Uecker, Regnerus, and Vaaler just published a very interesting study examining the social sources of young people in the U.S. losing their religion. They analyze data from the Add Health data set, and they test if leaving a faith is related to religious affiliation, educational attainment, family formation, or various behaviors. They studied 10,000+ kids who were interviewed twice seven years a part. They were labeled as dropping out of a religion if they said they affiliated with it in the first interview but not in the second interview (actually, wave III).
The average number of "drop-outs" for the whole sample was 17%. (Note, there were also many who went from no religion to a religion, but they are not the focus on this article).
Here are the drop-out rates by denominational/ religion standing.
Clearly there is a lot of variation across religious groups. Why do you think that is?