This is psychologist, Dr Wendy Walsh, discussing the merits of religion on CNN. Rather melodramatically, she is focusing on the purported advantage of being religious during catastrophic survival situations. According to Walsh, “Most studies on survivors show that the atheists die first“, suggesting that religious people survive longer. This news came as quite a surprise to health scientists working in fields like epidemiology, as well as to social scientists who study the impact of religion on society, many of whom publicly declared that they had absolutely no idea what Walsh was talking about. And why might atheists have trouble surviving? Well, according to Walsh, “if you don’t believe in something supernatural, how can you imagine that you yourself have supernatural abilities enough to survive?” In other words, all it takes to survive a catastrophe is sufficient imagination, something that being religious gives you. Hmm. At first blush, this might not sound like a scientific discussion. However, Walsh does state that her conclusion is based on “studies“, of which “most” allegedly support her position. Over the days since this interview was broadcast, Walsh’s claims have been extensively criticized by atheist and religious commentators alike. But presumably, in fairness, Walsh is not insane. So what studies might she be referring to? And can it be true that the evidence shows that “atheists die first“?

It is actually quite difficult to be sure which research Walsh is describing in her CNN interview. Most criticism of her claims (see, for example, here, here, and here) has been premised on the assumption that she is referring to survival in the event of natural disasters and the like. The big problem is that the scientific community has been unable to identify the relevant studies. However, Walsh has now clarified her position on her blog. It turns out that she was unfortunately unclear in her CNN comments and feels she has been badly misinterpreted by her critics. In a noble move, Walsh takes personal responsibility for the controversy — she says that her lack of clarity on CNN was the result of “live TV shorthand“, a fallibility for which she would now like to “deeply apologize“.  So that’s okay then. We can ignore any suggestion that “atheists die first”, right? Well, actually no, her clarification doesn’t go quite that far.

According to Walsh, atheists still die first in survival situations, it’s just that now we are talking about situations characterized by what Walsh calls “medical survival“:

Religiosity helps people cope with illness, and may even impede the progression of disease. People recovering from open heart surgery are three times more likely to survive if they have religious faith. Black women with breast cancer have lower survival rates than white women — unless they are religious. Then they are much more likely to outlive their white sisters. In India, doctors scratch their heads as babies born to strict Muslim families in poverty have higher survival rates than upper caste Hindus with less religion. Prayer can help manage anger and lower blood pressure. And meditation (a new age version of prayer) can increase memory, self-esteem and empathy, and even slow the progression of HIV.

Phew! That’s quite a lot of research. But at least now we can set about critiquing Walsh’s updated, non-shorthand, claim that “most studies” tell us that atheists die first in (medical) survival situations. Here’s a sneak preview of the critique: they don’t.

• The finding that “people recovering from open heart surgery are three times more likely to survive if they have religious faith” relates to a 1995 study published in the journal Psychosomatic Medicine. This study examined deaths following open-heart surgery in 232 patients. In the overall statistical model, 3 of the 23 persons declaring themselves to be non-religious had died within six months (i.e., 13%) compared with 18 of 209 self-declared religious adherents (i.e., 9%). The comparison here (13% vs. 9%) is “1.4 times more likely” rather than Walsh’s “three times more likely”, meaning that Walsh’s description is an exaggeration that more than doubles the actual finding. However, the overall figures in the study were so low that the comparison was statistically non-significant, which means the variations in survival were no different to what would be expected to happen by chance. It’s like ten coin-tosses yielding six heads and four tails rather than five heads and five tails. In other words, there were no statistical differences in survival between religious adherents and other patients. [The study did find a statistical effect when comparing patients’ responses to a question about the amount of comfort they find in religion. However, at least one major problem here is that of cause-and-effect. “Finding comfort in religion” will be strongly associated with the ability to find comfort in anything, which will likely be far lower in people who are critically ill. The finding was simply that sick people are less able to find comfort from religion than well people, and had nothing to do with “atheists dying first”.]

• The finding that “Black women with breast cancer have lower survival rates than white women — unless they are religious” comes from a 2003 study which appeared in the International Journal of Psychiatry in Medicine. While Walsh is correct that the authors report differences in survival rates associated with differences in both race and religion, what Walsh omits to mention is that the authors themselves acknowledged that the findings were statistically unreliable due to huge imbalances in the numbers. While the study included 303 “religious” women, it included only 11 women who declared themselves to belong to no religious denomination. Remember, only a subset of these 11 women were Black; and only a subset of this subset ended up dying from breast cancer. Interpretation of findings based on such tiny numbers is totally meaningless. To make matters even worse, some of the data were fictitious. This is because the study was complicated by such a problematic level of missing data, that the researchers had to extrapolate in places in order to fill in the blanks. Overall, the “finding” here was statistically neither meaningful nor reliable, and Walsh’s summary of it is a massive exaggeration.

