Sunday, August 15, 2010

Correlations between obesity and driving, and others

GOOD Magazine recently published a table ranking all 50 states and Washington DC in terms of actual levels of driving, walking, cycling, mass transit usage, and levels of obesity. From their table, I plotted all 10 possible correlations of the variables with each other. Note that the driving rank and obesity rank is opposite that of the other things; i.e. a driving rank of 1 means little driving, while a walking rank of 1 means lots of walking. An obesity rank of 1 means low levels of obesity (in this data, Colorado). Our state is indicated with the extra large marker in these plots.

The strongest correlation is between walking and driving, and is depicted here. The more people drive, the less people walk.




The correlation between obesity ranking and driving is almost as strong. Places where people drive a lot have higher obesity rates. As the slogan goes, "driving makes you fat."
A weaker correlation exists between obesity levels and the amount of walking. Nevertheless, the less people walked, the fatter they got.
A stronger correlation than that with walking can be found with cycling. The less people ride, the fatter they got. Or alternatively, fat people rode less. I will have more to write about causation below.
Mass transit usage also is correlated with decreased levels of obesity. The less people used mass transit, the fatter they got.
Apparently places that bike less drive more, and vice versa.
Mass transit usage and driving and fairly strongly correlated, which makes sense in that many mass transit trips would be replaced with car trips if they weren't taken.
Walking and biking levels are fairly strongly correlated too. Such makes sense in that people who are physically more active would tend to both walk and cycle more.
Interestingly, the correlation between mass transit usage and walking levels almost does not exist. Neither does a correlation appear to exist in terms of mass transit usage and cycling levels. Could it be that relatively few people engage in intermodal transportation, and that the amount of walking to get to a transit stop is almost negligible?

Now, of course the nature of our 50 states + DC are very different with some being rural and others being completely urban. So the opportunities are different for people in different places. Even people within a state can be faced with very different environments, for instance a farmer in rural middle Tennessee lives in an environment very different from a city dweller in say, Memphis. Nevertheless, this data with regard to obesity levels seem to suggest that increased physical activity is beneficial in leading to lower obesity rates.

Or does it? As promised earlier, correlations can be very interesting, but they cannot by themselves prove causality. We cannot know from this data alone whether increased physical activity reduces obesity levels, or increased obesity decreases physical activity, or whether there might be a third or even fourth confounding factor at work. Nevertheless, an observed correlation is usually grounds for looking further into the issue to see if benefits could be derived. Our state currently ranks quite low on all the metrics examined; however, we are definitely improving rapidly. It would be interesting to see these plots five years from now!

No comments:

Post a Comment