Did Racism Cost Obama Votes in 2008?

A new paper (PDF here) by Seth Stephens-Davidowitz, a Harvard Ph.D. economics student, attempts to measure whether “racial animus” cost Barack Obama votes in 2008. Using location-specific Google searches for racial epithets collected on Google Insights, and comparing Obama’s 2008 performance to John Kerry‘s in 2004, the study concludes that racism cost Obama 3 to 5 percentage points in the popular vote:

Traditional surveys struggle to capture socially unacceptable attitudes such as racial animus. This paper uses Google searches including racially charged language as a proxy for a local area’s racial animus. I use the Google-search proxy, available for roughly 200 media markets in the United States, to reassess the impact of racial attitudes on voting for a black candidate in the United States. I compare an area’s racially charged search volume to its votes for Barack Obama, the 2008 black Democratic presidential candidate, controlling for its votes for John Kerry, the 2004 white Democratic presidential candidate. Other studies using a similar empirical specification and standard state-level survey measures of racial attitudes yield little evidence that racial animus had a major impact in recent U.S. elections. Using the Google-search proxy, I find significant and robust effects in the 2008 presidential election. The estimates imply that racial animus in the United States cost Obama three to five percentage points in the national popular vote in the 2008 election.

To circumvent the problem of people under-reporting their own racist tendencies, Stephens-Davidowitz used the percentage of an area’s total Google searches that included the n-word as a proxy for an area’s level of racism. It turns out the word appeared as frequently as words like “charity,” “hispanic,” “nausea,” “sweater,” and “migraine(s).” 

The demographic factor most strongly correlated with racially charged searches is education level. Stephens-Davidowitz found that a 10 percentage-point increase in college graduates correlates with almost a one standard deviation decrease in racially charged search.

Searches for the n-word were most popular in West Virginia, upstate New York, rural Illinois, eastern Ohio, and southern Mississippi. They were least popular in Laredo, Tex. (a largely Hispanic market); Hawaii; parts of California; Utah; and urban Colorado.

Here’s a map of the results, with the darker colors representing areas with the most frequent searches for the term:

Search volume for the n-word from 2004-2007, at the media market level. Darker areas signify higher search volume. White areas signify media markets without data.

See page 8 of the study to see how the author addresses the issue that the n-word is a common reference in rap songs.

While Obama won 53.7 percent of votes in ’08, the study suggests he would have claimed between 56.7 and 58.7 percent if “the whole country had the racial attitudes of the most tolerant areas.” Stephens-Davidowitz concludes that racism gave John McCain “the equivalent of a home state advantage country-wide.”

(HT: Marginal Revolution)


Okay, please do the same for McCain (old, frail) and Palin (retarded) and see if the racism advantage McCain supposedly got still holds up.


I get your post, but in the spirit of taking everything seriously:

It's OK to not vote someone into office because you are concerned about them dying prematurely (old, frail) or because they are mentally handicapped (retarded). Not voting because someone is black (n-word) is a completely different level of social acceptability, because that factor has no impact on the ability of the individual to do the job (outside of things like assassination worries), while the previous two factors could have a real impact.


Let's do this same analysis for the Democratic Primary.


While racial animus may have costed Obama votes, a large fraction (more than 80% if I recall) of the Afro-American community voted for him. One study could find out he won because of and not despite the fact he was black...


Actually, 95% voted for him in 2008 as opposed to 88% of african americans who voted for Kerry in 2004. However there were also 2 million more african americans who voted in 2008 compared to 2004. This probably isn't enough to cancel out this "negative racism" factor, but is definitely enough that it should have been factored in. (Source: cnn and NYT election results websites)


Plenty of whites voted for Obama because of his race as well. Many of my white friends that voted for Obama considered his race a big plus, for a variety of reasons (conscious and unconscious). On balance, I would guess Obama's race led to a net gain of votes.


Looks like Lake Superior is pretty racist

Joshua Northey

As someone who grew up on the shore of Lake Superior I can tell you it is totally racist!

Elmer Fudd

That's right.....those Canadians just don't know how to behave......


Very poor thinking and methodology... This is the type of grasping at straws I would expect from a middle school student.

This is classic looking for an outcome and then bending data to fit your presumptions.


I think the entire study starts with bias. The first bias is that racism only had a negative influence. Where's the study showing the racist bias FOR Obama? I know of several people who voted for him specifically because he is black. Secondarily, when did the "N-word" become a barometer for racism? I guess that much of the black community must actually hate the black community if I take into account how often they use that term themselves.


Racism also enabled him to capture 96% of the African-American vote.

How much 'racial animus' did that cost McCain?

Nice try.

Jaime Velo

But how many white votes did he pick up from the "i'm not racist and I'll demonstrate it by voting for a black guy, that's how open-minded I am"-crowd? And btw, Obama is HALF white you know. Why do we call him black? In fact, you could argue he is more white than black given his upbringing.


