By Jonathan Oxley, Trinity University.
Most charitable donations in the United States either go to a religious organization or a religiously affiliated charity. Giving to religious organizations such as churches, mosques, and synagogues represents the largest share of donations received in the United States, with Giving USA (2021) reporting that 29% of all charitable giving goes to religious organizations. However, this only represents a fraction of religious giving in the United States. Giving USA (2021) estimates that including donations to religiously affiliated charities as religious giving brings this figure closer to 75% of all giving in the United States. This expanded definition includes donations to large, well-known, highly-rated charities such as World Vision, Habitat for Humanity, Lifewater International, World Hope International, and Opportunity International.
Despite receiving nearly 45% of donation dollars in the United States, Scheitle (2010) finds that 22% of all religiously affiliated, non-church charities do not include a religious keyword identifier on their Form 990. This number increases to 45% for the second-largest category of religiously affiliated, non-church charities, relief and development organizations. My paper “Examining Donor Preference for Charity Religious Affiliation,” uses a laboratory experiment to explore how donor behavior changes with the inclusion of religious language in a charity’s description, with the goal of determining if religiously affiliated, non-church charities have a financial incentive to selectively display their religious affiliation. I find such incentives exist.
In my experiment, subjects select one charity from a list of eight before earning a maximum of $10 in income that they are allowed to donate as much or as little as they chose to the charity. The subjects must select one charity from a list of eight, each of which includes the name of the charity and a description of the charity services. All charities used in the experiment were international relief and development charities, as these organizations were the most likely to not include religious keyword identifiers on their Form 990 (Scheitle, 2010). Sessions varied by the composition of charities included, as well as the charity descriptions. In some sessions, half of the charities have their religious affiliation information omitted, making it appear religiously neutral, while in others, all charities have explicit religious identifying information included. Additionally, in some sessions, I have Christian charities competing for donations against secular charities, while in others, I have Christian charities competing against Islamic charities, to determine if donation patterns change based on the composition of charity options.
When examining donor behavior over all experiment participants, I find that adding religious language results in an 11-percentage point decrease in donation likelihood to Christian charities who had previously appeared religiously neutral, but only when they are competing against Islamic charities. This corresponds to a decrease in conditional average donations by $1.667 or 16.67% of participant income. No other charity type experienced any change in donations with the inclusion of religious identifying language, and the lack of effect does not stem from insufficient observations. These results were somewhat counterintuitive, as the experiment took place at Florida State University, located in a state where Christianity is the majority religion. The lack of change in donation behavior is not explained by participant religious affiliation, as religious and secular participants donated at similar rates and levels throughout the experiment. However, I soon discovered that the changes in donor behavior were largely canceling each other out based on a different dividing line: political affiliation.
Nearly half of the participants in my experiment identified themselves as politically liberal, providing a natural dividing line within the data. Self-identified liberal participants respond to the inclusion of religious identifying language by decreasing donations to Christian charities competing against Islamic charities by 21.2 percentage points and decreasing donations to Christian charities competing against secular charities by 11.4 percentage points. This translated into a decrease in average donations by $0.901 and $0.281 respectively. No other type of charity experienced a statistically significant change in donation behavior among liberal subjects. Conversely, non-liberal subjects decreased donations to Islamic charities by 17.9 percentage points when religious identifying language was included in the charity description.
My results indicate that Christian charities, charities affiliated with the majority religion in the United States, have a financial incentive to selectively display their religious identifying information, particularly when trying to attract donations from politically liberal donors. While my study is exploratory in nature and has external validity concerns stemming from the mismatch between college students and the average donor, this finding is consistent with an existing relationship between fundamentalist Christianity and conservative political leanings in the United States, which is further detailed in the article. Furthermore, the incentive is likely to only grow as the political climate in the United States becomes increasingly polarized. The political climate in the United States is likely to put religiously affiliated, non-church charities in an increasingly tough position between maintaining a natural network through a religious affiliation and maximizing their appeal to donors. If this trend continues, it is possible that we see more organizations distance themselves from their religious affiliation, just like Child Fund International did in 2009 when the organization changed its name from the Christian Child Fund (Banks, 2009).
Oxley, J. (2022). Examining Donor Preference for Charity Religious Affiliation. Nonprofit and Voluntary Sector Quarterly, https://doi.org/10.1177/08997640221105656.