• The finding that “babies born to strict Muslim families in poverty have higher survival rates than upper caste Hindus with less religion” is drawn from a 2010 study in the Journal of Health Economics. In short, the paper examined the nature of differences between babies born to Muslims and to Hindus in India. First of all, we can note that the paper focuses on the survival of babies, not of parents. I don’t think it is an overly philosophical point to make, but the babies themselves can hardly be described as being adherents to, or believers in, either Islam or Hinduism. Similarly, differentials in survival rates can hardly be attributed to a baby’s capacity to — in Walsh’s words — believe in something supernatural and thereby imagine that they themselves have supernatural abilities enough to survive. So even at face value, this study is far removed from Walsh’s argument. Secondly, we should also note that this study did not examine atheists, which makes it completely unconnected to Walsh’s argument. Unless, that is, Walsh is asserting that Hinduism is somehow “more atheistic” than Islam, perhaps because, on some kind of vague spectrum of relativities, it lies further away from Christianity. (The idea that Christian fundamentalists believe non-Christians to be a type of atheist is not completely outlandish: just last week, aspiring US Presidential candidate Newt Gingrich warned that a “secular atheist” America would be dominated by “radical Islamists”.) Notably, while the original study authors identified better survival for newborn babies among (poor) Indian Muslims relative to (rich) Hindus, they concluded that this was a secondary statistical consequence of the fact that poorer mothers suffered a higher number of children dying in utero (i.e., miscarriages). For Walsh to distort this tragic finding into a partisan argument about atheism lacks compassion, to say the very least.

• The remaining two findings alluded to by Walsh can be dealt with very quickly. The finding that “Prayer can help manage anger and lower blood pressure” relates to a fairly banal principle that activities involving deep concentration and relaxation will be associated with reductions in cardiovascular functions such as blood pressure. As a result, several studies have found that when people pray, their blood pressure levels reduce. However, the crucial point here is that there is nothing especially unique about religious prayer in this context; any type of medidative concentration will have this effect. (Indeed, exuberant activities such as gospel singing or rapturous worship will likely elevate blood pressure). As such, this principle has nothing to say about differences between atheists and religious adherents. The finding that “meditation (a new age version of prayer) can increase memory, self-esteem and empathy…and slow the progression of HIV” falls into a similar category. However, as it relates to meditation, a practice which can be performed just as easily by atheists, and not to religion, it has no bearing on Walsh’s argument. Indeed, it probably tells us something about the shortage of relevant empirical evidence of any quality that Walsh feels the need to cite such studies of non-religion.

So, in summary, Walsh’s assertion that atheist die first in (medical) survival situations is not supported even by the studies she alludes to herself, never mind by “most studies” as she claimed in her CNN interview. Overall this case shows us the importance, yet again, of drilling down into the research evidence being cited by an apparently informed (but in actuality highly partisan) public commentator. It is sometimes said that nearly any position can be supported by research evidence. I’m not sure that’s true, but it is very much the case that a commentator can easily support any position with evidence if they are willing to distort and misrepresent the research they are citing. In most situations, few members of the audience will be in a position to pick up on the flaws.

If Walsh is really interested in becoming informed by the findings of research, and not just in marshaling studies that can be spun in ways that support her religiously predetermined position, then she may want to consider the following: (a) research suggests that strong atheism is associated with lower rates of depression; (b) several studies show that national average IQ is higher in countries that have more atheists; (c) data gathered by a US-based evangelical Christian lobby group show that divorce rates are lower among atheists than among religious adherents; and (d) cross-national comparisons show that more secular countries have lower rates of homicide, abortions, juvenile mortality, and adolescent STDs, and, surprise surprise, have longer life expectancies. No doubt these studies have plenty of flaws and require lots of caveats. However, it is clear that the overall picture on the merits of religiosity and atheism is far more complicated that Dr Wendy Walsh would have you believe.

It often appears that when religion is mixed with science, objectivity dies first.

Brian Hughes

Brian Hughes is an academic psychologist and university professor in Galway, Ireland, specialising in stress, health, and the application of psychology to social issues. He writes widely on the psychology of empiricism and of empirically disputable claims, especially as they pertain to science, health, medicine, and politics.