[morganvachon]

Back when I was going through cancer, my weight dropped to around 130 pounds on a 6'4", large build frame. Obviously, I looked like Death himself and I was seriously underweight. However, even when I got back up to my normal 210 pounds and I still looked a bit too skinny, the chart showed me as being "overweight" at a BMI of 26.

So I'd take that "overweight" indicator with a grain of salt. If this woman is on the tall side or has broad shoulders and hips, that BMI of 26 could mean she appears athletic or even skinny, and she can certainly be healthy with that misleading number.

[CaveTech]

BMI isn't perfect, but it takes height into account. You can actually work back words with her weight and BMI to calculate that her height is less than 5'1".

BMI is pretty accurate most of the time unless you're an athlete. Women putting on far less muscle on average than men, it's far more likely that she was actually overweight...

[Symmetry]

BMI is pretty accurate for heights that were common in 19th century Belgium. It assumes that people's weights ought to go up as the square of their height but as Galileo pointed out back in the day you would expect a person's weight to go up as the cube of their height.

This means that tall people will tend to have higher BMIs all things being equal and short people lower BMIs. Which actually suggests that the woman might have been overweight if she was only 5'1" despite what her BMI said.

[dragonwriter]

> BMI is pretty accurate for heights that were common in 19th century Belgium. It assumes that people's weights ought to go up as the square of their height but as Galileo pointed out back in the day you would expect a person's weight to go up as the cube of their height.

My understanding is that BMI doesn't "assume" anything about what weight "should" do, its instead a measure which has been empirically shown to have a useful (though rather loose) correlation to risks related to various health conditions, particularly cardiovascular conditions. (Mostly, AIUI, it serves as a proxy for a host of other measures, which together are more accurate but more involved to gather.)

The cube-of-height scaling you refer to is what you would expect if taller people were exactly like smaller people but linearly scaled equally in all dimensions -- but that's neither empirically what taller people are built like nor are taller people whose weight relates to that of shorter people that way on average as healthy in the areas BMI is used a risk measure for as the shorter people.

[Mz]

My ex husband was career military. When he was on recruiting duty, every single year they had some annual conference. Every single year, there were insurance reps there or something. Everyone in the office would get told they were "overweight" according to the (civilian) insurance reps height and weight charts. All of these people had to a) pass their PT test and b) meet weight. In the military, if your weight is too high for the chart, they tape-test you to check if you are actually fat or not. Bodybuilders in the military often cannot meet the weight requirement, but they pass the tape-test with flying colors.

tldr: have an upvote for agreement.