How Racist Are We? Ask Google

[cthulu2016]

June 9, 2012, 5:46 pm
How Racist Are We? Ask Google
By SETH STEPHENS-DAVIDOWITZ

...many Americans use Google to find racially charged material. I performed the somewhat unpleasant task of ranking states and media markets in the United States based on the proportion of their Google searches that included the word “nigger(s).” This word was included in roughly the same number of Google searches as terms like “Lakers,” “Daily Show,” “migraine” and “economist.” A huge proportion of the searches I looked at were for jokes about African-Americans. (I did not include searches that included the word “nigga” because these searches were mostly for rap lyrics.) I used data from 2004 to 2007 because I wanted a measure not directly influenced by feelings toward Mr. Obama. From 2008 onward, “Obama” is a prevalent term in racially charged searches.

The state with the highest racially charged search rate in the country was West Virginia. Other areas with high percentages included western Pennsylvania, eastern Ohio, upstate New York and southern Mississippi.

Once I figured out which parts of the country had the highest racially charged search rates, I could test whether Mr. Obama underperformed in these areas. I predicted how many votes Mr. Obama should have received based on how many votes John Kerry received in 2004 plus the average gain achieved by other 2008 Democratic Congressional candidates. The results were striking: The higher the racially charged search rate in an area, the worse Mr. Obama did.

Consider two media markets, Denver and Wheeling (which is a market evenly split between Ohio and West Virginia). Mr. Kerry received roughly 50 percent of the votes in both markets. Based on the large gains for Democrats in 2008, Mr. Obama should have received about 57 percent of votes in both Denver and Wheeling. Denver and Wheeling, though, exhibit different racial attitudes. Denver had the fourth lowest racially charged search rate in the country. Mr. Obama won 57 percent of the vote there, just as predicted. Wheeling had the seventh highest racially charged search rate in the country. Mr. Obama won less than 48 percent of the Wheeling vote.
...

http://campaignstops.blogs.nytimes.com/2012/06/09/how-racist-are-we-ask-google/

[LARED]

1. Interesting theory

He should get the help of a statistician to see if there actualy is a correlation between racially motivated google searches and voting habits within geographic areas.

[cthulu2016]

2. The blog entry is sumarizing his technical paper

He's an econmist at Harvard. The real paper is available as a pdf.

http://www.people.fas.harvard.edu/~sstephen/papers/RacialAnimusAndVotingSethStephensDavidowitz.pdf

[cloudbase]

3. I have some questions about the validity of his method.

"I predicted how many votes Mr. Obama should have received based on how many votes John Kerry received in 2004 plus the average gain achieved by other 2008 Democratic Congressional candidates."

Doesn't pass the smell test.

[Jamaal510]

4. Clearly this is still a very sick country.

Who knew that the N word would be searched as often as something like "Lakers" and "Daily Show"? Kinda makes me wonder if some of the folks I come across in person are screwed up in the head like this.

[iemitsu]

5. unless you hide at home most of the time

you have met some.
they are out there.

[Igel]

6. Dunno.

I just spent 45 minutes trying to figure out if I think he has a clue.

Don't know.

"Insights" is an interesting tool, but it's a bit of a vacuum cleaner. So for the n-word one of the most frequent searches was just the n-word repeated 3x.

All the instances of that search from 1/1/2004 to 12/31/2007 were in 2004. Huh?

And, as far as I can tell, it appears to be because of the lyrics of a song. A single song. By a white boy.

Did I mention it was an anti-racist song?

Yeah, that's part of his data showing racism.

In fact, when he compares the most common searche with the n-word versus "nigga" there's a large overlap. I'm not sure everybody shares his spelling conventions.

It doesn't help that he doesn't use numbers. Google "Insights" gives you rankings on a scale, the most common search/year/etc. being 100; it says nothing about how frequent a search is. I can't find that the author, for all his outrage, says how often the words or searches actually occurred. Now, "migraine" may be a very common search, with 50k searches per month in each state; or it may not. I don't think I've ever searched on it. I know I've never searched on "Lakers." Some hint as to actual frequency, as opposed to relative frequency, would be nice.

Some of his data plots are interesting. In some, you realize that a variety of curves could fit the data, and each curve would lead to a different interpretation. It's unclear why he picked the ones he did. Except that his thesis is on new techniques for data analysis. So I guess he used them. And they lead to the kind of politics he had before he started his grad program, so he kept them.

In some of his plots you have to think, "Dumbass." There were some clear endpoint outliers in his data. Get rid of one data point out of many and the skew of the line, the way the line would have to slant to get the fit, would change. The slope was important. And depends on one data point. That's wrong. Easy to do, but wrong.

Can't sort out how all the strengths and weaknesses fit together. It's a poli sci major, when push comes to shove.

[cthulu2016]

7. You can read the actual study, if interested

If you've read it, I repeat this link for others.

http://www.people.fas.harvard.edu/~sstephen/papers/RacialAnimusAndVotingSethStephensDavidowitz.pdf

The reason he uses data from 2004-2007 is explained. By 2008 the searches including both "nigger" and "Obama" is so high that the n-word cannot be used to seek isolated racism, independent of Obama.

And that in itself says a lot.