Randomizing Religion: The Impact of Protestant Evangelism on Economic Outcomes

That is a new and important paper by Gharad Bryan, James J. Choi, and Dean Karlan, and here are the results:

To test the causal impact of religiosity, we conducted a randomized evaluation of an evangelical Protestant Christian values and theology education program that consisted of 15 weekly half-hour sessions. We analyze outcomes for 6,276 ultra-poor Filipino house holds six months after the program ended. We find significant increases in religiosity and income, no significant changes in total labor supply, assets, consumption, food security, or life satisfaction, and a significant decrease in perceived relative economic status. Exploratory analysis suggests the program may have improved hygienic practices and increased household discord, and that the income treatment effect may operate through increasing grit.

File under “increased household discord”…


In Cambodia the Protestants pay families to follow that branch of Christianity, and you get absurd results. In my rush to be first I won't give an example. Perhaps the same thing is happening in the Philippines?

'Rice Christian is a term used, usually pejoratively, to describe someone who has formally declared himself/herself a Christian for material benefits rather than for religious reasons.

The term comes from Asian countries, such as India, China and Japan. Concerns have been expressed both by Christian missionaries and by those opposed to Christian missions that people in these situations are only nominally converting to Christianity in order to receive charity or material advancements. One of the earliest examples of this concept in English appeared in 1689 with the writings of William Dampier when he wrote regarding the French priests' effort to convert people of Tonkin that "alms of rice have converted more than their preaching."' https://en.wikipedia.org/wiki/Rice_Christian

Generosity to the poor has definitely been an effective Christian recruiting technique for 2000 years now. It's interesting that so few other movements have tried it.

Muslims are far more generous, thus Islam has been far more effective in gaining converts than Christians.

Sunni followers exceed Catholics, the largest, most effective Christian branch.

Got a source for that?

I did some quick googling, other than the edge case of Sikhs in the UK being the highest average for charitable giving, I didn't see any evidence for that. Of course, if you count the Saudi Kingdom's internal "giving" it would run the numbers up. But most people wouldn't' consider than necessarily charity.

Forgiveness works as well. Ex-Khmer Rouge Official Converts to Christianity Guided by Pastor She Once Enslaved...


I wanted to see if the authors were aware of the obvious Bayesian issue here, which is that if you test for enough variables a few are likely to appear "statistically significant". Evidently they do - they use something called "the minimum false discovery rate." I looked it up, it sounds like it basically means they have a higher bar for statistical significance than p<0.05. Can anybody comment on this so-called "q value"?

First of all note P 10 "We do not apply multiple hypothesis test corrections to our tests of hypothesized mechanisms and secondary outcomes because these analyses are exploratory" so only the q-values have been discounted. The exact FDR procedure used seems to be described at the footnote at the bottom of P 10.

I am inclined to compare the various q-values quoted in Table 1 PDF P 17 and be less than impressed to see that the really low numbers are all pretty much direct effects - religion - except possibly for the somewhat subjective Perceived relative Econ status.

If you wanted to do N tests and keep the risk of reporting any false result at no more than p, you could use Holm-Bonferroni. FDR is more generous than Holm-Bonferroni, especially when there are real effects to be found. In these circumstances the risk that one of the results reported is a false alarm may be higher, but most results reported should be OK (assuming various assumptions are met). I am somewhat conservative here but especially in a situation where there are strong direct effects but the really interesting results are weaker indirect effects I would be much happier to see Holm-Bonferroni used here than FDR. With lots of strong direct effects a guarantee from FDR that false alarms are relatively rare does not convince me that a few indirect effects mixed in with a load of strong direct effects are genuine.

A quick search shows me that Holm-Boneferroni method is a protocol for making it harder to have Type I errors but easier to have Type II errors. As such, it's hardly a cure-all. You can achieve the same effect by making the p-value smaller, say 0.01 rather than the conventional 0.05.

How do you get higher income, with no change to any of assets, consumption, or food security?

Probably assets are measured noisily or some sort of transfer payments or charitable giving are not counted as consumption.

That sounds wrong. I'd like to see what the story behind it is. Hopefully, it's detailed in the paper.

Well the detail amounts to they don't know what the cause is. (page 13)

Higher prices and wages. Instead of helping and sharing, you buy and sell the same things.

When you get paid to do what you did for free, you expect to be better off, except you must now pay others for what they did for free. Higher income, but no better off because consumption is the same, you just have to pay for it.

Thus income is higher but you feel worse off.

The charity didn't give everyone in the community money. Just a select group.

34 Think not that I am come to send peace on earth: I came not to send peace, but a sword. 35 For I am come to set a man at variance against his father, and the daughter against her mother, and the daughter in law against her mother in law. 36 And a man's foes shall be they of his own household.

Well that's American family values for you.

" a significant decrease in perceived relative economic status. "

Does this mean they viewed themselves as poorer compared to their peers?

Is this due to "prosperity gospel"?

From the study: "The Protestant doctrine of salvation by grace — a person cannot earn her way into heaven by performing good works, but can only be saved by putting her faith in Jesus, upon which God forgives her sins as a free act of grace — is taught. The proper response to God’s grace is to do good works out of gratitude. The final section of the curriculum covers what such good works would be. They include stopping wasting money on gambling and drinking, saving money, treating everyday work as “a sacred ministry,” and becoming active in a local church community." It may not fit the definition of "prosperity gospel", but it certainly does involve self-absorption. Would you define "good works" the way it's defined? [Works, as the term was used by St. Paul, meant Works of Jewish Law. Paul taught Gentiles that not only should they not practice Works of Law but if they did they would be denied Grace. Works was later conflated with Good Works, the "Good" implying charity, but that's not the correct usage. Thus, Evangelical Protestants can simultaneously believe in Good Works and self-absorption - including the prosperity Gospel.]

Perhaps all the results are being driven by the effect of the intervention on honesty?

Which to be clear includes how the treatment group respond to sensitive questions.

"education program that consisted of 15 weekly half-hour sessions. We analyze outcomes for 6,276 ultra-poor Filipino house holds six months after the program ended."

A more prosaic possibility is that people who have their shit together well enough to attend fifteen consecutive religious education programs are likely to be in a good patch in their precarious lives relative to ultra-poor people who can't make that commitment. So, it is hardly surprising that their income trend during this period a stability and leisure available for religious education for a sustained period is going to be more positive than a group of non-participants/controls who may not have shared the same life stability/leisure availability for four months.

Let's be blunt: converting devil-worshippers to Christianity would be the most effective help for the Third World.

Converting Christians to devil-worshippers might work even better.

No, it wouldn't!!

Santa Claus killed Christianity. No, seriously.

Christianity died in 325 AD at the Council of Nicaea when Nicolas of Myra, aka, Saint Nicholas, aka, Santa Claus, walked across the room and slapped Arius in the face. While he was arrested for committing violence in front of Constantine, the Roman Emperor, Constantine eventually sided with Nicholas and Arius' point of view that Jesus was created would then be known as the Arian heresy. God would now be a fraction, i.e., a third of three. Christians started worshiping a triune deity and have been confused ever since.

Santa Claus. Ho Ho Ho.

Comments for this post are closed