Dynamics of International Giving: How Heuristics Shape Individual Donor Preferences

Key Drivers of Individual-Level Philanthropy

Substantial research on charitable giving has looked at the motivations of individual donors (Bekkers & Wiepking, 2011; Wiepking, 2010). However, research on individual donor behavior has overwhelmingly examined giving to organizations working locally, rather than internationally. This literature has primarily focused on the importance of three main factors, especially when considering international philanthropy: (a) the role of social and associational capital, (b) the role of individual experiences such as higher education and levels of religiosity, and (c) third-party certifications.

Prior research has found that those who participate in a variety of associations and build social capital are more likely to make charitable donations as well as make larger donations (Hossain & Lamb, 2017). An individual’s social network also plays an important role in deciding to donate. Looking at relational configurations in social networks, Herzog and Yang (2018) find that having both a giver and a solicitor in one’s social network increased the likelihood of that individual’s charitable giving.

Individual experiences such as access to higher education and increased religiosity can also influence giving (Wiepking & Bekkers, 2010). Higher education fosters pro-social motivations and brings people into social networks that entail a higher level of solicitation (Bekkers & Wiepking, 2011). Education also increases abstract thinking (Wiepking & Maas, 2009), which is important for donating to distant international relief organizations than donating to local, more visible nonprofits (Micklewright & Schnepf, 2009). Overall, individuals with higher levels of income, education, and greater religious proclivities have been shown not only to be more likely to give internationally but also to give higher amounts (Casale & Baumann, 2015; Rajan et al., 2009).

Finally, information about organizational characteristics and third-party certifications can also affect levels of individual philanthropy, though results are mixed. Existing research shows that information on the efficiency of NGOs merely steers rather than encourages or discourages overall donations (Ryazanov & Christenfeld, 2018). The effects of changes in charity watchdog ratings and nonprofit accountability systems on individual donors are also mixed, showing that donors do not have the time to research charities thoroughly (Bekkers, 2010) and that ratings by charity watchdogs do not really affect donor support for these nonprofits (Szper & Prakash, 2011).

Even though these above factors affect individuals’ motivations to donate internationally, extant literature on international giving focuses on donating during crises and emergencies. However, as Micklewright and Schnepf (2009) point out, motivations for donating to ongoing international causes differ from donating money to natural disasters or emergency international causes. Under ordinary circumstances, “donors are more likely to donate to a charity operating locally than to a charity providing identical service abroad” (Tremblay-Boire & Prakash, 2017, p. 644). However, we lack adequate data and theory regarding international philanthropy. Most research on individual donors to international causes has been restricted to elite, high net worth donors—those with more than US$200,000 in annual income or US$1 million in assets (US Trust, 2014). However, given the increasing funds channeled by individuals to international nonprofits, it is crucial to understand what motivates these donors to donate.

Individuals rely on heuristics that differ from the benchmarks used by large governmental agencies and private foundations. These signals, we argue, can be even more important in giving to international causes. In theory, donors should scrutinize each aspect of an NGO’s organizational structure and programmatic performance prior to donation, but this rarely happens in practice (Tremblay-Boire & Prakash, 2017). Instead, donors respond to a host of heuristics when deciding whether to donate to an NGO, as seeking complete information about an organization is costly and time-consuming. These heuristics are commonly rooted in donor perceptions of organizational characteristics—donors make cursory judgments about an organization’s issue area, mission, vision, and values and seek out supplementary information from friends, family, and acquaintances (Sloan, 2009; Szper & Prakash, 2011). Nonprofits fundraising internationally may therefore rely on framing to control perceptions. Below, we theorize the effect of different frames surrounding a country’s legal environment and nonprofit organizational characteristics on individual donor preferences.

Sample

We test these hypotheses with a vignette-based survey experiment fielded through Amazon’s Mechanical Turk. ⁶ Our target population is the portion of Americans hypothetically willing to donate money for human rights and humanitarian work abroad. Our convenience sample (March 2018, N = 531) generally approximates the characteristics of our target population, as it is younger, more educated, wealthier, and more likely to donate to charities than nationally representative samples. In addition, the majority of the sample (≈90%) feels favorably toward human rights, humanitarian, and development NGOs.

