News for hackers. Stuff that matters.
Re:More than 73% of American adults are overweight or obese by viraptor on Tuesday December 29, 2020 @03:51PM (#25574818)
The measurements don't have to be precise to show a trend, relation, or infer a number of valid results. If someone calls the study into question based on using imperfect methods, rather than using those methods incorrectly... That's their opinion, but it's not useful.
Re:More than 73% of American adults are overweight or obese by kbenson on Tuesday December 29, 2020 @04:23PM (#25575213)
They don't have to be precise to show a trend, but there needs to be some assurance that the measurements affects subgroups similarly and a measurement of how it affects those subgroups to infer a number of valid results.

How does BMI accuracy scale with age? How does it scale with different ethnic groups? How do these interact? How do you infer an accurate number without knowing these variables and accounting for them and the relative sized of those cohorts?

If the measurements to create the BMI were taken from a statistically valid cross section of the current makeup of Americans, then I would have no problem. But how was the BMI generated? When? With what type of participants?

This article itself notes that Asians experience relatively more problems at a lower BMI. Without knowing what the BMI is tuned to, how an we interpret how it applies to a cross section of Americans for a singular study?

As a measurement of similar studies seeing a change over time, I agree it's less problematic (but could still be affected by trends in physiology over time, possibly from immigration).