The larger question here ought to be why this sort of information is even collected, or made available to any entity outside Google's search engine optimization team.


The real numbers have been doctored on every news site from the goverenment.

obama actually got about 16 million black votes as opposed to Kerry's 8 million

obama aslo got 30 million hispanic vote (many illegal) as compared to 11 milion for Kerry.

obama won because he was black.


I just wonder how the study accounts for the minorities who use the n-word but overwhelmingly supported his election.


This was controlled for in the study.


On the surface, there are so many factors that go into these things that its hard to believe the author is on to something. And if he is on to something, can any good come out of this study?

I can't help but think the results will come out differently in 2012. Then we'll say racism in voting doubled or vanished.

Cyril Morong

I recall that Obama got a higher % of the white vote than either Kerry (2004) or Gore (2000). It was close, only about 1 percentage point, in the low 40s.

"Clinton repeated the feat in 1996, but afterward, the Democratic performance among whites began to decline. As the party’s nominees became more liberal, the Clinton coalition slowly dissipated. Al Gore won only 44 percent of the two-party vote among whites nationally, while John Kerry won just 41 percent of the white vote. In 2008, for all the hype about Obama’s “broad” coalition, he only won 43 percent of the white vote, about two points better than Kerry. Obama’s win came almost entirely from turning out more minority voters, and doing better among them."

Gore's white % might be lower if Nader's votes are included. From



Cyril Morong

Todd Donovan of the Department of Political Science, Western Washington University published an article called "Obama and the White Vote" in 2010 in "Political Research Quarterly."

Here is the link



"This article draws on the racial threat thesis to test if white voters who lived in areas with larger African American populations were less receptive to Barack Obama in 2008. Racial context is found to structure white voters’ evaluations of Obama and, thus, affect where the Democrats gained presidential vote share over 2004. The overall Democratic swing was lower in states where a white Democrat (Hillary Clinton) had more appeal to white voters than Obama.
Obama increased the Democrats’ share of the white vote, but gains were associated with positive evaluations of Obama among white voters in places with smaller African American populations. The likelihood that a white voter supported Obama also decreased as the African American population of the respondent’s congressional district increased. The results are relevant to discussions of the future of the Voting Rights Act and to conceptions of a “postracial” America."

One passage reads

"...exit polls estimated white voters’ support for Obama was slightly greater than for John Kerry in 2004 or Al Gore in 2000."



Related question: how many votes did racism gain for him?


It's anecdotal I know, but I recall at least two conversations with friend who made the statement that they were voting for Obama because, in part, they "could not imagine voting against the first black President." Assuming this methodology is sound, I wonder, as I live in one of the highly-populated regions identified as "more tolerant," if there was a significant "reverse-racism" impact not just among blacks, as others here suggest, but among whites in those areas as well.

Joe J

Admittedly there is an inherent problem in trying to measure racism and it's effects if any. This seems to be more of a convoluted way than others. It assumes PC ism that the words you use is a direct indication of your racism, except we will exclude some groups for the use of those words.

Other studies have been done. including more direct polling, which also includes how racism benifitted Obama.


But does the study address the other side of the coin - the percentage of blacks that voted for Obama simply because he is black. Similarly, wouldn't the study be more complete if it looked at racial bias against both candidates. The assumption that racism only happens against people of color (any color) is naive.

Major Minor

"The assumption that racism only happens against people of color (any color) is naive." All people are people of color. The little girls in the Amazon Doll House post are pink and brown and equally adorable.

Artemis Fowl

This research method presumes that absolutely 0% of the population voted for him *because* he was black. If he gained, for instance, a 6% increase in turnout among voters who wanted to help elect the first black president... then in net being black was a benefit *not* a "loss".

And if anyone (and surely someone did) voted for him because he was black, then they were intrinsically also voting against his opponent because his opponent was not also black.

The entire premise here is that racism can only work as a benefit to a white candidate (John McCain) and not as a benefit to a black one. In other words, it only looks for racism against Obama.

The research itself starts from a racially biased assumption. It is hardly shocking that they managed to find a method to confirm that.


I'll be dammed I knew it, to bad that racism will probably cost him the re-election too. Also the economy


Yes, if you keep trying enough different methods, you can usually find *one* that will confirm what you "knew". Of course we'll just stroll past the dozens of studies that failed to confirm what you "knew"


Let's say Seth is right about the 3% - 5%, isn't that good news? Isn't this an indicator that racism is the exception, rather than the norm?

We all knew the election was a sign of improvement on racism, but given the noise the topic generates in our society, I would have expected the negative impact to be much higher.

At this point, I'm guessing short stature would have a more negative impact on election results than race.


As mentioned in a few previous comments, this study fails to consider the widely-known fact the black community often use the 'n-word' towards each other, not in a racist or derogatory way, but in a friendly manner. This therefore has most probably made your results both inaccurate and worthless, as such a term clearly doesn't embody racism in its entirety...