Okay Minnesota, another day of having our state government closed and we have no word on when budget talks are going to resume between Governor Dayton and the GOP-controlled state legislature. So, let’s move over to another topic–the health of the people of Minnesota.
And no, I am not going to talk about health just because I have been feeling lousy the past couple of days. Instead, I ran across this news story on obesity, which I found quite alarming.
Super size my heart attack, please.
So, Minnesota ranks 36th in obesity. Well, you might say: “that doesn’t sound so bad, we’re better than average.” The fact that we could put some faith in an obviously misleading ranking like this is alarming–all 50 of the states are failing in this regard so being the 36th worst failure is not something to be proud of. To be fair to Associated Press, the above article does allude that this statistic may be misleading when viewed in a historical context: “In 1995, no state had an obesity rate above 20 percent. Now, all but one does.”
For something really shocking, take a look at the obesity trends by state from the Centers for Disease Control and Prevention: http://www.cdc.gov/obesity/data/trends.html#State and watch the color of the states change in a very scary way.
Ronald McDonald: Scaring kids and cardiologists since 1940. (Photo: funnyrandompicture.com)
So, what is to blame for all of this? I decided to try to construct a regression model to predict state obesity rates. To all you statisticians out there, I would like to add that this analysis is meant to be exploratory and not necessarily explanatory (so assume that I have stated all the typical disclaimers about unit of analysis, model specification, etc). Okay, having said that, let me show you what I found in my 10 minute back-of-the-napkin analysis.
First off, does the number of McDonalds Restaurants per 100,000 people in a state predict the state’s obesity rates? Using Steven Graves’ data (thanks Steven!) for McDonald’s Restaurant data and CDC’s state-level data on obesity, the following model emerged:
The line goes up--which supports my argument.
Without getting into all of the gory statistical details of the analysis, the model indicates that for every increase of 1 McDonalds Restaurant per 100,000 people, the state’s obesity rate goes up by 1.29% (p-value = 0.054, so the level of confidence = 94.6%). The r-squared value of 0.075 means that 7.5% of the variability in state obesity rates can be attributed to proliferation of McDonalds Restaurants.
That’s nice, but I know what you are thinking–McDonalds is just one fast food restaurant among many. So, thanks to Prevention.com I also have the total number of all fast food restaurants per 100,000 people in each state. So I regressed obesity rates against the number of fast-food restaurants I got the following:
The line goes up even more this time...what did you expect?
Once again, without boring you too much with all the statistical details, the slope of the above line is 0.89, which means that every increase of one fast food restaurant per 100,000 increases the state obesity rate by 0.89% (p-value < 0.000, which translates into a level of confidence > 0.999%). Also telling is the r-squared value = 0.294, which means that 29.4% of the variation in a state’s obesity rate can be attributed to the proliferation of fast food restaurants in the state.
Now I’m not saying that we need to get rid of fast food restaurants, but I think we need to realize that fast food is an important indicator of obesity, both in Minnesota and nation-wide. By being aware of this, states can craft common-sense health awareness programs (and more controversially, regulations) to help reverse the obesity trend that is gripping our states and our nation.
This pi is not on the dollar-menu at McDonalds. (Image: tomdukich.com)