Experimental Treatments and Outcomes

We presented participants with a short paragraph with three manipulated frames, each highlighting different factors that might influence patterns of private philanthropy: (a) crackdown/no crackdown, (b) humanitarian assistance/human rights, and (c) government/private funding. We used the International Rescue Committee (IRC) as our example NGO because it fits within each possible frame and, as such, requires no respondent deception. IRC provides humanitarian assistance and engages in human rights advocacy for refugees, more than a quarter of IRC’s income comes from both government grants and private donations, and it works in countries with and without anti-NGO laws. ⁷ We use a 2 × 2 × 2 between-subject factorial design with participants randomly assigned to one of eight versions of the following vignette:

The International Rescue Committee (IRC) focuses on {humanitarian assistance for refugees | human rights for refugees} {and works in countries that have recently passed laws that harshly restrict nonprofit organizations | NOTHING}. A substantial proportion of IRC’s funding comes from {government | private} donors.

We measured two outcomes: (a) how likely participants would be to donate to IRC (measured with a 5-point scale ranging from extremely likely to extremely unlikely) and (b) how much participants would hypothetically donate to IRC if they had an extra US$100. Research finds that the factors driving donation willingness and amounts differ depending on donor income, pro-social attitudes, self-image, and other psychological benefits (Wiepking, 2007). Accordingly, our framing treatments might differ across the two outcomes. We collapsed the likelihood scale into a binary variable measuring whether the participant is likely (extremely likely and somewhat likely) or not likely (all other responses) to donate. ⁸ We also included an exploratory free response question asking respondents to justify their choice.

Each of the varied treatments correspond to our hypotheses: crackdown versus no crackdown, humanitarian assistance versus human rights, and government versus nongovernmental funding. Participants might misinterpret the vignette and assume that IRC’s host governments fund the organization rather than its home government. In either case, knowing about government funding should still serve as a signal of the organization’s deservingness. Similarly, there could be confusion with phrase “private funding,” but as we use it as the opposite of government funding, it should still act as a heuristic. Free responses tend to confirm the correct interpretation of the vignette, and many respondents explicitly justified their lack of support for IRC because of government funding, as expected (“I see that a large portion of its funding comes from government donors, so I feel it doesn’t need me as much” ⁹ ).

Estimation

We test our first hypothesis by calculating the differences in the average likelihood to donate and the average amount donated across the crackdown conditions. For our second and third hypotheses, we measure the difference in means spread across both issue and funding conditions. For additional exploration of the effect of issue and funding, we also measure the effect of crackdown within nested combinations of issue and funding conditions.

We use two Bayesian models to estimate the effect of crackdown on the likelihood of donating and the amount hypothetically donated. We model the proportion of respondents indicating they would likely donate as a binomial distribution:

n group 1 , group 2 ∼ Binomial ( n group total , θ group ) [ likelihood ] Δ n = n group 2 − n group 1 [ difference in proportions ] n : Number likely to donate θ group 1 , group 2 ∼ Beta ( 5 , 5 ) [ prior prob . of likelihood ]

We estimate the mean amount donated in each condition using a t distribution:

x group 1 , group 2 ∼ Student t ( ν , μ , σ ) [ likelihood ] Δ x = x group 2 − x group 1 [ difference in means ] x : Mean amount donated ν ∼ Exponential ( 1 / 29 ) [ prior normality ] μ group 1 , group 2 ∼ N ( x ¯ group 1 , group 2 , 10 ) [ prior donation mean per group ] σ group 1 , group 2 ∼ Cauchy ( 0 , 1 ) [ prior donation SD per group ]

We use median values from the posterior distributions as point estimates and calculate credible intervals using the 95% highest posterior density. We declare an effect statistically significant if the posterior probability of being different from zero exceeds 0.95. Because differences can be either positive or negative (i.e., in some frames, the crackdown condition causes larger donations; in others, it causes smaller donations), we report the probability that the difference is not equal to zero: When the median value is negative, we report the proportion of predicted values that are negative, and vice versa.

