There are two things I would say to that.
One is—and this is one of the challenges you always have in terms of statistical analysis of language mastery—that it tends to be self-identification. So the numbers that have always provided the basis for the analysis have been the degree to which people at that age self-identify as bilingual. Every now and then, there are tests that show that people either aren't as bilingual as they think they are, or are more bilingual than they say they are. It's a challenge in knowing exactly what we're talking about.
On one comment you made, challenging the idea that immersion is the only way to be bilingual, I didn't go to immersion. I'm a product of core French, as are many of my equally bilingual colleagues.