Crticial Index

A few months ago, I missed Columbia University's Andrew Gelman creation of a new, amusing, and astonishingly perceptive polimetric index: the Starbucks-Walmart Ratio. Take the number of Starbucks per capita in a state and compare that to the number of Wal-Marts per capita in a state. The results, color-coded by Garrett Dash Nelson at Legion and illustrated to the left, demonstrate a strong partisan correlation.

Briefly, it seems that the easier it is for you to get to a Wal-Mart, the more likely you were to cast your ballot for Bush in 2004. The easier it is for you to get to a Starbucks, the more likely it was that you would have preferred Kerry. This conforms nicely to the urban-rural split that county voting maps reported.

In fact, the correlation is stronger when you consider Wal-Mart density by itself. When you fall below 11 Wal-Marts for every million people in a state, only Nevada votes Republican (and that only barely). Above that ratio, very few states vote Democratic.

More research is clearly needed. It is not well understood why both Starbucks-dense Colorado and Wal-Mart deprived Nevada nevertheless voted Republican. (Perhaps in Nevada, Costco serves as a Wal-Mart surrogate?) New Hampshire's abberational voting behavior -- there are very few Starbucks and lots of Wal-Marts there -- is also worthy of closer examination. But overall, the trend is clear, and Starbucks closing one in twenty of its stores while Wal-Marts continue to spread like fungus across the map is an encouraging sign for John McCain. So remember, folks, correlation is not causation, and people of any political affiliation can consume the way all good Americans should.