Likelihood of Donation

In isolation we find that respondents are not more likely to donate to legally besieged INGOs (H_1a). As seen in Figure 2a, those exposed to the crackdown condition tend to have a slightly higher probability of donating (46.7%) than those in the control condition (42.9%), but the difference is not significant ( Δ = 3 . 8 % ; p ( Δ ≠ 0 ) = 0 . 82 ) . We find similar results when looking at the isolated effects of issue area and funding. Both differences are positive, but the posterior probability of being larger than zero is low ( Δ issue = 2 . 3 % ; p ( Δ issue ≠ 0 ) = 0 . 7 ; Δ funding = 2 . 9 % ; p ( Δ funding ≠ 0 ) = 0 . 75 ) . We thus do not initially find evidence that crackdowns (or issue or funding) alone increase the likelihood of donating (Table 2).

OPEN IN VIEWER

Figure 2.

Difference in likelihood of donation across crackdown and no crackdown groups, conditioned by other experimental frames.

OPEN IN VIEWER

Table 2.

Likelihood of Donation and Differences in Proportions in “Crackdown” (Treatment) and “No Crackdown” (Control) Conditions; Values Represent Posterior Medians.

H1a	% likelyTreatment	% likelyControl	Δ Δ ∆	%∆	p ( Δ ≠ 0 )
Crackdown—no crackdown	46.7%	42.9%	3.8%	8.9%	.82
Humanitarian assistance—human rights	45.9%	43.7%	2.3%	5.4%	.7
Private—government funding	46.3%	43.4%	2.9%	6.8%	.75
H2a and H3a	% likelyCrackdown	% likelyNo crackdown	∆	%∆	p ( Δ ≠ 0 )
Human rights issues	44.3%	43.6%	0.6%	1.5%	.54
Humanitarian assistance issues	49.4%	42.5%	6.8%	15.9%	.88
Government funding	41.3%	45.8%	−4.6%	−10.0%	.78
Private funding	52.4%	40.3%	12.1%	30.2%	.98
H2a and H3a (nested)	% likelyCrackdown	% likelyNo crackdown	∆	%∆	p ( Δ ≠ 0 )
Human rights issues, government funding	32.0%	47.4%	−15.4%	−32.5%	.97
Human rights issues, private funding	57.4%	40.4%	16.9%	41.5%	.98
Humanitarian assistance issues, government funding	51.4%	44.7%	6.8%	15.2%	.8
Humanitarian assistance issues, private funding	47.5%	41.2%	6.1%	14.8%	.78

Figure 2b and c shows the difference in the likelihood of donation for both crackdown conditions across NGO issue areas (H_2a) and funding sources (H_3a). Crackdown has almost no effect on the likelihood of donating to human rights NGOs ( p ( Δ ≠ 0 ) = 0 . 54 ) , which is not surprising, as people who donate to these kinds of NGOs likely know that their work is challenging to host governments. We hypothesized that the positive crackdown effect would be stronger for humanitarian NGOs, and although crackdown does have a positive effect under the humanitarian frame, the probability this difference is greater than zero is below our 95% threshold ( p ( Δ ≠ 0 ) = 0 . 88 ) .

We do find some evidence for H_3a, however. Crackdown has little effect on individual preferences to donate to NGOs receiving government funding ( p ( Δ ≠ 0 ) = 0 . 78 ) , but it does increase the likelihood of donating to privately funded NGOs by 30% ( Δ = 12 . 1 % ; p ( Δ ≠ 0 ) = 0 . 98 ) . This is likely because a donor might feel that their marginal donation would make a difference. For instance, one respondent reasoned that as IRC “is already receiving funding from governments, so funding from private individuals doesn’t seem as necessary. I’d prefer to give my money to an organization that is primarily run by private donations.” ¹⁰

