Is New Study on Altruism All True-istic?

Atheists across social media have joyfully been sharing this story:

It definitely evokes a reaction...    

For the nonreligious, it is cause to gloat over their latest victory .

Churchgoers are crying foul, meanwhile.

God-fearing bloggers are scrambling to discredit the study with stories like this:

Their rebuttal hinges on the fact that the original study was heavily skewed with Muslims, ignoring that the researchers broke it down by religion and showed that the Christian children in the story were also less giving than the atheist children.  (But everyone that watched the famous Bill Nye v. Ken Ham debate already knew not to expect strong debate skills from conservative Christians in the media.)

I can't blame the journalists entirely for this headline, for once.  The researchers spoon-fed their biased interpretation of the study to anyone reading their report.  The first thing that caught my attention (and made me suspicious of a biased approach) was this line from the summary:

"Together these results reveal ... how religion negatively influences children’s altruism, challenging the view that religiosity facilitates prosocial behavior."

And this line, from the results:

"...the religious rearing environment fundamentally shaped how their altruism was expressed."

 Both of these statements imply a causal relationship, i.e. - being religious causes a person to be less giving.  All scientists know that causality is extremely difficult to prove.  A study like this only shows correlation, i.e. - religion and altruism are related somehow.

I am always on the hunt to expose correlations that are masquerading as causation, so I dove into reading the actual study instead of just news reports about the study.  The researchers spend a lot of time discussing the significance of religion in predicting how giving a child will be.  They have this nice graph that shows how Christian and Muslim children are both less charitable than secular children:

Figure 1, Decety et al., The Negative Association between Religiousness and Children’s Altruism across the World, Current Biology (2015), http://dx.doi.org/10.1016/j.cub.2015.09.056

What caught my attention was where the children came from, however.  The study recruited children aged 5-12 from six different countries.  The numbers are given in the following table from the study.

Decety et al., The Negative Association between Religiousness and Children’s Altruism across the World, Current Biology (2015), http://dx.doi.org/10.1016/j.cub.2015.09.056

The researchers do not give a breakdown of which religions came from which countries, but it does not take a genius to figure out that Jordan and Turkey are supplying most of the Muslim participants.  Each of those nations is over 90% Muslim.  It is likely that most of the atheist children are from China, Canada, and the U.S.  China is over 50% non-religious, while Toronto and Chicago (the cities in the study) have non-religious rates of 21% and 40%, respectively.

Since the study does not provide this information, I am left to formulate my best guess at where the children came from.  Using the demographics of each country, I developed the following chart.  It's not exact, but I'm sure it gives us the general picture of things.

You'll notice three things about the people in this study:

• Most of the atheists are coming from China, Canada, and the U.S.  
• Half of the Christians are coming from South Africa, with the rest from Canada and the U.S.
• The vast majority of Muslims are from Jordan and Turkey.

The authors make no mention of collecting equal proportions of each religion from each country, nor do they make any mention of correcting for country or socioeconomic status.  So we must ask the questions: Does the country a person is raised in affect their altruism?  Does the socioeconomic status of a person affect their altruism?

The researchers provide us with answers to both of those questions.  I'm not going to get into statistics, but there is a number known as a "βstandardized" that tells you how predictive a characteristic can be.  The higher the number, the better you can predict things based on that characteristic.  Here are the top four predictors found in the study: (The negative number on the religion just means as one goes up, the other goes down.  It is the magnitude of the number that is important.)

Age was by far the strongest predictor of whether or not a child was giving.  The older the child, the more they shared.  Socioeconomic status is a STRONGER predictor of giving than religion.  The six countries of the study have different socioeconomic factors.  Also, what country a child is from is only slightly less predictive than their religion.

I would classify this study as "Mostly Useless" until the authors provide data to show that the country and socioeconomic status of the children didn't outweigh the effects of their religion.    This is a classic case of having too many variables.  Why not do the study in one school, in one country, with children from one socioeconomic class?  I might be a little more charitable toward the authors of this study if that were the case.