The Latest Scientific Fraud You're Reading About In The News And On The Web

Taking a break from the topic of the foundation in Darwin's natural selection for eugenics, in both its English and German forms, someone has hurled the newest made-for-media-attention slam at religion, especially Christianity and Islam in the form of a piece of social science as evolutionary biology.  It's a good example of how the Evolutionary Psychology, from here on in to be denoted "evo-psy" for ease of tying and, I hope to indicate what I think of it, and how it, among other things, replaces the construction of a story about the entirely unrecorded past for any actual evidence of what happened in the past which led to who had children and who didn't have children, thus passing on genes, etc.  But, for this post, I'll concentrate on the how of the fraudulent study instead of the alleged framing of the results as biology.

Here is the study, The Negative Association between Religiousness and Children’s Altruism across the World

The study claims

Decety and colleagues assessed altruism and moral cognition in six countries.  Parents in religious households reported that their children expressed more empathy and sensitivity for justice in everyday life.  However, religiousness was inversely predictive of children's altruism and positively correlated with their punitive tendencies. 

Wow, "across six countries" as I heard it reported, this morning,  that sounds like a massive study.  It makes you wonder how many of the people in the media and the twits on twitter who are talking about this read any farther than this "In Brief" description  It says everything that a social scientist could want to be the take-away message which, these days, is guaranteed to get your study media exposure.

What six countries?  Canada, China, Jordan, Turkey, USA and South Africa.   Well, for a start, there is no representation of Latin America or Europe, two rather significant populations with a range of different religious cultures in them.  And, yet, this study is supposed to represent entire worlds of religion.  I would wonder how representative their samples of  religious and non-religious populations in just one of those countries can be believed to be representative of those two populations in any one of those countries, the United States, for example.

Well, all of the children from the United States in their sample came from Chicago, hardly a place which is typical and which, as any large city will, has its own particular cultural quirks.  And, I will point out, hardly likely to be similar to a sample of children taken from a rural population or, perhaps, a suburban one. And which children in Chicago?  Those who go to a prep-school?  A parochial school?  A public school?  A public school in a poor neighborhood or in an affluent neighborhood?  The recruitment was done in school settings, somehow, from what follows, I don't think the children in Chicago were recruited from a range of schools.

That becomes an interesting problem for taking this study at all seriously when you look farther into their methodology, which is typical of these kinds of experimental studies.

The number of children studied in all is 1, 170, to continue with the U.S. portion of that number, I can't, from my quick, sleep-deprived glance at this study find what percentage of those where American children who live in Chicago.  But if, say, it is one sixth of that number, that would be 195 children from the Chicago area who represent the hundreds of millions of Christians in the United States, most of whom were raised in Christian homes.  Only, no, that isn't right either because, presumably, some of those 195 children must have been, or, ideally, should have been Muslims because there are Muslims mentioned in the Study, as well as members of other religions and those brought up in non-religious households. The study says that its universal sample were 23,9% Christian, or a sample of 280 children to represent all of the Christians in the world as characterized by their results.   If you take a sixth of that, you find that about 46 and two-thirds of children in the Chicago area might be purported to yield significant results telling you something about the character of well over two hundred million Americans who are Christians, in vastly different denominations.

And that number of 1,170 children isn't what the results were based in because they threw out the numbers for religious groups which were too small to yield what they purport are significant results.   Oh, and just so you can guess the nature of the science of this study, while they base their characterization of the billions of Christians of the world on 280 children, they define the more than a billion and a half Muslims on a population of 510 and of "not religious" on  323 children.  While noting that the sample sizes are skewed to not reflect the percentages of Muslims and "not religious" populations in the world, the size of those samples are inadequate, as well.

I don't think a reputable statistician would hold those are valid sample sizes.  In her extremely valuable book for non-statisticians, Seeing Through Statistics, Jessica Utts in discussing the issue of sample size in judging the validity of the results says,

The more divers, or variable, the individuals are within each group, the larger the sample needs to be to detect a real difference between the groups.  

Given that logical and sensible rule, the stupendous inadequacy of a sample of 280 children to represent the set of all Christians in the world is easily seen from this:

As defined here, world Christianity consists of 6 major ecclesiastico-cultural blocs, divided into 300 major ecclesiastical traditions, composed of over 33,000 distinct denominations in 238 countries, these denominations themselves being composed of over 3,400,000 worship centers, churches or congregations.

World Christian Encyclopedia by Barrett, Kurian, Johnson, Oxford Univ Press, 2nd edition, 2001

Clearly the world wide numbers of billions of children brought up in Christian homes can't be characterized by 280 children sampled in Chicago, Toronto Canada, Amman Jordan, Izmir and Istanbul Turkey, Capetown South Africa and Guangzhou China, which is purported to yield a characterization of billions of Christians around the world.

And note, those are cities, the "in six countries" claimed in the study is a fraudulent claim, it is actually in seven municipalities, in which of any, the samples are so tiny that they couldn't possibly indicate anything typical of those cities,  the hundreds of thousands, hundreds of millions and on to billions of people in any of the purported parts of the world population they pretend to define.  

On that basis alone, this study is a scientific fraud.

And that's only the problem of their clearly and absurdly inadequate sample.  Their methodology to discern differences in "altruism" and "punative" characteristics in the huge range of real world childrens' lives is ludicrous.   And there are more problems with their sampling

The children who were "recruited" (aka, volunteers, to some extent, not a randomly chosen sample)  from schools (in some of those countries school attendance is related to income and in other cities, such as Chicago, given schools don't have a reliable distribution of incomes to be representative).

The ages of those children studied ranged from 5 to 12.  OK, there's a red light right there for anyone with real life experience of real children.  There is no way to guess how even one child would react to their study, based on a lame "Dictator Game" at age 5 and at age 12, there is no way to guess if a kid who is as generous as anything at age 5 won't turn into a totally self-absorbed, selfish brat under the influence of puberty, which may well be just taking over at 12.  Not to mention any single child at any of the other ages.  A 12-year-old can display entirely different behaviors one day to the next.

And that's in real life.  Their study was based on a totally artificial "Dictator Game" the point of which, I don't get and I doubt any of the children in the study got.   Here is the entire description of it in their study.

Dictator Game 
In this task, children were shown a set of 30 stickers and were told to choose their ten favorite.   They were then told “these stickers are yours  to keep.”  Children were instructed that the experimenter did not have the time to play this game with all of the children in their school, so not everyone would be able to receive stickers.

Huh?  Where's the game?  I defy anyone to tell me what the point of it is for an adult imposing it on these kids, never mind what it's supposed to mean to the kids being studied.   And, really, STICKERS?  Those are supposed to be of some great value to children ages 5-12, from totally different backgrounds and cultures?   I would bet that you would find the full range of interest in stickers in a class room of any one of those ages but I can guarantee you that by the time those kids are 12 a lot of the kids would consider STICKERS baby stuff and expressing an interest in them would get you mocked by some of the kids.

I could go on all day pointing out why this study is a complete piece of crap as so many social science and, these days, neuro-sci studies are, based in ridiculously tiny samples which are in no way randomly chosen or representative of the population they are supposed to represent.   And this study is so bad that I will accuse its authors and Cell: Current Biology, where it is published to have had an ideological motive in its publication, which was guaranteed to get into the news as another slam at religion.