Summary

• In May 2022, EA for Christians conducted a survey of 400 Christians from the US and UK on attitudes to charity and altruism, including questions about donations, attitudes towards some EA causes, and ethics
• Respondents were recruited through online survey platform Prolific
• The sex and age distribution was approximately the same as the average of the Christian populations in the UK and US

Highlights from Results

• Reported concern for effectiveness and impact was higher than I (Vesa Hautala, Research Coordinator at EA for Christians) expected 
  • 65% said they worry whether their donation has any effect at all and 
  • 63% said they always consider how many people the charity they’re donating to is able to help with their donation
• Christians prioritized Christian causes and helping other Christians much less than I would have thought
  • Only 15% agreed that Christians should prioritize helping other Christians over non-Christians
  • 26 % agreed that Christians should primarily donate to Christian charities and 
  • 26% agreed that Christians should prioritize evangelism over other causes
• Preference for helping those socially, geographically, and temporally near was correlated with prioritizing Christian causes and helping other Christians
• About 80% of participants agreed that animals can suffer and about 75% agreed that they have rights
• The participants’ responses had less correlation with their age than I expected
• 66% agreed that pulling the lever in the classic trolley problem is the right thing to do

Introduction

At the end of May and early June 2022 EA for Christians ran a survey about Christian attitudes to charity and altruism. The purpose of the survey was to provide information for a book we are working on and in general provide EACH with data about Christian attitudes relevant to Effective Altruism.

The survey participants were 400 people from the US and UK who identify as Christian and participate regularly in public and private religious activities.¹ The participants were recruited and prescreened using Prolific, a service that provides respondents to online studies. The participants filled out an online questionnaire of 35 multiple-choice questions.

The survey was run in six different segments with varying numbers of respondents: age groups 18–29, 30–49, and 50–64 for males and females. This was done to achieve a sex² and age distribution that approximates the demographics of the Christian population in the UK and US.³ This procedure was necessary because Prolific’s user base is younger and has more women compared to the Christian population in the UK and US. Ages 65 and above were excluded because there were not enough respondents available in Prolific.

Limitations of the survey

• The participants in this study were all people who fill surveys for money
• Respondents were paid for completing the questionnaire which might result in filling it without carefully considering the questions. Attention checks were implemented in the study to mitigate the negative effects from this, but they only guarantee a minimum level of attention
• The survey was designed and executed by a non-expert, though with input from experts
• No corresponding survey for a non-Christian population, so hard to compare Christians to the whole population

Results

The correlation mentioned are Bayesian Kendall’s Tau correlations calculated using JASP. The correlations I talk about in this post mostly had a Bayes factor (log₁₀) of over 100, so were highly statistically significant. Usually if a correlation had a smaller Bayes factor, I have mentioned it in the text. A table of correlations (not containing all questions in the survey and not including demographic data) can be found here.

Donations

Of the respondents, 65% say they worry whether their donation has any effect at all and 63% say they always consider how many people the charity they’re donating to is able to help with their donation and by how much. Ideas about how much more effective the best charities are compared to average charities were not strongly correlated with responses to other questions. (There was weak positive correlation with considering how many people a donation helps and by how much; 0.13 BF₁₀ 97). This might imply that ideas about effectiveness of charity aren’t clearly tied to other ideas which might make them more susceptible to influence.

Preference for Christian causes

The survey had a number of items gauging whether the respondent prioritizes helping other Christians over non-Christians and Christian causes over other charity (I use “Christian causes” as an umbrella term including evangelism, financial support for local church, and Christian charities). A majority of the respondents did not show partiality to Christian causes or helping other Christians. Of all the statements in the survey, “Helping other Christians should take priority over helping non-Christians” provoked the highest amount of disagreement with 42% of respondents strongly disagreeing and 29% somewhat disagreeing. Only 27% agreed that Christians should prioritize evangelism over other causes and in a question comparing funding missionaries vs funding life-saving charity, 21% agreed they would fund the missionaries. However, when it comes to supporting one’s local church over charity, agreement and disagreement were more evenly split, even with the additional specification that the money donated to charity could save lives.

