value of a religious affiliation for a minority religion. Thus, there is an element of competition for donors between religious organizations, which the article defines as intrareligious competition:

Definition 2: Intrareligious Competition: A competition for donors within the subsector of religiously affiliated charities, among religious charities affiliated with differing faiths.

Each type of competition for donation dollars, as well as the base idea of altering how a charity advertises its religious affiliation, factors into the experiment design through charity choice. Table 1 lists each charity used in extrareligious and intrareligious competition, as well as their masking status in masked and unmasked sessions. In all sessions, the eight charity options consist of four charities with Christian affiliation and four of either Islamic or secular affiliation. Masked sessions omit the religious affiliation information of half the charities, specifically two Christian charities and two either Islamic or secular charities depending upon the realm of competition. Doing so effectively makes the charity descriptions read as “neutral” charities, making it impossible to tell what their religious affiliation is. These “neutral” charities define the difference between masked and unmasked sessions. Simulating the change in the way a charity highlights their religious affiliation, unmasked sessions provide religious affiliation information for all charities, including those previously masked.⁷

Table 1.

Experiment Design.

	Charity name	Intrareligious competition		Extrareligious competition
	Charity name	Masked sessions	Unmasked sessions	Masked sessions	Unmasked sessions
Christian 1	World Vision	Unmasked	Unmasked	Unmasked	Unmasked
Christian 2	World Hope International	Unmasked	Unmasked	Unmasked	Unmasked
Christian 3	Opportunity International	Masked	Unmasked	Masked	Unmasked
Christian 4	World Concern	Masked	Unmasked	Masked	Unmasked
Islamic 1	Islamic Aid	Unmasked	Unmasked
Islamic 2	Islamic Relief USA	Unmasked	Unmasked
Islamic 3	Life for Relief & Development	Masked	Unmasked
Islamic 4	Mercy-USA	Masked	Unmasked
Secular 1	S.H.A.R.E. through C.F.I.			Unmasked	Unmasked
Secular 2	Direct Relief International			Unmasked	Unmasked
Secular 3	Oxfam			Masked	Unmasked
Secular 4	Mercy Corps			Masked	Unmasked

Note. Placement of each charity is randomly determined before each session.

The same charities are masked in every masked session within both intrareligious competition and extrareligious competition. Thus, over masked and unmasked sessions, the summary of information content is varied as opposed to the charities themselves. Additional variation comes in the form of religious competition. The intrareligious sessions show how specific religious in-grouping motivates charitable giving, whereas the extrareligious sessions show how religious affiliation motivates giving versus secular affiliation. Thus, the experiment has a 2 × 2 design [half charities masked, fully identified charities] × [intrareligious competition, extrareligious competition]. Specific differences between intrareligious and extrareligious competition are both explained below.

The intrareligious competition features four Christian and four Islamic charities. These specific religions were selected to have a clear majority and minority religious group for the region where the experiment was conducted. Each religious group received a balancing of the emphasis on religious affiliation within the charity based on the descriptions on their websites and fundraising materials. This was done intentionally to match the varying levels of emphasis on religious affiliation used by all charities in fundraising materials. A similar process was followed for extrareligious competition.

Of the two realms of competition, extrareligious competition is what most people probably first think of in terms of the value of a charity’s religious affiliation. As the majority religion in the United States is Christianity, all four Christian charities from intrareligious competition treatment are kept as the religious charities. The four Islamic charities are substituted for four secular charities. The differences between masked and unmasked sessions are identical in extrareligious competition to those in intrareligious competition. Extrareligious competition only changes the charities used in the experiment. For the secular charities, descriptions were constructed to match the same level of variability in emphasis on lack of an affiliation as the emphasis placed on religious affiliation for the religious charities used in this experiment.

In Phase 2, participants can earn money through a real effort task. The real-effort task is necessary to counteract the “windfall effect,” where participants donate more often and more of their money in experiments with endowed charitable donation allowances (Reinstein & Riener, 2012). This is only important in the context of the analysis if the effect threatens to systematically disrupt the distribution of charity selection and donation dollars given to the charities. Such a risk does exist. In addition to finding differing rates of donation with and without a real effort task, Reinstein and Riener (2012) find differing rates of donation between men and women. In addition, men and women have differing rates of religious affiliation in the United States, particularly when it comes to Christianity (Pew Research Center, 2016). Internationally, Christians are more likely to be women than men; conversely, nonaffiliated individuals are more likely to be men than women (Pew Research Center, 2016). While interactions between these effects are not explored, and gender is controlled in the regression analysis, to mitigate threats to external validity further, a real-effort task is included to better simulate the charitable giving process.

Using the counting grid from Abeler et al. (2011), participants have 10 min to count as many 10 × 10 matrices of 1s and 0s as they possibly can. Each correct grid earns the participant completing it US$2.50. There is a US$10 earnings cap placed on the participants, set at a low enough level to ensure that all participants should hit the earnings cap. Establishing the earnings cap allows for donation amounts to be comparable across participants, avoiding income effects. Over 99% of participants achieved maximum earnings in the actual experiment, and those who did not complete four tables were dropped from the analysis.

Prior to the 10-min earning session, participants complete a practice grid to make sure they understand the concept of the task. After the task is finished, participants can donate as much or as little of their earnings as they wish to the charity they selected. They are not allowed to donate their show-up fee. To ensure that participants believe that their donations will go to the charity that they select, they are explicitly informed that the experimenter will write a check to these charities for each of them. At the time of payment, the experiment shows the participants their donation check and places it in a corresponding envelope, addressed to the selected charity with appropriate postage. Participants are allowed to accompany the experimenter to the mailbox on Florida State’s campus to verify that the checks are mailed if they wish.

