Being oveweight as a function of region & race

Tyler Cowen linked to a Time article on the phenomenon of Southern Americans being relatively overweight vis-a-vis Americans from other regions of the country. Several reasons are offered, from the lower per capita income of Southern states, to the fact that Southern food tends to be fried and less healthful. But the article doesn't mention one very salient fact: black Americans are heavier than white Americans, and are disproportionately concentrated in Southern states. What is a regional disparity could be accounted for by underlying differences in the distribution of races.

State Health Facts reports that 70% of African Americans are overweight, vs. 60% of white Americans. Using state-by-state data one can see how accurate assessments of interregional variation are when you control for race. The chart below shows the relationship between the proportion of whites who are overweight to the total population in each state*.

i-93779e625de1c298e2a4e0a33fc29e62-totalobesityscatter.png

As you can see, there's a relatively close relationship between white overweight rates and total overweight rates. This makes sense since ~2/3 of the latter is composed of the former. There are many states, such as New Hampshire, where the white (non-Hispanic) rate for any given statistic is an excellent approximation of the total rate. On the other hand, there are states such as Hawaii or Mississippi where the white rate for a statistic shouldn't necessarily be used as a proxy because of a large non-white population. I say not necessarily because there is likely to be some correlation between white and non-white values in a given state (this is evident in the South where the whites exhibit social views closest to black Americans when set against other regions of the nation). So, what is the correlation between black and white rates of being overweight by state? In other words, do black and white Americans vary in weight by state in a similar manner, or are they totally uncorrelated?

Again, I looked for data at State Health Facts. Because of the sample size for blacks was very small in many states I eliminated those (the cut off was determined by State Health Facts, with those states being left empty). Additionally I removed Washington D.C. from the data because whites and blacks exhibit such a huge difference that this is an outlier. Only 38% of whites in the District of Columbia are overweight, while 66% of blacks are. This is probably due to the fact that D.C. isn't equivalent to a state, but is a segregated city. The issue of size of course distorts regional assessments on a state-by-state basis, as California or Texas probably have more variation within them than many groups of states (e.g., the Upper Midwest states for example), so the results below should be interpreted cautiously. In any case:

i-3ed0805fd592c36cd98881d19664475f-blackwhiteobesitymap.png

33% of the black overweight rate variation between states can be predicted by the white overweight rate variation. There is some general underlying commonality then even when you account for race by region which causes disparate rates of being overweight. Obviously blacks and whites in the South both share the experience of living in the same climatic regime, and also aspects of their culinary traditions.

Finally, there are obviously other variables. How about looking at the % in a state who are "low income," defined as no more than 200% of the federal poverty line.

i-a5b58fed362b0a01b4950a4f0333d2bf-povobesity.png

Only 17% of the variance in being overweight can be explained by the proportion who are low income when it comes to federal criteria. Since this doesn't take into account huge national differences of cost of living, I suspect that these data probably exaggerate the real variance of weight explained by income. I suspect that 17% is higher than it would be if these factors were taken into account.

Finally, a table which shows the states where there are enough blacks to be counted for in the data, and the black-white difference in proportion who are overweight. Note that in Wisconsin the black and white overweight rates are almost matched.


Total White Black Ratio
Alabama 67.9% 62.2% 75.8% 1.22
Arkansas 65.6% 62.6% 71.1% 1.14
California 61.3% 56.9% 68.8% 1.21
Colorado 59.7% 51.6% 60.0% 1.16
Connecticut 59.7% 56.5% 65.2% 1.15
Delaware 63.6% 60.9% 72.0% 1.18
Florida 60.1% 56.0% 73.9% 1.32
Georgia 64.6% 60.0% 71.9% 1.2
Illinois 63.2% 58.7% 72.6% 1.24
Indiana 63.5% 61.4% 68.3% 1.11
Kansas 65.5% 63.2% 74.2% 1.18
Kentucky 66.6% 64.7% 73.7% 1.14
Louisiana 63.6% 59.4% 65.1% 1.1
Maryland 63.3% 58.7% 71.3% 1.22
Massachusetts 58.0% 55.2% 63.2% 1.15
Michigan 64.6% 61.9% 69.6% 1.13
Mississippi 67.4% 62.4% 71.3% 1.14
Missouri 65.4% 63.3% 73.7% 1.16
New Jersey 62.0% 57.0% 69.9% 1.23
New York 60.2% 58.5% 63.9% 1.09
North Carolina 65.7% 61.3% 72.3% 1.18
Ohio 63.3% 61.1% 67.6% 1.11
Oklahoma 66.4% 63.3% 70.0% 1.11
Pennsylvania 64.2% 61.0% 72.8% 1.19
Rhode Island 59.9% 57.5% 69.4% 1.21
South Carolina 65.8% 60.5% 70.5% 1.16
Tennessee 67.9% 63.5% 73.2% 1.15
Texas 66.1% 59.7% 68.5% 1.15
Virginia 61.5% 60.3% 66.4% 1.1
Washington 61.8% 59.8% 65.7% 1.1
Wisconsin 63.5% 61.3% 62.1% 1.01