Although the crackdown frame has no effect on the probability of donating to human rights NGOs alone, conditioning this finding on the source of NGO funding reveals competing trends. Figure 2d shows the difference in donation likelihood across all experimental conditions. When donors know that a human rights organization is primarily government-funded, they are substantially less likely to donate if the organization faces government crackdown abroad. This follows our expected hypotheses for both issue area and funding ( Δ = − 15 . 4 % ; p ( Δ ≠ 0 ) = 0 . 97 ) . This finding possibly reflects donor suspicion about the mission and goals of government-funded human rights programming. For instance, one respondent in the government and human rights conditions explained that they did not donate because “government donors might have interests that actually hurt refugees, and are donating to have some sway in the organization.” ¹¹

In contrast, when donors know that a human rights organization is privately funded, they are more likely to donate when there is a legal crackdown than when the NGO faces no legal hurdles ( Δ = 16 . 9 % ; p ( Δ ≠ 0 ) = 0 . 98 ) . As explained previously, each of the heuristics on their own influence the propensity to donate, but we do not yet know how the interaction between heuristics affects donor behavior. Here, the private funding heuristic appears to override the dampening in likelihood that should come from focusing on more contentious human rights issues. For humanitarian assistance NGOs, however, legal crackdowns do not have a significant effect on the likelihood of donation when nesting either funding or issue frames—neither of the nested experimental frames play a heuristic role.

These results give partial support to H_3a. The source of funding has little influence on the preference to donate to humanitarian NGOs, but on average, donors are substantially more likely to want to give to privately funded human rights NGOs than government-funded organizations. In addition, donors appear to both punish government-funded human rights NGOs facing crackdown and rally behind privately funded NGOs facing those same hurdles. For instance, one respondent who was presented with a government-funded human rights version of IRC explained that “I’m not exactly sure what they did was right or wrong, I think it just seems bad, so I’m less likely to want to donate to them.” ¹² Other respondents showed increased support when presented with a privately funded version of IRC, explaining that they would donate a substantial amount because “[t]hey [IRC] are doing good work in countries where it is tough for groups like them to operate and they need all the help they can get.” ¹³ Individual donors thus seem to be more willing to support besieged human rights organizations when they are unencumbered by government funds.

Amount Donated

On their own, crackdowns do not substantially influence donors’ likelihood to donate, but they do increase the amount of money that respondents are willing to contribute (see Figure 3a). Informing participants that IRC faces legal hurdles abroad increased donations by US$3.39 on average, a 26% increase from the no crackdown condition ( p ( Δ ≠ 0 ) = 0 . 97 ) . This provides good evidence for H_1b—donors appear to give more money to besieged INGOs. In contrast, neither issue area nor funding on their own have an effect on the amount donated (Table 3) ( Δ issue = − 0 . 78 ; p ( Δ issue ≠ 0 ) = 0 . 66 ; .

OPEN IN VIEWER

Figure 3.

Difference in amount donated across crackdown and no crackdown groups, conditioned by other experimental frames.

OPEN IN VIEWER

Table 3.

Mean Values and Differences in Means for Amount Donated in “Crackdown” (Treatment) and “No Crackdown” (Control) Conditions; Values Represent Posterior Medians.

H1b	AmountTreatment	AmountControl	∆	%∆	p ( Δ ≠ 0 )
Crackdown—no crackdown	16.35	12.96	3.39	26.3%	.97
Humanitarian assistance—human rights	14.01	14.84	−0.78	−5.3%	.66
Private—government funding	15.11	13.65	1.44	10.6%	.78
H2b and H3b	AmountCrackdown	AmountNo crackdown	∆	%∆	p ( Δ ≠ 0 )
Human rights issues	17.43	14.94	2.49	16.8%	.83
Humanitarian assistance issues	16.01	11.74	4.27	36.5%	.96
Government funding	13.87	12.26	1.64	13.5%	.74
Private funding	18.97	14.33	4.55	31.6%	.96
H2b and H3b (nested)	AmountCrackdown	AmountNo crackdown	∆	%∆	p ( Δ ≠ 0 )
Human rights issues, government funding	10.55	15.04	−4.49	−29.9%	.9
Human rights issues, private funding	23.71	14.5	9.09	62.7%	.99
Humanitarian assistance issues, government funding	21.42	11.97	9.18	76.4%	.99
Humanitarian assistance issues, private funding	15.73	15.79	−0.03	−0.2%	.5

