What does science say about faith healing?
While skeptics have taken on individual faith healers, scientists have conducted research on the effects of prayer on health. Numerous studies and meta-analyses have been done, but their results raise more questions than answers. In general, a few studies have shown small positive effects from prayer; most studies do not show any effects from prayer; and a few studies even found that prayer worsened the subjects' medical outcomes [source: Carey].
A study on intercessory prayer is usually constructed like this: A population of patients with similar medical problems is divided into two groups. Strangers from a church pray for one group and don't pray for the other. After a certain amount of time, the researchers measure how the prayer group is faring as compared to the nonprayer group.
Research has been conducted on AIDS patients, people with blood infection and people recovering from coronary artery bypass graft surgery, among many other types of illness. The coronary bypass study is notable because it divided patients into three groups: no prayers given but the patients didn't know whether they were receiving prayers, prayers given but the patients didn't know whether they were receiving prayers, and prayers given that the patients were aware of. The patients who were prayed for and knew it suffered a higher rate of complications following the surgery than the other groups. The researchers supposed that the knowledge of extra spiritual care may have either made them worry that their condition was severe (why else would they need prayers?), or that it put pressure on them to respond to the prayers, causing anxiety, which is known to be detrimental to health [sources: Benson; Schneiderman; Carey].
The real problem with studying intercessory prayer becomes clear when you look closely at the methods of the studies. Some suffer from common issues with statistical studies: small sample sizes, unclear measurement of health outcomes and the sharpshooter effect (in which so many possible outcomes are considered that it's inevitable the researchers will find a result among them, like drawing the bull's-eye around the bullet holes after you've already fired).