Phase 3 consists of a demographic survey, used to elicit participant religious background and other controlling factors for regression analysis. Importantly, participants are asked their familiarity with any of the charities listed.⁸ To control for other identifying factors that could cause participants to donate, it is important to know participant familiarity with the charities. If a participant has a prior history with any of the charities listed, the role their religious identity plays in the charity choice and donation decisions is likely compromised. The survey questions are constructed with questions and responses consistent with racial, gender, and religious belief categories used in the General Social Survey and the Baylor Religion Survey. Finally, to obtain a measure of religiosity in addition to affiliation, the survey includes all questions from the Duke University Religion Index, an index ranging from 5 to 27 used to measure the intensity of the participant’s religious belief (Koenig & Bussing, 2010). The questions used in the Duke University Religion Index are consistent with Brenner’s (2011b) findings that questions on religious activities such as attendance can be used as a proxy for determining the importance of religiosity to a participant’s identity.

Experiment Data and Baseline Analysis

The experiment consisted of 16 sessions with four masked and four unmasked sessions for extrareligious and intrareligious competition treatments, respectively. Sessions had between nine and 26 participants, resulting in 164 observations for intrareligious competition and 157 observations for extrareligious competition. Three participants failed to reach their maximum and maximum earnings and were dropped to allow for data analysis without having to control for income effects. Those observations were not included in the total count of observations. Participants earned an average of US$16.51, including their show-up fee. Sessions lasted 65 min on average.

Demographic balance tables across sessions are reported in Table 2. The dashed line separates the three religious variables used to measure participant religious identity, including indicator variables for secular and Christian affiliated individuals, as well as the composite score of the Duke University Religion Index (Koenig & Bussing, 2010) from other demographic control variables used in the experiment. The demographic controls include participant age as well as indicators for female participants, participants who are economic majors, Black participants, Latino participants, participants from states outside of Florida, participants who indicated that they had either previous experience of either donating or volunteering for a charity listed in the experiment, or had previously been exposed to or heard of charities listed in the experiment, and participants who answered “slightly conservative,” “conservative,” or “extremely conservative” when asked about their political beliefs. Sessions are mostly balanced along these variables, with the exception of participants in extrareligious unmasked sessions having an 11.8 percentage point higher level of prior exposure compared with masked sessions, and a 13.4 percentage point higher rate of identifying as secular. Conversely, intrareligious sessions only differ with a difference in participant age of 0.853 years between masked and unmasked sessions. Finally, there is a 19-percentage point difference in the differences at the means of previous exposure across all sessions.

Table 2.

Demographic Control Balance Tables.

	Intrareligious competition				Extrareligious competition				Diff-in-diff
	Pooled	Masked	Unmasked	Difference	Pooled	Masked	Unmasked	Difference	(Extra − intra)
Secular	0.268(0.444)	0.230(0.423)	0.312(0.466)	0.082(0.069)	0.293(0.457)	0.233(0.425)	0.366(0.485)	0.134^‡ (0.073)	0.052(0.100)
Christian	0.530(0.501)	0.540(0.501)	0.519(0.503)	−0.021(0.079)	0.452(0.499)	0.453(0.501)	0.451(0.501)	−0.003(0.080)	0.018(0.112)
Strength of religiosity	13.409(6.073)	13.701(6.075)	13.078(6.093)	−0.623(0.952)	12.764(5.850)	12.988(5.779)	12.439(5.964)	−0.495(0.940)	0.128(1.339)
Female	0.646(0.480)	0.678(0.470)	0.610(0.491)	−0.068(0.075)	0.585(0.494)	0.581(0.496)	0.591(0.495)	0.010(0.079)	0.078(0.109)
Age	19.664(3.006)	19.264(1.505)	20.117(4.052)	0.853^‡ (0.467)	20.299(4.510)	20.384(4.875)	20.197(4.055)	−0.187(0.725)	−1.039(0.855)
Economics major	0.061(0.240)	0.069(0.255)	0.052(0.223)	−0.018(0.038)	0.051(0.221)	0.035(0.185)	0.070(0.258)	0.036(0.035)	0.052(0.053)
Conservative	0.177(0.383)	0.195(0.255)	0.156(0.365)	−0.040(0.060)	0.261(0.441)	0.244(0.432)	0.282(0.453)	0.038(0.071)	0.077(0.093)
Black	0.165(0.372)	0.161(0.370)	0.169(0.377)	0.008(0.058)	0.153(0.361)	0.128(0.336)	0.183(0.390)	0.005(0.058)	0.047(0.082)
Latino	0.244(0.431)	0.218(0.416)	0.273(0.448)	0.054(0.067)	0.172(0.379)	0.174(0.382)	0.169(0.377)	−0.005(0.061)	−0.060(0.091)
Not from Florida	0.226(0.434)	0.253(0.444)	0.195(0.426)	−0.058(0.068)	0.204(0.379)	0.209(0.360)	0.197(0.401)	−0.012(0.061)	0.046(0.092)
Previous experience	0.250(0.434)	0.264(0.444)	0.234(0.426)	−0.031(0.068)	0.172(0.379)	0.151(0.360)	0.197(0.401)	0.046(0.061)	0.077(0.092)
Previous exposure	0.116(0.321)	0.149(0.359)	0.078(0.270)	−0.072(0.050)	0.146(0.355)	0.093(0.292)	0.211(0.411)	0.118* (0.056)	0.190*** (0.075)
N	164	87	77	164	157	86	71	157	321

Note. Standard deviation in parentheses in Columns 2 to 4, and 6 to 8. Standard error in parentheses in Columns 5, 9, and 10.

‡

p < .10. *p < .05. **p < .01. ***p < .001.

