I suppose the reason why some bristle at your partitioning of American obesity rates into racial groups is that it's unclear that it's particularly relevant. Obesity varies by income, urban/suburban/rural-ness, red state/blue state, and more.
Any of these dimensions could be used to partition America into tranches. Then, throw out the bottom tranches. Then, say that America isn't especially unhealthy.
Imagine person A says: There's an epidemic of gun violence in America.
Person B responds: No. Just ignore the urban gang violence, the mentally ill mass shooters, and the suicides. It turns out that among mentally stable Whites, America isn't that violent.