The finding about evangelism is interesting because applying a utilitarian logic might seem to suggest that evangelism should trump all other endeavours in a Christian worldview because of the infinitely large value of an eternity in heaven. I’ve also personally encountered this suggestion from non-Christian EAs multiple times. Opinions about prioritizing evangelism did not have either a positive or a negative correlation with statements that related to consequentialist morality, which should be the case if a more consequentialist outlook would cause Christians to value evangelism more highly – or then our measures of consequentialist thinking were inadequate. There is of course also variance within the views of Christians about eternal salvation and evangelism.

Prioritising Christian charities, evangelism, and helping other Christians were correlated with each other and with prioritising support for the local church over charity and, interestingly, also agreement with the statement “I aspire to live on as little as possible, and give the rest away.” Agreement with the statement “My money is mine to do whatever I want with it” also had a slight negative correlation with prioritising Christian causes. Prioritizing other Christians and Christian causes correlates with higher reported giving to church but not with higher giving to charity. (See table linked above for the precise values).

Local vs global

Several items measured preference for helping those socially, geographically, and temporally near over helping those further away. All of these were correlated with each other and with prioritizing Christian causes and helping other Christians. This could be explained by other Christians and Christian causes counting as socially near to respondents who have a local preference or with another underlying factor that influences both Christian preference and local preference.

Animals

In a question asking respondents to rank different cause areas based on how important it is for them to receive additional resources, 140 respondents out of 391 ranked preventing farmed animal suffering the last. In a question asking the participants to select from a list of causes those they would not personally donate to, about 25% chose preventing farmed animal suffering (compared to about 50% not saying this about any cause, 4% about global health and poverty, and 28% about X-risk).

This is curious considering that about 70% agreed that animals have rights and about 80% agreed that animals can suffer. It could be that the respondents still consider animal welfare important but mitigating current human suffering even more so, or that they are not making a connection between their stance on animal suffering and rights and the situation of farmed animals.

Agreement with “Animals have rights” was negatively correlated with prioritising Christians and Christian causes and helping locally versus globally.

Politics

Several questions measured how important the respondents considered certain things in choosing who to vote for. Overall, the respondents did not consider foreign aid particularly important or particularly unimportant. Results were similar for GDP growth. Pacifist/nonproliferation policies were mostly considered unimportant or moderately important.

Greater importance placed on pacifist/nonproliferation policies, foreign aid, and social justice was negatively correlated with prioritising Christians and Christian causes and helping locally versus globally. These issues were also strongly correlated with each other and with placing greater importance on the religious beliefs of individual politicians and protection on traditional social values. These were also correlated with prioritizing other Christians and Christian causes.

Demographics

The participants’ responses had surprisingly little correlation with their age. My initial expectation was that older participants would tend to prioritise Christian causes, helping other Christians, and helping locally, as well as believe less in animal rights and animal capacity to suffer. This turned out not to be the case. Older respondents reported more agreement with considering how many people their donation is able to help and by how much (correlation 0.18), but less worry about whether their donation is effective at all (-0.13), agreed less that their career was their most important resource for making the world a better place (-0.18), and with the end justifying the means in everyday decision making (-0.13 BF₁₀ 89). Older respondents tended to report donating more to charity (0.17) but not to churches.

Females were less likely to prioritise helping Christians over non-Christians (-0.14) and donations to Christian charities (-0,12), less likely to agree it is acceptable to cause harm when it is necessary to help others (-0.23) or that it is permissible to prioritise spending on family and loved ones over those in need (-0.17). They were more likely to place greater importance on social justice as a voting criterion (0.17).

American respondents were more likely to support training missionaries over using the same money for charity (0.14), to agree Christians should prioritise evangelism over other causes (0.21), and to agree Christians have a stronger moral obligation to support their churches than donate to charity (0.21). In the voting questions, American respondents tended to give somewhat more right-wing coded answers, placing less value on social justice (-0.25), foreign aid (-0.19), and pacifist/nonproliferation policies (-0.16) while placing greater value on protecting individual liberties (0.20) and traditional social value (0.13).