Donation behavior across charity types in extrareligious and intrareligious competition is plotted in Figure 2 for changes in the frequency of donations, and in Figure 3 for changes in the average donation received conditional upon having donated to that type of charity. Each plot contains 90% confidence intervals, robust to heteroskedasticity. Starting with the donation frequency changes in Figure 2, the only group of charities experiencing a statistical change in their donation behaviors were masked Christian charities in intrareligious competition, seeing donation frequencies decrease by 11.0 percentage points. The change in conditional average donations plotted in Figure 3 shows that unmasking decreases average donations to previously masked Christian charities in intrareligious competition by US$1.667. Conversely, unmasking changes donations in previously masked secular charities saw an increase in average donation, conditional upon donating to a secular marked charity, of approximately US$1.408 after unmasking. Putting these results into context, as participant earnings were US$10 in the experiment, these results indicate a decrease in conditional average donation by 16.67% of subject earnings and an increase in conditional average donation by 14.08% of subject earnings, respectively.

Figure 2.

Mean changes in donation frequency with unmasking.

Figure 3.

Mean changes in conditional average donations with unmasking.

To determine how donation behavior changes across the four treatments, this article estimates a regression model with the following form:

y_{i} = α_{i} + β {u n m a s k e d}_{i} + δ {e x t r a}_{i} + θ {(u n m a s k e d \times e x t r a)}_{i} + u_{i},

where y_i represents donation frequency (measured as a linear probability model) and amount, unmasked_i is an indicator for unmasked sessions, extra indicates extrareligious competition sessions, (unmasked × extra)_i is an interaction term of each indicator, and u_{i ~}(0,V) represents idiosyncratic errors robust to heteroskedasticity. Analysis is run over donations to all charities, as well as a narrowing of analysis to donations to Christian charities, and finally donations to Christian charities masked in the experiment. These restrictions of y_i are due to the fact that Christian charities are the only charities used in both extrareligious competition and intrareligious competition, and masked Christian charities are the only charities subject to unmasking used in both extrareligious competition and intrareligious competition.

The results from the regression are reported in Table 3. Donation frequency and amounts do not change across any of the four treatments when analyzing donations to charity as a whole or donations specifically to Christian charities. However, donation frequency does decrease by 11.0 percentage points for Christian charities in unmasked sessions. In addition, unmasking decreased unconditional average donations to previously masked Christian charities by US$0.341, or 3.41% of subject income. As shown in the previous figures, the majority of this effect is driven by changes in donor behavior in intrareligious competition.

Table 3.

Differences in Donor Behavior Across Specific Charities.

	All charities		All Christian		Christian unmasked
	Donated	Donation	Donated	Donation	Donated	Donation
Unmasked	−0.097(0.077)	−0.218(0.268)	−0.038(0.071)	−0.031(0.214)	−0.110* (0.044)	−0.341* (0.142)
Extrareligious competition	−0.004(0.073)	−0.182(0.270)	−0.101(0.067)	−0.125(0.232)	−0.033(0.052)	−0.099(0.190)
(Extrareligious Competition × Unmasked)	0.164(0.107)	0.675(0.414)	−0.003(0.095)	−0.045(0.319)	0.065(0.064)	0.250(0.219)
N	321	321	321	321	321	321

Note. Standard errors robust to heteroskedasticity in parentheses.

‡

p < .10. *p < .05. **p < .01. ***p < .001.

Hurdle Model Analysis

Empirical Strategy

In addition to the baseline analysis in the preceding paragraphs, the data collected in the experiment allow for regression analysis to determine donor preferences for charity religious affiliation over a variety of religious demographic factors, including strength of religiosity and affiliation of the donor. The regression model utilizes a hurdle model featuring a Probit selection model and a Tobit model censored at US$0.00 to measure the donation amount. The choice to censor the Tobit model at only US$0.00 is driven by the experimental data. Donation amount frequencies are provided in Table 4. Roughly 35.51% of participants chose not to donate any of their income to charity, compared with only 1.25% of participants who chose to donate the entirety of their earnings.

Table 4.

Donation Frequency by Amount.

Donation	Frequency	Percent of donations
0.00	114	35.51
0.10	1	0.31
0.25	1	0.31
0.40	1	0.31
0.50	6	1.87
0.75	1	0.31
1.00	82	25.55
1.25	1	0.31
1.50	5	1.56
1.80	1	0.31
2.00	57	17.76
2.25	1	0.31
2.34	1	0.31
2.50	3	0.93
3.00	12	3.74
3.40	1	0.31
3.50	2	0.62
4.00	6	1.87
5.00	14	4.36
6.30	1	0.31
7.00	4	1.25
8.00	2	0.62
10.00	4	1.25
N	321	100

The choice to use a hurdle model as opposed to separate Probit and Tobit models stems from the fact that the standard errors of the Probit selection model likely help identify the decision on how much to give to philanthropy. For analyses of specific types of charitable giving with insufficient donation observations for the hurdle model to converge, separate Probit and Tobit models of the same form as described below are utilized.⁹ Unlike the preceding analysis, the hurdle model analysis does not pool extrareligious and intrareligious competition data due to the use of a Probit selection equation. The marginal effects of interaction terms are not able to be computed, rendering it impossible to determine differences in behavior across intrareligious competition and extrareligious competition (Ai & Norton, 2003).

The selection model equation takes the following form:

P ({D o n a t e d}_{i} = 1 | Z_{i}) = Φ (α_{i} + {R e l}_{i} φ + δ {D u k e}_{i} + γ {U n m k}_{i} + X_{i} β),

where Donated_i is a binary variable indicating that the charity received a donation from individual i, Z_i represents the total control vector with all the independent variables of interest and controls, Rel_i indicates religious affiliation as reported, Duke_i is the composite Duke University Religion Index score measuring the strength of religiosity, Unmk_i is a binary indicating those in unmasked sessions, and X_i represents control demographic variables such as major, age, gender, political beliefs, charity experience and exposure, whether the participant is from outside of Florida, and racial background. Doing the above analysis answers how religious identity drives donations in terms of the donation decision. Thus, to see how the change in information alters donations, the analysis above is repeated with the sample restricted to individuals selecting specific types of charities. In intrareligious competition, these include selection of Christian or Islamic charities, as well as an examination specifically on the charities affected by unmasking. The same procedure is followed for extrareligious competition with restrictions to either religious or secular charities, as well as further restrictive analysis on the charities altered specifically by unmasking. The results of the selection equation demonstrate how the probability of donation changes with a change in charity religious affiliation information.

