Well, I would argue that what they know by the time they get to college has a lot to do with high-stakes assessment, in math and every other subject. We don't teach students anything permanent when their assessments are so arbitrary and ineffective.
I'd have to look into the literature, but I'd say that students that are taught well with more conceptual methods do have better retention of concepts. In fact, I'm sure that's what the literature says because that's what my peers in math education are championing lately. The problem, as always, is in translating what we know works from research into what is actually achieved in the classroom environment.