I also didn't find a link to the study. I found out that Matthew Feinstein is a medical student. Given my experience with such medical studies and their use of statistics, I would be very surprised if their empirical methodologies are sound. Maybe that's also why he seem to stress his result much.
Although this is along the sentiment that 'all generalizations are bad,' it has been my nearly life-long experience that most church-going folks are losers, socially-speaking.