Misleading Statistics

Analysis of “Road Hogs”

In a June 15, 2011 post on their website, The Economist published a chart purporting to show a causative link between driving and obesity in the United States (The Economist). Their post was based on a recent paper titled, “A Note on the Relationship Between Obesity and Driving,” in which two members of the University of Illinois Department of Computer Science and a member of the University of Illinois Department of Industrial and Enterprise Systems Engineering state that, “Vehicle travel rates and obesity in the United States have surged in recent decades.” The abstract continues, stating, “This paper...draw[s] attention to a very close relationship between trends in miles driven per licensed driver and adult obesity rates six years later.” (Jacobson)

The authors present data for Vehicle Miles Traveled (VMT), Licensed Drivers (LDs), Adult Population and Gross Domestic Product (GDP) in the United States for the period of 1985-2007. Additionally, the obesity rate in the United States is included for the period between 1995 and 2007, as obesity rates prior to 1995 were not available. In order to provide evidence supporting their data for a lag of six years between a change in VMT/LD (Vehicle Miles Traveled/Licensed Drivers), the authors cite a 2008 paper that purports to show that weight changes become permanent of 2000 days. The authors show a regression analysis with a near perfect correlation and an even better Coefficient of Determination (R2) of 98.44% for their claim that a 1% decrease in VMT causes a 0.8% decrease in obesity rates (or vice versa) (Jacobson).

There are several misleading ideas and either lying with statistics or massaging statistics to fit the outcome desired. Perhaps the most egregious error – and one that is compounded later – is using the six-year lag between VMT/LD and obesity rates. Jacobson, et al. appear to admit on page 2 that they modified their approach when no correlation was demonstrated...