Thursday, December 18, 2014

Lies, Damned Lies, and Statistics

In a sustainable agriculture forum I follow on Linkedin, a farmer questioned whether food deserts actually exist and suggested that the problem may be exaggerated. The statistics he cited suggested, for example, that it takes poor people in low-access areas only 4.5 minutes longer to get to the grocery store than it takes wealthy people in well-served neighborhoods.

"Using the American Time Use Survey (ATUS), researchers found individuals in low-income low access areas spend an additional 4.5 minutes traveling to the grocers over the national average and average a trip every 8 days over the national average of 7 days."


The problem with this is that, as is typical in these kinds of studies, it's using averages. In most higher-income, higher access areas, almost all individuals will be traveling by the same means: privately-owned automobiles. This means respondents from such areas will all fall within a very narrow statistical band; whether they make $30K a year or $300K a year, their car still gets them to the grocery store in the same amount of time.

In poor, low-access areas, though, transportation methods used will be more diverse. Some will have automobiles. Some with a little money might take the bus. Those with slightly less money might ride a bicycle. Many will walk. Of the walkers, some will have backpacks, carts, or wagons; others will be able to carry only the grocery bags they can hold in their hands. Elderly and disabled people might pay for taxis or use publicly funded or non-profit taxi services, or they may rely on friends or relatives with cars to give them a ride. The statistical result is that you're going to have a really wide spectrum, with people on one end starving unless they walk several miles every day, and people on the other end whose numbers match those in the high-income neighborhoods. Neither the mean nor the median in that set will give us a clue of what the people at the bottom are going through unless we also know the percentage of households in the poor neighborhoods that have on-demand access to a functioning automobile.

No comments:

Post a Comment