From here, the Tobit model takes the following form:

{D o n a t i o n}_{i} = α_{i} + {R e l}_{i} φ + δ {D u k e}_{i} + γ {U n m k}_{i} + X_{i} β + u_{i},

where Donation_i stands for donation amount, all other dependent variables being the same as above, and finally u_i~ N(0,V) represents the idiosyncratic errors. As in the first stage of the selection equation, analysis over the specific affiliations is rerun for both intrareligious and extrareligious competition, along with an analysis directly on those charities altered by unmasking. The results of the Tobit model demonstrate how the average donation amount changes with a change in the charity’s religious affiliation information.

Hurdle Model Results

Results for the hurdle model analyzing intrareligious competition are reported in Table 5 and Table 6, and results for Extrareligious Competition are reported in Table 7 and Table 8. In all cases, the dashed line separates the estimates of the unmasking impact on donations within the specific type of competition from a variety of religious demographic controls of interest. Each subset of charities examined is analyzed twice, once only with religious demographic controls and once with the full set of demographic controls reported in the balance table in Table 2.

Table 5.

Probit Selection Equation Marginal Effects, Intrareligious Competition.

	All charities		All Christian		Christian unmasked		All Islamic		Islamic unmasked
Unmasked	−0.094(0.078)	−0.098(0.082)	−0.036(0.072)	−0.045(0.073)	−0.114** (0.043)	−0.145** (0.048)	−0.053(0.073)	−0.075(0.073)	−0.050(0.070)	−0.065(0.069)
Strength of religiosity	0.022* (0.009)	0.032** (0.010)	0.013(0.008)	0.015^‡ (0.009)	−0.006(0.005)	−0.006(0.006)	0.006(0.008)	0.010(0.009)	0.002(0.008)	0.008(0.008)
Christian	0.031(0.102)	0.055(0.155)	0.201* (0.305)	0.178^‡ (0.104)	0.075(0.055)	0.078(0.068)	−0.151(0.099)	−0.160(0.103)	−0.082(0.097)	−0.070(0.100)
Secular	0.065(0.121)	0.126(0.118)	0.157(0.144)	0.163(0.140)	0.067(0.089)	0.063(0.091)	−0.074(0.109)	−0.076(0.111)	−0.037(0.329)	−0.025(0.110)
Full demographic controls		X		X		X		X		X
N	164	164	164	164	164	136†	164	164	164	164

Note. Standard errors robust to heteroskedasticity in parentheses. Controls include those for gender, ethnicity, political beliefs, charity exposure and experience, economics majors, out of state subjects, and age. Regressions with N denoted by † represent models where certain observations are dropped as they predict failure with certainty, and the hurdle model does not converge.

‡

p < .10. *p < .05. **p < .01. ***p < .001.

Table 6.

Tobit Model For Donors, Intrareligious Competition.

	All charities		All Christian		Christian unmasked		All Islamic		Islamic unmasked
Unmasked	0.009(1.015)	−0.002(0.881)	0.322(2.090)	−0.285(0.670)	−3.865* (1.776)	−3.720** (1.384)	−0.411(1.110)	0.076(1.023)	−0.260(1.289)	0.192(1.194)
Strength of religiosity	−0.058(0.101)	−0.086(0.089)	−0.240(0.299)	0.090(0.081)	−0.220(0.139)	−0.199(0.140)	0.071(0.132)	0.029(0.117)	0.057(0.147)	0.003(0.130)
Christian	0.199(1.314)	0.790(1.119)	1.010(2.769)	1.750^‡ (1.040)	2.161(1.605)	1.862(1.593)	0.432(1.249)	1.056(1.329)	0.348(1.566)	1.002(1.722)
Secular	−0.873(1.896)	−0.864(1.519)	−1.073(4.152)	1.161(1.296)	1.639(2.234)	1.281(1.868)	0.166(1.776)	0.016(1.780)	0.012(2.040)	−0.216(2.135)
Full demographic controls		X		X		X		X		X
N	100	100	48	164†	164†	164†	52	52	45	45

Notes. Standard errors robust to heteroskedasticity in parentheses. Controls include those for gender, ethnicity, political beliefs, charity exposure and experience, economics majors, out of state subjects, and age. Regressions with N denoted by “†” are Tobit regressions used because the number of subjects donating to this charity type are too small for the hurdle model to converge.

‡

p < .10. *p < .05. **p < .01. ***p < .001.

Table 7.

Probit Selection Equation Marginal Effects, Extrareligious Competition.

	All charities		All Christian		Christian unmasked		All secular		Secular unmasked
Unmasked	0.095(0.075)	0.082(0.080)	−0.039(0.249)	0.013(0.068)	−0.039(0.040)	−0.016(0.042)	0.126(0.083)	0.073(0.094)	0.007(0.069)	−0.009(0.067)
Strength of religiosity	−0.003(0.008)	0.003(0.008)	0.017** (0.006)	0.029*** (0.008)	0.005(0.005)	0.008^‡ (0.005)	−0.027** (0.009)	−0.032** (0.010)	0.001(0.007)	0.001(0.007)
Christian	−0.036(0.103)	−0.031(0.107)	0.035(0.068)	−0.058(0.093)	−0.009(0.049)	−0.066(0.060)	−0.081(0.107)	−0.005(0.120)	−0.170* (0.081)	−0.145(0.076)
Secular	−0.230* (0.116)	−0.191(0.126)	−0.123^‡ (0.074)	−0.178** (0.081)	−0.079^‡ (0.048)	−0.119** (0.047)	−0.187(0.116)	−0.123(0.133)	−0.109(0.084)	−0.067(0.080)
Full demographic controls		X		X		X		X		X
N	157	157	157	157	157	126†	157	157	157	157

Notes. Standard errors robust to heteroskedasticity in parentheses. Controls include those for gender, ethnicity, political beliefs, charity exposure and experience, economics majors, out of state subjects, and age. Regressions with N denoted by “†” represent models where certain observations are dropped as they predict failure with certainty, and the hurdle model does not converge.

‡

p < .10. *p < .05. **p < .01. ***p < .001.

Table 8.

Tobit Model for Donors, Extrareligious Competition.

	All charities		All Christian		Christian unmasked		All secular		Secular unmasked
Unmasked	1.824(2.268)	2.293(1.844)	1.137(2.669)	3.081(2.080)	4.762(5.696)	−0.509(1.319)	1.718(1.760)	1.292(1.191)	3.894(2.804)	1.039(1.066)
Strength of religiosity	0.615(0.437)	0.578^‡ (0.320)	0.751(0.512)	0.641* (0.270)	−0.133(0.266)	0.231^‡ (0.121)	0.229(0.225)	0.295(0.197)	0.347(0.339)	0.474** (0.181)
Christian	−3.375(3.458)	−2.997(2.737)	1.552(3.146)	0.122(2.471)	−2.346(2.676)	−2.023(1.565)	−3.024(2.956)	−3.236(2.467)	−3.212(4.109)	−2.408(1.922)
Secular	6.287(4.630)	4.227(2.908)	22.629* (9.674)	16.113*** (3.965)	8.743*** (1.333)	−3.446^‡ (1.980)	1.397(1.446)	1.095(1.223)	4.201(2.162)	3.436** (1.086)
Full demographic controls		X		X		X		X		X
N	107	107	30	30	15	157†	77	77	36	36

‡

p < .10. *p < .05. **p < .01. ***p < .001.

Intrareligious competition

Beginning with the results of the Probit selection equation reported in Table 5, the analysis finds that unmasking does not affect donation rates when looking at all charities. The only subgroup reporting decreases in donation frequency were unmasked Christian charities, seeing a drop in donation rates by 14.5 percentage points. In addition, the results reported in Table 5 demonstrate the importance of strength of religiosity in donation preference rather than donation affiliation alone. Individuals with a higher overall strength of religiosity are 3.2 percentage points more likely to donate, corresponding to an additional point at the mean on their Duke University Religion Index Score. Furthermore, individuals with a higher strength of religiosity are more likely to donate to a Christian charity, increasing the likelihood of donation by 1.5 percentage points for an additional point at the mean on a respondent’s Duke University Religion Index Score. Affiliation of the donor does not affect the likelihood of donating overall, but there are some differences across specific subgroups of charities. Christian-affiliated individuals are more likely to donate to Christian charities overall at a rate of 17.8 percentage points higher than non-Christian participants.

Turning to the Tobit model measuring changes in conditional average donation amounts reported in Table 6, unmasking does not change conditional average donations received in aggregate across all charities. Analysis over all subgroups shows that unmasking again only impacts Christian charities who were previously masked. The number of donations to this subgroup of charities was too small for the hurdle model to converge; however, analysis using a Tobit model over all observations shows a decrease in the average donation received by US$3.720. In addition, the results reported in Table 6 show that differences in conditional average donations received are not driven by either the strength of religiosity or religious affiliation, with the exception of Christian-affiliated individuals donating to Christian charities. In donations to Christian charities, Christian-affiliated individuals donate US$1.750 more on average than non-Christian donors.

Extrareligious competition

Starting with the results of the Probit selection equation reported in Table 7, the analysis shows that unmasking does not change donation behavior across all charities in extrareligious competition, nor among any specific subset of charities analyzed. In addition, the analysis finds that religious and secular individuals do not donate at different rates of donation frequency when all demographic controls are included. Furthermore, the strength of religiosity is not a predictor of donor likelihood. However, Table 7 demonstrates that donors sort by their religious preference. Donors with higher Duke scores are less likely to donate to secular charities and are more likely to donate to religious ones. In addition, secular individuals are 17.8 percentage points less likely to donate to any Christian charity and 11.9 percentage points less likely to donate to an unmasked Christian charity.

Turning to the Tobit model for donation amounts in Table 8, the analysis finds that unmasking does not statistically change the amount of an average donation received. This implies that earlier results indicating increases in conditional average donations to secular unmasked charities are largely explainable by the religious demographics of participants. The significant predictor for donation amounts in both religious and secular unmasked charities, when the hurdle model converges, appears to be respondents who are secular. Finally, the analysis finds that there is no significant difference between the giving patterns of Christian or secular individuals regarding conditional donation amounts. However, there is a relationship between a higher religiosity score and an average donation, with an increase of a point on the Duke University religion index at the mean corresponding to an increase in donations by US$0.578. This positive correlation extends to all Christian charities, including analysis of the unmasked charities, as well as secular unmasked charities. However, there is no relationship between average donations to secular charities as a whole and the strength of religiosity.

Political Affiliation and Participant Behavior

While the regression analysis controls for political affiliation and the political composition of the participant pool is not significantly different across treatments, it is possible that participants with differing political leanings behave differently when presented with religious affiliation information in a charity description. In the United States, a relationship between fundamentalist Christianity and conservative political leanings has existed from at least the 1980s onward with the “Moral Majority” and has roots tracing back to the 1960s (Durham, 2000; Johnson et al., 1986; Phillips-Fein, 2011). The relationship between American Christianity and American conservatism persists today, in behaviors and beliefs such as voting patterns (Baker et al., 2020), opposition to same-sex couple adoption (Whitehead & Perry, 2016), and disregarding COVID-19 precautions (Perry et al., 2020). The combined literature indicates that conservative-leaning participants may react differently to the inclusion of religious language than liberal-leaning participants, particularly when it comes to the inclusion of Christian affiliation.

To explore donation patterns by political affiliation, participants are separated by their political based on whether they self-identify as liberal. This grouping is used rather than the conservative binary used as a control in the regression analysis and balance table to have a more evenly distributed number of participants.¹⁰ Plots of the raw data from the experiment are provided in Figure 4.

Figure 4.

Mean changes in donation frequency among liberal participants.

Beginning with analysis on donation frequency among self-identified liberal participants, the raw data indicate a large decrease in donations to Christian charities subject to unmasking in both intrareligious competition and extrareligious competition, decreasing donations by 21.2 percentage points in intrareligious competition and decreasing donations by 11.4 percentage points in extrareligious competition. No other type of organization saw a statistically significant change in donation frequency among self-identified liberal participants.

Figure 5.

Mean changes in conditional average donations among liberal participants.

Due to a lack of participants donating to certain types of charity organizations, it is impossible to calculate changes in average donations conditional upon donating to the same organizational type as done analyzing the full data set. Instead, plots demonstrate the change in average donation received, conditional upon the participant donating at all. While this will bias average donations toward zero, it will still provide a measure of how adding religious affiliation information impacts donation amounts for self-identifying liberal and nonliberal participants. Among self-identifying liberal participants, the data indicate that the process of unmasking charities results in decreased average donations to Christian charities subject to unmasking, with a decreased average donation of US$0.901 in intrareligious competition and a decreased average donation of US$0.281 in extrareligious competition. These changes represent a decrease in average donation corresponding to 9.01% of participant income and 2.81% of participant income, respectively.

Figure 6.

Mean changes in donation frequency in nonliberal participants.

With regard to the analysis on donation frequency among self-identified nonliberal participants, the raw data indicate changes to secular charities not subject to unmasking in extrareligious competition, with an increased donation likelihood by 18.1 percentage points. In addition, Islamic charities subject to unmasking saw a decreased donation likelihood of 17.9 percentage points. While the data are insufficient to run regressions controlling for religious affiliation, it is likely that the increase in giving to secular charities among nonliberal participants is driven by a greater number of secular participants participating in unmasked sessions.

Figure 7.

Mean changes in conditional average donations among nonliberal participants.

Finally, in examining average donations among self-identified nonliberal participants, the raw data show that average donations received only change for Christian charities not subject to unmasking in intrareligious competition. These organizations saw an increased average donation of US$0.841, corresponding to 8.41% of the subject income.

Discussion

This article finds that adding religious language results in a reduced selection of previously masked Christian charities by an 11.0 percentage point likelihood when competing exclusively against Islamic organizations. In addition, adding religious language reduces average donations by US$1.667, or approximately 16.67% of participant income, for Christian charities competing for donations against Islamic charities. Much of this drop in donations come from self-identifying liberal participants, as their donations to Christian charities subject to unmasking decrease in both extrareligious and intrareligious competition. These results are exploratory in nature, providing some of the first building blocks in understanding the impact of charity and religious affiliation on donor behavior.

In terms of overall donation patterns, the article’s findings are consistent with other experimental papers finding no difference between religious and secular individuals in donation frequency or conditional average donation when controlling for demographic factors. These results provide supporting evidence to the theory that the differences in giving between secular and religious individuals stem from differences in opportunities to give. In addition, the article’s results provide corroborating evidence to Eckel and Grossman (2004), Anderson and Mellor (2009), and Anderson et al.’s (2010) results indicating that nonreligious individuals were as likely to make a charitable donation compared with religiously affiliated ones.

Building off their research, this article finds that a connection between the religious identity of an individual and their preference for religious affiliation of a charity does exist; however, it seems to be driven by the strength of an individual’s religiosity rather than affiliation. This result is consistent with the religious salience theories put forward by Brenner (2011a, 2011b, 2011c) and is also consistent with donor preferences found in Chapman et al. (2018). This article finds that individuals who have a higher overall strength of religiosity donate at a higher frequency in intrareligious competition and in larger amounts in extrareligious competition.

This article finds differences in participant behavior by separating participants based on self-identified political affiliation. Participants self-identifying as liberal donate less to Christian charities, subject to unmasking, in terms of donation frequency and average donations received, in both intrareligious competition and extrareligious competition. Furthermore, self-identified nonliberal participants decrease donation frequency to Islamic charities with the inclusion of religious identifying language. These results build on a literature demonstrating a relationship between American conservatism and American Christianity and provide evidence that politically conservative and liberal donors behave differently in response to the inclusion of religious language in a charity description.

Combined, the article’s exploratory results indicate that a financial incentive to selectively display religious affiliation does exist for charities. Religious charities must consider their target audience’s political affiliation as well as the religious affiliation or lack thereof associated with their closest competitors. Financial incentives could explain why 45% of religious relief service organizations, as well as 22% of all religious charity organizations, choose not to report at least one religious keyword in their description to the IRS on their Form 990 (Scheitle, 2010).

However, these results are subject to the participant pool generating the data. External validity concerns exist due to the age and demographic composition of the participants. Being college students at Florida State University, these participants are likely young in their religion and charitable giving life cycles. Furthermore, the participant pool consists mainly of 18- to 22-year-old participants that are either Christians or religiously nonaffiliated. The participants are also nearly all from the United States, with between 75% and 80% of the participants from the state of Florida. In addition, a small minority of the participants identify as politically conservative, ranging between 15% and 30% of the participant pool for a given session. Finally, due to the nature of using college students, the sample by construction excludes individuals with only a high school level of education or less. Changing the composition of participants could potentially change the results, which is left to be explored in future research.

Due to the nature of the experiment, it is impossible to determine if religious charities have an incentive to increase their religious affiliation ties for a specific religious audience to potentially substitute for church giving, as no outside church giving option is available for participants to select. In addition, it cannot be determined if a charity’s religious affiliation is a primary driver in the donation decision or a secondary driver. Based on the construction of the charity descriptions and the donation patterns of each participant, the data from the experiment suggest that donors may react differently to a religious affiliation in intrareligious and extrareligious competition. This changes the primary factor of the donation decision and necessarily the financial incentive for religious affiliation emphasis. The experiment instead focuses on appealing to donors in an overall general context, as if they were selecting from a menu of charities to potentially donate to in a workplace charity drive. These questions could be answered in a “laboratory in the field,” which would also address the external validity concerns present due to the participant pool, which is left for future research.

Conclusion

This article demonstrates that increasing information on religious affiliation results in decreasing charitable donations for Christian charities, both in terms of frequency as well as the total dollars received within the donation dollar distribution when competing in intrareligious competition. The decrease in donations is driven by self-identified liberal participants in the sample. This article demonstrates that religiosity strength does predict preference for charity religious affiliation among experiment participants.

The article indicates that Christian charities in the United States would be better served to not acknowledge their religious affiliation when fundraising from a politically liberal audience. Still, maintaining enough of a religious affiliation to keep their church networks available may be in a charity’s financial interest, as indicated in Bottan and Perez-Truglia (2015). The maintenance and selective display of affiliation mirrors the unmasking process of religious affiliation described in the discussion section of the article and is consistent with 45% of religious relief service agencies not reporting a religious connection (Scheitle, 2010).

The outcomes of the experiment, as well as the questions raised by the experiment’s shortcomings, have real-world implications for charities trying to maximize their appeal to donors. The article tells a story consistent with current charity behavior and provokes interesting questions for the future. Ideally, future research will take the experiment to a “laboratory in the field” setting to see if the results hold with a different participant pool more representative of charity donors. In addition, a field setting would provide the opportunity to further test if individuals are shifting their donations from direct religious giving into more indirect measures such as international relief charities. Finally, a field setting would provide the opportunity to investigate the driving decisions in charity selection in reference to charity emphasis on religious affiliation.

Supplemental Material

sj-docx-1-nvs-10.1177_08997640221105656 – Supplemental material for Examining Donor Preference for Charity Religious Affiliation

Supplemental material, sj-docx-1-nvs-10.1177_08997640221105656 for Examining Donor Preference for Charity Religious Affiliation by Jonathan Oxley in Nonprofit and Voluntary Sector Quarterly

Footnotes

Acknowledgements

I would like to thank David Cooper for sharing his experimental software. In addition, I extend my deepest gratitude to the Quinn Graduate Student Fellowship and the Institute for Humane Studies for helping fund this project and making the paper possible. I would like to thank the editors at Nonprofit and Voluntary Sector Quarterly and the three anonymous referees for their helpful comments. Finally, I would like to thank R. Mark Isaac, Carl Kitchens, Luke Rodgers, Jens Grosser, Jeremy P. Thornton, Erik Kimbrough participants in the Florida State University Experimental Readings Group, Southern Economic Association, Science of Philanthropy Initiative, Association for Research on Nonprofit Organizations and Voluntary Action, the Institute for the Study of Religion, Economics and Society, and the Association for the Study of Religion, Economics and Culture for their helpful advice and assistance.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received funding from the Quinn Graduate Student Fellowship at Florida State University and a Doctoral Fellowship from the Institute for Humane Studies for support in pursuing their dissertation. The author used these funds to pay for the sessions in this experiment.

ORCID iD

Jonathan Oxley

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biography

Jonathan Oxley is a visiting assistant professor in the economics department at Trinity University. He primarily studies topics within behavioral and public economics relating to nonprofit organization structure, donor behavior, and religion, through quantitative methods using applied microeconomic and experimental techniques.

References

Abeler

Falk

Goette

Huffman

(2011). Reference points and effort provision. The American Economic Review, 101(2), 470–492.

Norton

E. C.

(2003). Interaction terms in logit and probit models. Economics Letters, 80(1), 123–129.

Anderson

L. R.

Mellor

J. M.

(2009). Religion and cooperation in a public goods experiment. Economics Letters, 105(1), 58–60.

Anderson

L. R.

Mellor

J. M.

Milyo

(2010). Did the devil make them do it? The effects of religion in public goods and trust games. Kyklos, 63(2), 163–175.

Baker

J. O.

Perry

S. L.

Whitehead

A. L.

(2020). Keep America Christian (and White): Christian nationalism, fear of ethnoracial outsiders, and intention to vote for Donald Trump in the 2020 presidential election. Sociology of Religion, 81(3), 272–293.

Banks

A. M.

(2009, May 15). Christian children’s fund drops “Christian” from name. Christianity Today.

Bekkers

Schuyt

(2008). And who is your neighbor? Explaining denominational differences in charitable giving and volunteering in the Netherlands. Review of Religious Research, 50(1), 74–96.

Bekkers

Wiepking

(2011). Who gives? A literature review of predictors of charitable giving part one: Religion, education, age and socialisation. Voluntary Sector Review, 2(3), 337–365.

Benjamin

D. J.

Choi

J. J.

Fisher

(2016). Religious identity and economic behavior. Review of Economics and Statistics, 98(4), 617–637.

10.

Bottan

N. L.

Perez-Truglia

(2015). Losing my religion: The effects of religious scandals on religious participation and charitable giving. Journal of Public Economics, 129, 106–119.

