NY Times Correlates Trump Support With Racism — Based on Decade-Old Google Searches

January 2nd, 2016 2:19 AM

On Wednesday, Nate Cohn at the New York Times, who by some accounts is being anointed the next Nate Silver of polling, made a clumsy and despicable attempt to inject race into his political "analysis" of the Donald Trump phenomenon.

Cohn's tediously long writeup, which made Page A3 in the New York version of the Old Gray Lady's print edition on Thursday, attempted to identify and characterize Donald Trump supporters. Apparently troubled by finding that Trump's support crosses into a number of groups with whom Republican presidential candidates have usually fared poorly, he felt the need to go far afield for evidence of something sinister. Thus, he attempted to correlate the level of current support for Trump's presidential candidacy to regional levels of racism as seen in Google searches. That's right, Google searches — from 9-12 years ago.

The basis for Cohn's writeup is a pile of rubbish disguised as a scholarly paper by Seth Stephens-Davidowitz at Harvard.

Stephens-Davidowitz claims that one can estimate an area's relative degree of "racial animus" by determining "the percent of Google search queries that include racially charged language."

Stephens-Davidowitz's analysis is a one-way street with no pull-offs, as the only search terms he deemed to be "racially charged" were variations on the N-word (excluding the version that ends in "-ga," because rappers use that).

Google's worldwide search volume in 2013 was 1.216 trillion, with about 220 billion searches coming from the U.S. Stephens-Davidowitz's work from that year reported that variations on that "racial epithet ... (are) now (i.e., in 2013) included in more than 7 million searches annually." That's a whopping 0.0006 percent of all worldwide searches (i.e., 6 out of every million) and .0032 percent of all U.S. searches (3 out of every 100,000). As far as I could tell, Stephens-Davidowitz did not make it clear whether his "7 million" searches were U.S.-based only — but even if they were, it's quite clear that the nation is far less obsessed with the N-word than he is.

Stephens-Davidowitz's work analyzed N-word searches from 2004-2007. On that basis, he tarred entire areas as having enough "racial animus" to affect voting patterns — especially if those voting patterns ended up showing less-than-predicted support in 2008 for Barack Obama's general-election candidacy in comparison to notoriously inaccurate pre-election polls. The end result of his work is a ridiculous conclusion that Obama lost more votes than he gained in 2008 because of his race (which really isn't African-American, but managed perception is reality in our brave new word). Never mind the over 2 million additional African-American votes Obama got than John Kerry did in 2004, or that period's intense, media-driven "it's time we had a black President" sentiment.

Now let's get to Cohn's further abuse of an already flawed study in his Trump "analysis" (bolds are mine; HT to an emailer):

Donald Trump’s Strongest Supporters: A Certain Kind of Democrat

... (Trump's) geographic pattern of support is not just about demographics — educational attainment, for example. It is not necessarily the typical pattern for a populist, either. In fact, it’s almost the exact opposite of Ross Perot’s support in 1992, which was strongest in the West and New England, and weakest in the South and industrial North.

But it is still a familiar pattern. It is similar to a map of the tendency toward racism by region, according to measures like the prevalence of Google searches for racial slurs and racist jokes, or scores on implicit association tests.

... Racially charged searches take place everywhere — they are about as common as searches for “The Daily Show” or the Los Angeles Lakers. But they are more common in some parts of the country than others.

That Mr. Trump’s support is strong in similar areas does not prove that most or even many of his supporters are motivated by racial animus. But it is consistent with the possibility that at least some are. The same areas where racial animus is highest in the Google data also tend to have older and less educated people, and Mr. Trump tends to fare better among those groups — though the effect of Google data remains just as strong after controlling for these other factors.

Got it? Trump's biggest bloc of support comes from dumb, old, white racists.

Cohn's memo to America: You don't want to be associated with "those people," do you?

Cohn noted that "Mr. Trump’s best state is West Virginia." In his fevered mind, that of course has to be because Stephens-Davidowitz's work tells us that West Virginia was the nation's worst hotbed of N-word search — 9-12 years ago.

Horse manure. The real "anti-black" sentiment is in Washington. Have any of you in Manhattan ever heard of the left's eight-year war on coal, that black fuel which still provides 39 percent of the nation's electricity, aimed straight at the heart of the Mountain State?

Good heavens, 2016 is going to be a very trying year.

Cross-posted at BizzyBlog.com.