This trend also holds when nesting crackdown within NGO issue area. The crackdown condition elicits higher donations for both the human rights and humanitarian NGOs, though with varying levels of significance (see Figure 3b and c). Emphasizing legal crackdown increases donations to human rights NGOs by US$2.49, but with a lower probability of significance ( p ( Δ ≠ 0 ) = 0 . 83 ) . In contrast, crackdowns boost donations to humanitarian NGOs by US$4.27, a 37% increase beyond the control condition ( p ( Δ ≠ 0 ) = 0 . 96 ) . We thus find support for H_2b: Donors give more to besieged humanitarian NGOs. We also find evidence for H_3b. Donors give US$4.55 more (a 32% increase) to legally besieged NGOs that are privately funded, as hypothesized ( p ( Δ ≠ 0 ) = 0 . 96 ) . However, the crackdown condition has little effect on the amount donated to government-funded NGOs ( p ( Δ ≠ 0 ) = 0 . 74 ) . This is possibly because the effects cancel each other out—legal crackdowns should increase the amount donated, but reliance on government funding dissuades donors from giving, as predicted by the literature showing that governments can crowd out private funding.

Combining issue areas and funding sources provides more texture. We previously found that donors are less likely to donate to government-funded human rights NGOs and more likely to donate to privately funded human rights NGOs when they face legal issues. These trends also apply to preferences in the amount donors donate. Respondents were willing to donate US$9.09 more to besieged privately funded human rights NGOs compared with the no crackdown control condition ( p ( Δ ≠ 0 ) = 0 . 99 ) , increasing their donations by 63%. Respondents explained that they chose to give additional money to privately funded human rights NGOs precisely because “the country limits non-profits.” ¹⁴ The punishment mechanism found previously does not translate to the amount donated—although there is a decrease of US$4.49 in average donations to government-funded human rights NGOs facing crackdown, the chance this difference is less than zero is lower than our threshold ( p ( Δ ≠ 0 ) = 0 . 9 ) .

Emphasizing legal difficulties increased donations to government-funded humanitarian NGOs by 76%, or US$9.18 ( p ( Δ ≠ 0 ) = 0 . 99 ) . The crackdown condition causes this increase despite some hesitation from respondents who often indicated that government-funded NGOs do not deserve private donations. Many respondents in the crackdown condition argued that “[m]y tax paying dollars go to the government and if the government is helping to fund it then they are in turn getting some of my money.” ¹⁵ This hesitation is not universal, however, and in aggregate, most respondents chose to give substantially more. Many echoed this respondent, who identified the crackdown condition as one of the reasons for giving the full hypothetical US$100: “If they can function without being affected by the harsh laws against non-profits, I would want them to have as much help as possible to do their humanitarian work.” ¹⁶ Thus, legal crackdowns can increase individual donor urgency to support an organization, and this heuristic is potentially processed differently compared with other organizational-level heuristics or information from social networks.

In contrast, crackdowns have no substantial effect on the amount donated to privately funded humanitarian NGOs ( p ( Δ ≠ 0 ) = 0 . 5 ) . No respondent in these conditions mentioned funding sources in their justification, and many explained that they would not give solely because of the crackdown and that they “would keep all $100 because it would not even go to helping humanitarian efforts due to the new laws.” ¹⁷ It is thus possible that donors see government-funded humanitarian NGOs as more legitimate and more capable of handling difficult legal restrictions than privately funded organizations. Donors might assume that an NGO that receives government funding would have support from the home government when facing legal challenges, whereas an NGO funded by small donors would be less able to confront such challenges. Further research is needed to probe this trend.