Waldo recently posted a link to a story about an intercessory prayer experiment published in the American Heart Journal. I disagree with the conclusions reached by Waldo and others, who claim that the study shows that prayer can have a harmful effect on patients.
In essence, the researchers told some patients in the study that they might be prayed for, and others that they were being prayed for, and recorded the percent who had major post-surgery complications. The three roughly 600-person groups were
- told they may or may not be prayed for, and were actually prayed for (51%),
- told they may or may not be prayed for, and were not prayed for (50%), or
- told they were being prayed for, and were actually prayed for (69%).
The researchers compared groups 1 and 2 to test the double-blind nature of prayer, and compared groups 1 and 3 to test whether certainty about receiving prayer would have any effect. Unsurprisingly (to me), prayer had no noticable effect in the double-blind comparison (groups 1 and 2), indicating that intercessory prayer had no effect. However, the certainty/uncertainty comparison revealed quite a disparity: being told you were indeed being prayed for made you 35% more likely to develop major post-surgery complications than if you were simply unsure. The AHJ article notes that the construction of the experiment may have contributed to this effect, thereby contaminating the conclusions. (Would they have noted this if the study had found the opposite result? I don't think so.) Though the researchers were very successful in standardizing the prayer technique and removing extraneous variables, they failed to include what I consider proper control groups.
With something as non-scientific, non-rigorous, psychological, and diverse as prayer, placebo and placebo-like effects need to be controlled carefully, necessitating a wider set of control groups. There needs to be at least another three groups:
- Told they were being prayed for, and were not prayed for
- Told they were not being prayed for, and were actually prayed for
- Told they were not being prayed for, and were not prayed for
Those should all again be compared to the statistics of those not participating in the study, giving a firm baseline. Without the extra components, no interesting conclusions can be reached. I think the study was a huge step in the right direction wih respect to analyzing intercesory prayer, but that they didn't go far enough.
Here's something to think about. What if we added two more groups, unaware they were in a study?
- Not told they were in a study, and are prayed for.
- Not told they were in a study, and are not prayed for.
Unfortunately, three of the extra groups I have suggested would be regarded as unethical because they involve lying about treatment. One of the extra groups would not even be informed that they were taking part in a study. With most studies, this would be an instant ethics-board-smackdown. But prayer is widely (or perhaps instead officially) regarded by the scientific community as having no effect on an unaware subject. Does this mean that it can be excluded from ethics considerations? Or, since that is the very question to be investigate by the first half of the study, is it all the more important?
I personally believe that since the first half of the study concludes that in double-blind conditions intercessory prayer has a solely placebo effect, informed consent is not necessary as long as the subjects are never informed of their participation, that is, the last two groups I suggested. However, I still have trouble with the groups who would be lied to. What are your thoughts?[EDIT: I recently saw an interesting post on helmintholog about Americans and functional atheism, talking about the same study. The author thinks that even religious Americans don't actually think intercessory prayer works, and that therefore prayer has a nocebo effect.]