Friday, May 25, 2007

Miracles and other nonsense by Victor Stenger

T: My belief in God is not based on blind faith. It is based on evidence. How can you deny the countless reports of miracles going back thousands of years?
A: They are all anecdotes. Anecdotal evidence alone is never sufficient. It must be supported by controlled experiments and observations.

T: But how can you explain all those reports?
A: Without details on the observations, I can only offer possible explanations.

T: OK, what are those?
A: Delusions, hallucinations, even outright lying and fraud.

T: You cannot prove any of these. And how unjust of you to accuse people of fraud without proof! Whatever happen to the principle that you are innocent until proven guilty?
A: That principle may be fine in a law court, but these are not legal questions but scientific ones. In science you are guilty until you prove yourself innocent. Besides, I made no accusations. I only said I was going to offer possible explanations. Fraud is one possible explanation among the others. That is the problem with anecdotal tales. Not enough data are available to determine the true explanation.

T: Then your possible explanations are no better than my equally possible explanation--that God enacted these miracles.
A: No, I disagree that your explanation is equally possible. My explanations are based on well-established facts: People have been known to be deluded, have hallucinations, and lie. These explanations are unexceptional and so must be ruled out before you can consider extraordinary explanations that are not based on well-established facts.

T: Well, many miracles are well-established facts. Take weeping icons and stigmata, for example. They have been witnessed by many people, including skeptics.
A: Yes, these phenomena have been observed by skeptics. But they are not miraculous. Natural explanations have been given and the effects dupicated by investigators. They could be honest psychological effects. Or, they could be faked. For example, go into a church early in the morning and rub some Mazola oil on a stained glass window, say on the face of the Virgin Mary. When the sun later shines through the window, the oil will warm and trickle down like tears. Other substances placed on statues in churches produce a similar effect, looking like tears or blood. As the church fills with worshippers, it warms and the substance melts. As for stigmatists, many have been discovered to wound themselves. For detailed explanations of these types of miracles, see Looking for a Miracle by Joe Nickell.

T: I understand that hundreds of scientific studies have been reported showing the health benefits of religious behavior.
A: These cannot be taken as evidence for the existence of God. They merely indicate that certain types of behavior can be good for you. Going to church may lower your blood pressure for a few hours, but so can many other, secular relaxation techniques. What's more, much religious behavior is not beneficial. Many innocent children die each year because their parents rely on faith healing rather than modern medicine.

T: I heard of a study that showed that church-goers specifically are more healthy than non-church-goers.
A: I have heard of such studies too. One failed to account for the fact that a lot of sick people are too immobile to go to church. When the authors later corrected for this, no differences were found.

T: Still, there are many such studies. There must be something to it.
A: Epidemiological studies are notoriously difficult to interpret because of all the so-called "confounding factors" that are not controlled for. Just because a correlation is observed, that does not mean a causal connection has been observed. For example, if a study revealed that people who carried around matches had a higher rate of lung cancer, that would not mean that matches cause lung cancer.

T: Evidence has been reported that prayer helps healing.
A: That can be explained by people feeling better because others care or them. Although, the data indicate that this does not work for mental patients and alcoholics, who seem to recover more slowly because of the stresses caused by family interference.

T: But carefully controlled studies have shown that prayer has healing power even when the patients do not know they are being prayed for.
A: There were two recent studies that made such a claim, but neither was statistically significant. Other studies, such as a larger recent one from the Mayo Clinic, have failed to confirm these claims.

T: You say they were not statistically significant, yet they were published in reputable medical journals.
A: Medical journals have a low publication threshold which may be suitable for their purposes of quickly disseminating information about possible therapies but are unsuitable for extraordinary claims.

T: Who are you to say they are unsuitable?
A: I am simply stating a fact. Let me make it precise. Medical journals, and those of other healing sciences such as psychology, typically allow a paper to be published if it has a statistical "p-value" of 5 percent or lower. At this level, 1 in every 20 experiments will report effects that are nothing more than statistical artifacts. Since only positive results are often reported, 19 similar experiments showing no effect could easily be lying around unpublished for every one that is published. By contrast, in my field of physics the typical p-value threshold in 0.01 percent. That is, of every 10,000 experiments where there is no real effect, only 1 (on average) will be published. No claim is made about the proportion of experiments that are published with p<=0.0001 where the effects are real. Moreover, no new phenomenon is accepted until it is independently replicated several times with the same quantitative effect size. This has not happened for prayer or other alternative therapies. Even when replication is reported, you will find it does not duplicate exactly the results being claimed, often referring to a totally different observation. The science here is very bad and undeserving of so much attention.

T: I don't see how the prayer studies can prove anything anyway. How could you ever control that a patient is not prayed for? The Pope prays daily for the ill.
A: I agree that a negative result can be explained away in this fashion. But you raised the argument that the efficacy of prayer provides evidence for the existence of God. I agree that, in principle, the existence of a supernatural power who answers prayers could be demonstrated by a well-designed experiment that produces a huge effect that cannot be explained naturally. This has not happened.

T: OK, but if even if the absence of convincing evidence is a fact, it does not prove that God does not exist.
A: True, but you would think that with the billions of prayers that have been made over thousands of years, we would by now have some evidence that they work. This seems to be a pretty good indication that a God who answers prayers is highly unlikely to exist. And, of course, it says nothing about a god who does not answer prayers.

T: What about psychic phenomena? I read that the existence of ESP is now solidly confirmed. Does this not indicate that the human mind is more than a purely material phenomenon of the brain, that "spiritual forces" beyond matter exist?
A: This claim is incorrect. ESP is far from confirmed. Like the prayer studies mentioned above, no ESP report meets the standards for the acceptance of a new phenomena that are conventional in physics and other "hard" sciences.

T: But the ESP reports are published in peer-reviewed journals.
A: Yes, but almost all of these journals are produced by believers and the "peers" who do the reviewing are also believers. Few reports of ESP have been published in reputable journals which utilize a wider selection of peer reviewers. Those reports that have managed to get published in top scientific journals like Nature have all been refuted in those same journals.

T: What about the results from Princeton in which the mind has been shown to affect the output of a random number generator? They have a a very low p-value. And, they have been replicated.
A: These results have never been independently replicated at the same effect size, which is extremely tiny. Such a small effect can be produced by any number of plausible systematic errors and their experiment has many flaws that the investigators refuse to correct.

T: In his book The Conscious Universe, Dean Radin reports a meta-analysis of many ESP experiments which shows that, while none themselves may be significant, taken together they comprise a body of data that establishes the reality of the phenomenon beyond the shadow of a doubt.
A: Meta-analysis is notoriously unreliable because it is so sensitive to how it is done. Garbage in, garbage out. Besides, Radin's meta-analysis method has been proven to be in error.

No comments:

Post a Comment