* I mislabeled the as "obesity" what should have been "overweight," just just replace appropriately.

More like this

The stats don't take into account the greater muscle mass of blacks; it says on the site they used BMI's of 25.

Weren't there a lot of issues with this (Blacks get overweight at higher, and East Asians lower, BMI's than whites) so that the stats would be different if this was corrected for?

Weren't there a lot of issues with this (Blacks get overweight at higher, and East Asians lower, BMI's than whites) so that the stats would be different if this was corrected for?

yes. you want to do the correction? the data is there. just send me the file and i'll post an updated with corrected proportions.

Do blacks have greater muscle mass on average? Is that upper or lower body, or both? How come it's always white nordic guys in the worlds strongest man competition? Previously I've heard nordic whites have more upper body mass -- no idea if that's true though.

If we consider the possibility that the standard medical definition of overweight (BMI 25 to <30) confers no increased overall health risk, then we should be careful as to where we set the threshold for measuring obesity. It's not so clear when increased mortality attributable to weight begins. It seems that weight has a protective effects against death from accidents or certain illnesses. Of course, at a certain point, the increased risks of diabetes, MI, stroke, etc. supercede any advantages.

At the risk of being a stinker, have you looked at the interaction terms (ie., ANOVA & ANCOVA). I'm thinking they would be very illuminating.

I am not sure that this can get anywhere - the strongest determinant of weight is body type and genetic history - we continue to try and "blame" large people for being large -Kenyans and Etheopian marathon runners are not 5' 6" and 140 pounds because of anything they do... Nor are Samoan adult males 6' plus and over 280 pounds because of the number of McDonalds on their islands.

This crap is both silly and dangerous...

By Lonely Libertarian (not verified) on 26 Jul 2009 #permalink

I read an article showing a positive correlation between weight and income for very poor people, which is inverted from some level of income and further. So, if that article is right, then "% who are low income" is not a good measure, overlapping two contrary effects.

the strongest determinant of weight is body type and genetic history - we continue to try and "blame" large people for being large -Kenyans and Etheopian marathon runners are not 5' 6" and 140 pounds because of anything they do... Nor are Samoan adult males 6' plus and over 280 pounds because of the number of McDonalds on their islands.

this is dumb. just because something is strongly heritable doesn't mean environment doesn't matter.

At the risk of being a stinker, have you looked at the interaction terms (ie., ANOVA & ANCOVA). I'm thinking they would be very illuminating.

i would prefer to do that in R. perhaps i will at some point since people are interested.

Interesting that blacks are reported as having slightly greater muscle mass and lower body fat than a caucasian of a given weight on average, yet blacks nevertheless suffer from diseases which are increasingly commonly described as linked to weight at higher rates, sometimes at significantly higher rates. If the issue is weight and its impact on health and health care costs, it doesn't appear that BMI variations matter all that much.

BMI is a useful population level number. you have to control for population. the "natural" median BMI for europeans may be higher than for asians, for example. whether it has that much individual level information, well, there's a debate about that (depends on how much statistics you think a typical doctor understands enough to transmit to a patient).

I don't have data on this, but my impression is that Southern cities have more fast food restaurants. Folks love the cheeseburgers and fried chicken.

There might be errors in the data. Colorado's total obesity rate is about 55% in the last plot, but the table at the end has it at about 60%.

we have fat poor people in the usa. poor people world wide are actually poor. our poor have cell phones and massive amounts of food. how great is the usa? seriously.