Right, there will be a correlation, but that's not the point. The point is that class as a concept exists very much outside of race.
You can easily read the essay pretending that everybody in the US who isn't white simply doesn't exist, and it still makes completes sense.
You may just be getting numbers back based on the ethnicity that predominantly lives in your state.
Look at relative poverty numbers. i.e., of the number of people who are of ethnicity X, how many are poor? I think you'll surely find a correlation that OP describes as a decent first-order approximation.
For example: white 9%, hispanic 18%, black 37%. That's 2x and 4x more likely to be poor vs white depending on if you're hispanic or black.
[0] http://kff.org/other/state-indicator/poverty-rate-by-raceeth...