BOB GARFIELD:

With election season in full flower, swarms of pollsters have emerged like bees to gauge the fluctuating preferences of American voters. But there are some crucial questions to which pollsters are unlikely to get honest answers. For instance, how much did race factor into your vote for President in 2008.

Seth Stephens-Davidowitz, a  PhD  candidate at Harvard, has found an ingenious way to plumb America's seemingly impenetrable psyche. He sifts through people's Google search results. Just a quick heads up. Davidowitz uses the N word in the explanation of his research.

SETH STEPHENS-DAVIDOWITZ:

We are able to see how often a word is searched over time and in different areas. So basically, I used the most salient racially charged word, nigger, and I’m able to see in what areas of the country this word is searched more often than in other areas of the country.

BOB GARFIELD:

And you found out that there were significant differences in different places around the U.S.

SETH STEPHENS-DAVIDOWITZ:

Yeah, that's exactly correct. West Virginia, for example, the state that searched it the most often, is gonna search it more than three times as often as Utah, the state that searches it the least often.

BOB GARFIELD:

Now, when you say the most, West Virginia’s a small state. You mean the most as a percentage of, of population?

SETH STEPHENS-DAVIDOWITZ:

As a percentage of Google searches.

BOB GARFIELD:

So how do you know when that word is used in the search that it implies some sort of racial animus and isn't a search for a song or for a book report on Huckleberry Finn, or what have you?

SETH STEPHENS-DAVIDOWITZ:

Yeah. When I initially started the research I was surprised ‘cause I thought that it would be mostly searches for rap lyrics. Rap songs tend to use “nigga,” the word ending in “A,” which Google considers a separate search from the epithet ending in “ER”. So I actually used that as a control in my paper.

Also, I'm not saying that everybody who searches this, you know, I've seen into their soul and they deep down harbor racial animus. That’s certainly not the case, and there are many perfectly benign motivations of searches of the word. But the idea is that individuals with racial animus are on average going to be more likely to search the term.

BOB GARFIELD:

You mentioned that the lowest incidence of the search for that word was in Utah, which is as lily white a state as we have in this Union.

SETH STEPHENS-DAVIDOWITZ:

People were a little surprised to see Utah in the bottom spot. It’s actually not inconsistent with some of the other imperfect measures of racial animus. Utah recently has tended to score very low. One might also suspect, if we were going to go based on stereotypes, that individuals in Utah are less likely to use strident language in Google. So in my paper I control for profane language, and it doesn't really change the results.

There is a theory that a number of sociologists and psychologists have developed called racial threat, which is this idea that racial animus can increase the more African- Americans you have in the community.

I actually found in the data a somewhat more subtle relationship. When there are few African-Americans, such as an area like Utah or Boise, Idaho, racial animus tends to be very low.

But then as you have a very high percentage of African-Americans, racial animus, again, gets very low. So I found that racial animus hits its maximum point when an area is between 20 and 30 percent black, which is not inconsistent with some of the work of other psychologists and sociologists.

BOB GARFIELD:

So you ran all of this data, and there was lots of it, through your social sciences statistical extruding machine, and you were able to come up with an actual value for what racism cost Obama in the 2008 election. It was pretty striking.

SETH STEPHENS-DAVIDOWITZ:

Yeah, about three to five percentage points of the national popular vote Obama lost. One way to think about it is that racial animus gave Obama's opponent John McCain the equivalent of a home state advantage countrywide. So political economists have kind of detected how much advantage a candidate gets from being from a certain home state, and that tends to be about four percentage points of the national popular vote.

BOB GARFIELD:

Now, political polling has always been problematic. It would be great if mining Google or other aspects of social media could replace polling, but does your methodology really expand beyond a few extremely particular situations where honest self-reporting just doesn't take place?

SETH STEPHENS-DAVIDOWITZ:

I don't think polls have to worry about going out of business or anything. There are obviously a lot of advantages to surveys for a number of questions.

I, I think Google data will prove useful in a number of areas, besides just racial animus.

BOB GARFIELD:

Give me a for instance.

SETH STEPHENS-DAVIDOWITZ:

Health will prove a big area. Researchers have used Google data to detect influenza by area. Some scholars say that we can't trust anything in self-reported health data. I think that might be a little extreme, but I believe Google data will - prove very helpful on those research topics.

BOB GARFIELD:

I've got to ask, where are the communities wherein racial animus most resides?

SETH STEPHENS-DAVIDOWITZ:

West Virginia. Most of the areas of West Virginia tend to be very high on racial animus. Parts of western Pennsylvania and eastern Ohio, southern Mississippi, upstate New York, rural Illinois, as well.

BOB GARFIELD:

And apart from Utah, where are the results most benign?

SETH STEPHENS-DAVIDOWITZ:

Laredo, Texas is a largely Hispanic area in Texas. Boise, Idaho scores low, parts of Montana, San Francisco, California, Washington, D.C., Honolulu, Hawaii, Obama's place of birth.

BOB GARFIELD:

Ha, that’s what you say. Show me the birth certificate.

SETH STEPHENS-DAVIDOWITZ:

“Obama birth certificate” is another search term that yields some interesting results. [LAUGHS]

BOB GARFIELD:

Seth, thank you very much.

SETH STEPHENS-DAVIDOWITZ:

Thanks for having me, Bob. I, I really appreciate this.

BOB GARFIELD:

Seth Stephens-Davidowitz is a PhD candidate in economics at Harvard.