Re:Further reading... (Score 1)

Don't just read the first page! :)

If you read further you will see that this original study was deeply flawed.

Besides only having a pitiful sample size of 20 (10 in each treatment group) the study originally intended to look at mortality. However only 1 person died in the period of the study. They then unblinded the data to search for data to confirm their hypothesis - something that the journal was not told when they submitted their manuscript.

This sort of thing is a major statistical no-no. Testing, and retesting again and again alters the bounds of "statistical significance" or alpha-value.

"I'm even more troubled by the multiple endpoints than the unblinding," he said with increasing concern. "It's a little post hoc. Normally, we accept the standard that a finding must have less than a 1-in-20 chance of randomly occurring. When you're on your third or fourth attempt, it's much more likely a 1-in-20 event will occur, so the standard has to be higher. You divide the alpha by the number of attempts, thus 1 in 60, 1 in 80, et cetera. There was no indication of this recast standard."

Furthermore, the data was confounded by age -

Most of the 20 participants were in their mid-twenties to early thirties, but four were older. Three in their late thirties, and one in his sixties. Those oldest four patients died. They were all in the control group. In other words,

the study provided fairly convincing evidence that if you had AIDS back in the mid-1990s, the older you were the more likely you were to die.

Lastly, the prayers did not do her any good did they? She died of her cancer (after being urged by her psychic buddy to cease radiation treatment).