This is a copy of a post from my old blog that I’m no longer update
Purpose: Show the relationship between obesity and car crashes by state, by year
Data retrieved: 06/30/15
Tools used: R (ggplot2, plyr)
Script and Data zip: 615_obesity_fatal_car
Summary: There is evidence that obese individuals are at greater risk of fatality in car accidents. What can we learn looking at the relationship between obesity and fatalities in general on a state by state basis?
The obesity data shows state by state obesity rates and the accident data gives deaths per million people. These data are for 2005-2013.
First, let’s look at car accident death rates over time.
Car accident deaths have been going down for a while so this isn’t totally novel, but still good to see!
What about obesity rates?
Another clear relationship, though doctors might be a bit more concerned!
Because obesity is increasing and car accident deaths are decreasing by year, it is almost necessary to look at these data year-by-year. These things can be corrected for, but this is simpler (though it can raise other questions!).
Let’s look at a single year before getting the correlations for each year.
The higher the state obesity rate the more fatal car crashes! Well that’s 2013, let’s at least get the correlations for the others.
So in this plot each point is a correlation between obesity and deaths for all states in a given year. I think this is particularly noteworthy. If you recall, obesity rates are going up but car fatalities are going down overall. Could this mean that despite cars getting safer overall, the safety features that are implemented do no benefit obese people as much as non-obese counterparts? These data are all correlational so I’ll happily speculate, but there is much more that needs to be done before I believe that. Regardless, every year there is a fairly strong relationship between obesity and car deaths.
NOTES AND CAVEATS:
1. Because this is state by state, the strong relationship could be due to State differences in car fatalities. I didn’t explore this but would be happy if someone was interested enough to dig deeper!
2. In the downloadable dataset and R-code I included plots for each year’s correlation.
3. Correlation does not equal causation. Correlation does not equal causation. Correlation does not equal causation. ad infinitum
4. Photo credit: