One of the big assumptions about the 2016 election is Donald Trump cannot win due to the diversity of the country. Indeed, this author has already seen a plethora of articles suggesting CO and NV are out of reach for the GOP nominee and how AZ, GA and NC could flip this cycle with increased minority turnout.
Many prominent GOP strategists worry that Trump cannot win without increasing Mitt Romney’s share of the minority vote. Joe Scarborough, the token conservativeish pundit for MSNBC put it thusly, “There are not enough white voters in America for Donald Trump to win while getting routed among minorities.” But, maybe there are!
An analysis conducted by Nate Cohn at the Upshot suggests millions more white voters participated in 2012 and that, contrary to the popular themes of 2012, it was Northern whites who put Obama over the top.
The most popular theme of the 2012 election was that Romney lost because he did so poorly among non-white voters. This is largely what drove many GOP strategists to conclude the party needed to expand its racial reach after 2012 and win larger shares of Hispanics and Asians. But, these conclusions were based on exit polls. The same exit polls that in 2010 overestimated GOP support and underestimated it in 2014.
The data from the Upshot gathered from the Census Bureau, voter registration files, polls and the finalized results suggest that Obama actually ran better among whites outside the South than both Al Gore and John Kerry. It was his weakness among Southern whites that gave Romney a 20 point lead among whites on election night. Put simply, Obama would have been reelected even if the electorate resembled 2004.
This makes sense if you think about it more. In the key swing states of Ohio and Colorado (to name 2 examples), Bush carried Araphoe and Jefferson Counties by the mid-single digits and he carried Hamilton County, Ohio by a similar margin. Obama basically flipped these margins in 2012.
Obviously, this benefited Obama in 2012 and it portends good things for Mr. Trump in November if he plays his cards right. The downside is that in appealing Northern whites Mr. Trump may hurt himself with college-educated voters and women.
Exit polls obviously play a clear part in the GOP’s worries about its appeal to a diversifying electorate. When tens of thousands of voters in precincts across the country report their vote it is hard to ignore. But, exit polls are imperfect instruments. They rely on self-reporting in many cases. Further, perhaps due to this self-reporting and the fact pollsters massage the data to account for missing variables, the electorate appears more educated and diverse than it really is.
The Census and registered voter files paint a different picture of the electorate than exit polls do. The Current Population Survey, used by the Census Bureau to report unemployment numbers also asks individuals if they had voted in the last election.
Registered voter files, a compilation of local voting records, are used by both parties to gauge turnout and results. They tend to be less biased due to their reliance on what the data tells them. Using Catalist, a Democratic firm that offers an academic subscription, the UpShot found white turnout increased from 1 to 3 percent in 2012 compared to exit poll results. Those without college degrees increased their share of the electorate from 4 to 10 percent and the share of the electorate over 45 increased by almost 10 percent.
This has profound implications if true. First-off, it shows how inaccurate exit polls are. Secondly, how difficult it is to ensure they are accurate. Lastly, that major disagreements are fought over data from voter files or CPS data. Indeed, the Voter ID law up for debate in North Carolina is being fought in the courts using voter file data.
One can take examples from battleground states to illustrate exit polls woes. In Ohio, exit polls showed an electorate that was 15 percent black. This means 250,000 more black voters showed up in Ohio than in 2008 despite overall turnout in the state dropping. In North Carolina, exit polls showed more Hispanics voted with college degrees than actually live in the state (according to Census data).
This runs flat against the narrative the Democratic coalition is dependent on the young, minorities and women. While Democrats cannot win without these groups they still need the support of white working class voters.
This is illustrated by the 2014 Colorado Senate contest. In 2008, Senator Mark Udall carried 62 percent of white women and 58 percent of white men. But, in 2014, after running a campaign targeting only women, his share among white men dropped to 42 percent and dropped by 4 points among white women. County-wise, he lost ground in the populous Denver suburbs and conservative Northern Colorado.
Relating to 2012, Obama carried 34 percent of whites without a college degree compared to the 25 percent shown by exit polls. His margins among these whites increased in almost every state outside the South (minus a few conservative Western states, Pennsylvania and Massachusetts). In every battleground state in 2012 minus Iowa, the CPS and voter file data indicates Obama ran better among whites than exit polls showed.
This plays down the theory that Hispanics cost Romney the election. In Florida, you can bet they did with his weakness in Cuban-American dominated Southern Florida and the heavily Puerto-Rican Osceola and Orange counties. But, in Virginia and elsewhere, Hispanics played a marginal role in giving Obama his wins.
If this is the case Republicans should be happy and worried. They should be happy because it suggests non-whites are not a baked-in wing of the Democratic coalition. But, it also suggests Northern whites backed Obama for a reason.
I disagree with Cohn the auto bailout and painting Romney as a plutocrat did not play a large part in the election. If they didn’t, it is true they drifted to Obama over cultural issues. But, an extraordinary number of polls have shown white, non-college educated voters agree on many issues with Trump. Trump, rarely agrees with Obama on anything culturally.
Another theory this new data runs against is the theory of the “Missing White Voter.” In 2012, due to exit polls, it was assumed fewer whites voted than past elections. Yet, voter registration files suggest many of these voters were registered Democrats. Indeed, the CPS data suggests turnout among Republicans 60 and older increased in 2012 relative to 2008. It is also unclear whether these voters were young and older. Regardless, they were not in the battleground states that decided the election.
So can Trump really pull this off? A compilation of polls show he is beating Clinton among whites without a college degree 58 percent to 31 percent nationally. That is better than Romney did in polls but it is unclear whether these gains are coming in battleground states or not.
The same polls showing Trump doing better among non-college educated whites also show him losing college educated whites to Clinton. Just for reference, even McCain did not lose college educated whites to Obama in 2008. So, if Trump loses college educated whites he would need astronomic margins among non-college educated whites to win.
Now, it’s possible the polls could be underestimating Trump’s support. Voters can feel uncomfortable admitting to a stranger over the phone they support Trump (especially with the media going overboard calling him a racist and women-hater). They may also suffer from social desirability bias. In a test of over 2,300 Republicans last year it was found in online surveys Trump did 6-10 points better among voters along education lines.
Even if so, the dearth of polls in battleground states makes dissecting the individual results in national polls difficult. But, if Cohn is right, Trump does have a shot this cycle, contrary to the narrative of a diversifying electorate automatically carrying Clinton to victory.