As a recent study claiming Obama lost up to 5 percent of the vote in 2008 because of his race is an innovative way to rely on Google search results to measure what people won't tell pollsters, but it also suggests some limitations of using search engines to get inside people's heads. Researcher Seth Stephens-Davidowitz looked at how often the n-word was searched for in Google, using that as a proxy for racism, and then compared it to that area's change in democratic vote shares between the 2004 and 2008 election. Using that method, he found that areas with high racial animus had a higher vote for Kerry than Obama. "A one standard deviation increase in an area’s racially charged search is associated with a 1.5 percentage point decrease in Barack Obama’s vote share, controlling for John Kerry’s vote share," he wrote. That seems to make sense, but then there seem to be a few problems.
Google Searches Don't Have an Age Requirement Only those 18 and older can vote. Googling, on the other hand, anyone can do. Let's say there's an entire family of racists googling the n-word willy-nilly. They may contribute to the "racial animus" metric that Stephen-Davidowitz measures, but not all those votes count.
Google Searches Are Not Limited to Amount of Times The same racist can search the n-word as many times as they like. A really active racist might make an area look more racist than it actually would be in terms of votes.
Google Searches for the N-Word Don't Reveal Latent Racism Using the n-word as a metric for racial animus is being kind to all the racists out there. There are people who would never vote Obama because of the color of his skin, but also know enough to never use the n-word, even in the privacy of their web browser. We suspect racial animus to be even deeper than those who just Google the slur.
People Using the N-Word Aren't Likely to Vote Obama Anyway And while the Google test might serve as a good metric for evaluating racial animus in a community, the people looking up the n-word probably aren't the type who would vote Democrat in the first place.
The N-Word Isn't Always Associated with Racism Some might also aruge, that not everyone who searches the n-word has racist intentions. Some might search the n-word for other reasons. Like, kids wondering what it means, academics, those looking for Hip Hop lyrics, or, say, people discussing Huck Finn -- at least for now. Stephen-Davidowitz actually anticipated for this, he told us, controlling for terms he thought might be searched along with the n-word, like joke or "nigga."
As for the other issues, Stephen-Davidowitz wrote, "It is a proxy for an area's racial animus," and not a measure of individual behavior. "It is based on the hypothesis that individuals with higher racial animus will be more likely to type the racial epithet into Google." Meaning, the search numbers does not (and cannot) look at these individual behaviors, but rather, relies on an aggregate. "There are definitely things you can't do with area-level data that you can do with individual-level data," he added.
Stephens-Davidowitz also told The Wall Street Journal, "The idea for using Google in this way was based on a number of papers studying the conditions under which individuals discuss social taboos. Google searches provide a useful database for attitudes not easily accessed by surveys." We bet people are more willing to express themselves on Google than admit to being racists in an exit poll. But, alas, it's hard to measure these types of things.