There have been quite a few such studies over the years, with different results. Aside from the article from Duke itself, I think it's important to read through comments posted to this discussion. Can we control god through his obligation to answer prayers? Will god refuse to answer an honest prayer because he's being watched by scientists? (Several contributors complained that the test was unscientific, as we are instructed not to test god, and that therefore the results would presumably be skewed by god being purposefully contrary.) Is god distinguishable from background statistical noise, or is this a hit-and-miss fallacy (that is, counting the hits and not counting the misses?)