You're getting it wrong, it's that people who get in, but don't go, have similar outcomes. That suggests that a lot of the value of the signal is in the inital filter, which seems like what you're saying.