Jamelle Bouie wrote an interesting article over at The American Prospect this morning, Did Obama Lose Votes Because of His Race? In it, he reported on some very clever research by Harvard economist Seth Stephens-Davidowitz. He combined regional voting patterns from 2004 with regional Google searches of “racially-charged material.” By comparing how voting in different locations changed based upon the prevalence of these searches, he was able to discern a racist signature.
An example will help. Consider Denver, CO and Wheeling, WV. Both of these areas voted about 50% for John Kerry in 2004. Taking into account demographic changes, Denver voted as much for Obama as they had for Kerry. Denver, had a very low rate of “racially-charged material” searches. So that’s as it should be: little racial effect, little effect on voting. However, support for Obama in Wheeling was way less than it was for Kerry. What’s more: lots of “racially-charged material” searching in Wheeling. Do enough comparisons like this and you can figure out how many people were not voting for Obama because he was black.
Stephens-Davidowitz found that the effect in 2008 and 2012 was between 3 and 5 percentage points! Not excited? That’s just because you are one of those dull human beings who does not live and breathe statistics. Let’s assume that it is the lower number: 3 percentage points. That would mean that correcting for the bigot vote, Obama would have beat Romney 55%-45%: a ten percentage point defeat. What’s more Obama would have beat him by more than 12 million votes. (For the 5% number: 57%-43%, almost 18 million votes.)
Admittedly, there are enough caveats to this research to fill my new 2 TB hard drive. One that I don’t think matters, but I’ll bring up because it relates to my life: the meaning of “racially-charged material.” The truth is, I very often search for things that make me feel very uncomfortable. Now, I know that the nature of my research is always clear in the context of many searches. Just the same, I often feel queasy that people will see my search string in their Google Analytics. (For the record, it mostly has to do with Nazism and fascism, not racism; but I’m not keen on people thinking me a neo-Nazi.)
But assuming that the research is correct (in direction if not amount), Bouie notes that this means that 2016 looks very good indeed for the Democratic Party. And screw it, I’m going to take any good news I can get right now.