4 ms·
> How do they know members of the control group weren't being prayed for? Randomization should balance confounders like that: there should be on average an eq
by gwern 12y ago
> How do they know members of the control group weren't being prayed for?
Randomization should balance confounders like that: there should be on average an equal number of pre-prayed-for people in both the experimental and control groups.
(And even if all subjects had been prayed for, it seems reasonable that prayer would be additive and one would still expect to detect an average treatment effect.)