The population of a specific Google office could be two std deviations from the mean. My office area, SoHo NYC, has a dramatically lower obesity rate than the USA average. Maybe 10-20% of the population is overweight, let alone obese, and itβs mostly the service workers.
That seems incompatible with the numbers that show something like 3/4 of the US population being overweight.
(And yeah, BMI isn't great, but it works pretty well with an aggregate population of sedentary individuals, which the population at large is.)