11.

Brenner

P. S.

(2011a). Exceptional behavior or exceptional identity? Overreporting of church attendance in the US. Public Opinion Quarterly, 75(1), 19–41.

12.

Brenner

P. S.

(2011b). Identity importance and the overreporting of religious service attendance: Multiple imputation of religious attendance using the American time use study and the General Social Survey. Journal for the Scientific Study of Religion, 50(1), 103–115.

13.

Brenner

P. S.

(2011c). Investigating the biasing effect of identity in self-reports of socially desirable behavior. Sociological Focus, 44(1), 55–75.

14.

Brown

A. L.

Meer

Williams

J. F.

(2017). Social distance and quality ratings in charity choice. Journal of Behavioral and Experimental Economics, 66, 9–15.

15.

Chapman

C. M.

Louis

W. R.

Masser

B. M.

(2018). Identifying (our) donors: Toward a social psychological understanding of charity selection in Australia. Psychology & Marketing, 35(12), 980–989.

16.

Chen

Liu

Mei

(2017). Does team competition increase pro-social lending? Evidence from online microfinance. Games and Economic Behavior, 101, 311–333.

17.

Durham

(2000). The Christian right, the far right and the boundaries of American conservatism. Manchester University Press.

18.

Eckel

C. C.

Grossman

P. J.

(2004). Giving to secular causes by the religious and nonreligious: An experimental test of the responsiveness of giving to subsidies. Nonprofit and Voluntary Sector Quarterly, 33(2), 271–289.

19.

Eckel

C. C.

Priday

B. A.

Wilson

R. K.

(2018). Charity begins at home: A lab-in-the-field experiment on charitable giving. Games, 9(4), 95.

20.

Fischbacher

(2007). Z-tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10(2), 171–178.

21.

Giving USA. (2021). Giving USA 2021: The annual report on philanthropy for the tear 2020. The Giving Institute, Indiana University–Purdue University Indianapolis.

22.

Greiner

(2004). An online recruitment system for economic experiments. In Kremer

Macho

(Eds.), GWDG Bericht: Vol. 63. Forschung und wissenschaftliches Rechnen 2003 (pp. 79–93). Gesellschaft für wissenschaftliche Datenverarbeitung.

23.

Gruber

(2004). Pay or pray? The impact of charitable subsidies on religious attendance. Journal of Public Economics, 88(12), 2635–2655.

24.

Hoge

D. R.

(Ed.). (1996). Money matters: Personal giving in American churches. Westminster John Knox Press.

25.

Hrung

W. B.

(2004). After-life consumption and charitable giving. American Journal of Economics and Sociology, 63(3), 731–745.

26.

Iannaccone

L. R.

(1992). Sacrifice and stigma: Reducing free-riding in cults, communes, and other collectives. Journal of Political Economy, 100(2), 271–291.

27.

Johnson

S. D.

Tamney

J. B.

Halebsky

(1986). Christianity, social traditionalism and economic conservatism. Sociological Focus, 19(3), 299–314.

28.

Koczanski

Rosen

H. S.

(2019). Are millennials really particularly selfish? Preliminary evidence from a cross-sectional sample in the philanthropy panel study. American Behavioral Scientist, 63(14), 1965–1982.

29.

Koenig

H. G.

Bussing

(2010). The Duke University Religion Index (DUREL): A five-item measure for use in epidemiological studies. Religions, 1(1), 78–85.

30.

McCleary

R. M.

(2007). Salvation, damnation, and economic incentives. Journal of Contemporary Religion, 22(1), 49–74.

31.

Michel

Rieunier

(2012). Nonprofit brand image and typicality influences on charitable giving. Journal of Business Research, 65(5), 701–707.

32.

National Center for Charitable Statistics. (2017). Charitable giving in America: Some facts and figures. National Center for Charitable Statistics.

33.

Perry

S. L.

Whitehead

A. L.

Grubbs

J. B.

(2020). Culture wars and COVID-19 conduct: Christian nationalism, religiosity, and Americans’ behavior during the coronavirus pandemic. Journal for the Scientific Study of Religion, 59(3), 405–416.

34.

Pew Research Center. (2016). The gender gap in religion around the world. Pew Research Center.

35.

Phillips-Fein

(2011). Conservatism: A state of the field. The Journal of American History, 98(3), 723–743.

36.

Reinstein

Riener

(2012). Decomposing desert and tangibility effects in a charitable giving experiment. Experimental Economics, 15(1), 229–240.

37.

Reitsma

Scheepers

Grotenhuis

M. T.

(2006). Dimensions of individual religiosity and charity: Cross-national effect differences in European countries? Review of Religious Research, 47(4), 347–362.

38.

Scheitle

C. P.

(2010). Beyond the congregation: The world of Christian nonprofits. Oxford University Press.

39.

Schnabel

Bock

(2017). The persistent and exceptional intensity of American religion: A response to recent research. Sociological Science, 4, 686–700.

40.

Smith

Emerson

M. O.

Snell

(2008). Passing the plate: Why American Christians don’t give away more money. Oxford University Press.

41.

Wang

Graddy

(2008). Social capital, volunteering, and charitable giving. Voluntas: International Journal of Voluntary and Nonprofit Organizations, 19(1), 23–42.

42.

Warner

C. M.

Kılınc

Hale

C. W.

Cohen

A. B.

Johnson

K. A.

(2015). Religion and public goods provision: Experimental and interview evidence from Catholicism and Islam in Europe. Comparative Politics, 47(2), 189–209.

43.

Whitehead

A. L.

Perry

S. L.

(2016). Religion and support for adoption by same-sex couples: The relative effects of religious tradition, practices, and beliefs. Journal of Family Issues, 37(6), 789–813.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.22 MB