Without debating the actual factual/statistical merits of the claims made in this article and the paper it reflects, would your prognostication have predicted that a somewhat higher than “desirable” BMI would have a lower death rate?
Actually, it occurs to me that I’m not aware of any well-conducted study that generated a BMI vs. lifespan chart, such as you describe, much less capable of addressing arbitrary distinct “groups and subgroups” vs. the general population. I’d estimate that such a study would need a sample size in the high hundreds of thousands (likely millions, to attempt accurate prediction for arbitrary subgroups).
Since I’m not aware of such a large study for BMI vs. lifespan, I’d really appreciate a link or citation for my own education. It’s something I should know.
Or am I being whooshed? You did only say you can make “reasonable predictions” – anyone can make “reasonable” [seeming] predictions. Accurate predictions, as proven by verification, are what most of us would really want. each of those verifications, testing a prediction for a [distinct, non-random] subgroup against a an actual 10K-sample data set) would be a pretty sizable undertaking by itself.
I’m not being picky. I just didn’t know there were such studies, and I feel I should.