The scientific study of prayer

Previously: What good evidence for the supernatural would look like

Careful investigation of miracle healing and other oddball paranormal claims is, when done right, an valuable application of scientific thinking. But it isn’t a very typical example of science. So now I want to talk about another story from Craig’ Keener’s Miracles–which is also, as it happens, a more typically example of the book’s content:

Even solid medical documentation is not adequate by itself to surmount strongly held presuppositions, because one may insist in every case (even if there are thousands of them) that another explanation is possible. My colleague in Hebrew Bible, Emmanuel Itapson, was told that his third child had “the death chromosome” and would likely die before birth if not aborted. The family prayed, and the boy is now nine years old. Because 1 percent of those with this chromosome are known to live beyond infancy, one cannot prove beyond any doubt that prayer is the factor that helped him to live so long; yet I am prepared to grant that likelihood in view of the significant number of extraordinary answers to prayer in Emmanuel’s circle, including one mentioned in chapter 9 and another in chapter 12 (p. 666–I did not notice this page number until after choosing the quote).

In spite of Keener’s implication that cases like these are evidence of how closed-minded skeptics are, a couple of problems jump out. First, while “the death chromosome” presumably refers to a lethal chromosomal abnormality, since there are many lethal chromosomal abnormalities, there’s no such thing as “the” death chromosome. Such imprecision is not surprising when a story is being filtered through people without medical or scientific training, but it makes the story harder to evaluate.

More importantly, stories like this are not just explainable, but completely unsurprising without even assuming no miracles. By definition, for every hundred people faced with 100 to 1 odds, one person will beat the odds. In more religious parts of the world, including the United States, I have no doubt that most people pray when they or their children are faced with a serious illness. In that case, most odds-beating recoveries will appear to be “answers to prayer.”

Even the “the significant number of extraordinary answers to prayer in Emmanuel’s circle” means nothing without more information. Humans are terrible at just eyeballing probabilities. And one place we go wrong is in assuming that clusters of seemingly improbable events must have some deeper explanation.

To see why that’s wrong, consider that truly random events won’t discriminate against people who’ve had improbable things happen to the past. This means it’s very unlikely that lucky breaks will be perfectly evenly distributed throughout the population. Probably some people (or “circles” of people) will get more lucky breaks than others. We should even expect a very few people (or “circles”) to get a lot of lucky breaks. (This doesn’t mean some people have an enduring trait of “being lucky”; truly random events don’t favor people who’ve been lucky in the past either!)

This is why carefully-designed scientific studies are vital to medicine. Instead of just giving one person a new drug and seeing if they get better, we ideally take a large group of people and randomly assign them to take the drug or not, and see if one group is more likely to recover than the other. Then we do statistical analysis rather than just eyeballing the probabilities. And ideally we do other things to avoid other potential pitfalls, like giving the people who aren’t getting the medicine placebos to account for the placebo effect.

In fact, in 2006 we finally got a carefully-funded study on the effects of prayer on medical recovery, funded by the Templeton Foundation. The study was specifically designed to correct for faults in previous studies. And it found no benefits from prayer.

  • Reginald Selkirk

    “The death chromosome”? Somehow that was never covered while I earned a bachelor’s in biochemistry, nor while I earned a Ph.D. in molecular biology. Specifics would be helpful, but I can’t say as I feel motivated to read such a book to look for it.
    The Mayo Clinic and Duke (MANTRA) prayer studies were also reasonably good. The only large scale studies in intercessory prayer which have shown a positive effect were those by Elisabeth Targ, and by Columbia University, both of which were fraudulent. So we can say that fraud is more likely than effective intercessory prayer.
    A Prayer Before Dying
    Wired, 2002

  • sailor1031

    A single wooden leg hung up at Lourdes by its formrt owner (not by some priest who….) could answer this question once and for all. Think doG’ll do it?
    Nah – me neither!

  • Greg G.

    The poor pray for wealth. The unhappy pray for happiness. They seldom get rich or happy. The rich andhappy pray or don’t. It doesn’t make a difference.

  • Pingback: The problem with asking science to be religion-friendly

  • stuart

    Just a comment on “By definition, for every hundred people faced with 100 to 1 odds, one person will beat the odds.”
    Not exactly- you could get a run of 100 out of 100 surviving, or none out of a hundred. But over a large sample, the probability of any one person surviving will be 1/100. Statistics need to be applied across large samples, not selected individuals.

    • Chris Hallquist

      Yer, I should reword that.

  • bones

    “By definition, for every hundred people faced with 100 to 1 odds, one person will beat the odds. ”

    –actually, not quite: for every 100 people faced with 100 to 1 odds, ALL of them could die. For every 101 people faced with 100 to 1 odds , one person will beat the odds.