Appendix: Questions and responses

Note, the attention check questions are omitted.

Questions 1–25

How much do you agree with the following statements?

Questions 26–35

Q26. How much more effective do you think the best charities are compared to typical ones? (meaning how much they achieve with the same donated sum)

• 23,7% usually there are no large differences between typical and best charities
• 11,9% less than 2x
• 35,9% 2–10x
• 24,5% 10–100x
• 4% 100x or more

Q27. When I think about how much people in my situation ought to donate to charity, I think that the amounts I donate are

• 9,1% much less than I feel is right
• 31,1% less than I feel is right
• 53,3% about right
• 6,6% go beyond the call of duty
• 0,0% go much beyond the call of duty

Q28. About what percentage of your income did you donate to a church in 2021?

• 14,1% — I did not donate to a church in 2021
• 19,4%  — 1%
• 12,1%  — 2%
• 17,7%  — 2 – 5%
• 22,2%  — 5 – 10%
• 13,1%  — 10 – 20%
• 1,3%    — 20% +

Q29. About what percentage of your income did you donate to charity (excluding donations to churches) in 2021?

• 11,9% — I did not donate to a church in 2021
• 23,2 %  — 1%
• 18,2%  — 2%
• 23,7%  — 2 – 5%
• 14,1%  — 5 – 10%
• 7,3%   — 10 – 20%
• 1,5%   — 20% +

Q31. How important it is for these areas to receive additional donations? Rank from most to least important.

Q32. Check all of the following causes you would personally NOT donate to

• fighting extreme poverty — 4%
• preventing farmed animal suffering — 24.5%
• improving people's lives in the next centuries and millennia — 27.5%
• reducing the risk of humanity's extinction 25%
• reducing global catastrophic risks such as pandemics and nuclear war (i.e. risks which will harm many people massively but are unlikely to cause human extinction) — 14.9%
• I might donate to all of these causes — 50.5%

Q33. How important do you consider the following for career choice? Rank from most to least important. [Note, the numbers are numbers of respondents who selected the option for the particular career, e.g., how many respondents selected  "least important" for "salary/material benefits".]

Q34. When choosing who to vote for, how important the following are to you?

Q35. The trolley problem (classic version)—the right thing to do is to pull the lever

• Strongly disagree 6,3%
• Somewhat disagree 7,1%
• Neither agree nor disagree 20,7%
• Somewhat agree 45,7%
• Strongly agree 20,2%

(Trolley problem wording used in the survey: A runaway trolley is heading down the railway tracks. Ahead, five people are tied up and unable to move on the tracks. You are standing next to a lever that you can pull to divert the trolley to a different set of tracks, where one person is similarly tied up. You have two and only two options: 1. Do nothing. The trolley will continue ahead and kill the five people on the track. 2. Pull the lever. The trolley will be diverted to the side track and kill the one person there. To what extent do you agree with the statement: “Pulling the lever (i.e. option 2) is the right thing to do?”)

1 The exact wording of the prescreening questions was set by Prolific and could not be altered. For religious participation, the question was “Do you participate in regular religious activities?” Our survey was filled by respondents who answered this question with the option “Yes. Both public and private” (as opposed to only public, only private, or rather not say). The other prescreening question was “What is your religious affiliation?”, with the answer option “Christianity (e.g. Baptist, Church of England, Roman Catholic, Methodist, Jehovah [sic] Witness, etc.)”. This unfortunately did not distinguish different types of Christians or Christian-adjacent groups. Prolific is supposed to show the study only participants who match the prescreening criteria but allows validating the prescreeners and filtering out those who provide incompatible information.

2 A sex distinction of male and female was used instead of a gender distinction and no ‘Other’ or ‘Do not wish to answer’ option was included in order to match the available demographic data about the US Christians population, which only had a binary sex distinction. The data used was from the Pew Research Center’s 2014 Religious landscape study. Pew Research Center’s documentation is somewhat unclear on whether their distinction is a gender or sex distinction, but the closest match in Prolific seemed to be sex.

3 The age distribution was based on both US and UK data whereas the sex ratio was based on US data alone because UK data was not easily available.