Are the exit polls that nearly every election analyst relies on flawed? That’s the question posed by New York Times Upshot reporter Nate Cohn — a question whose answers have profound implications for how you view the 2016 general election.
Standard analysis says Democrats have a built-in advantage because the electorate is increasingly nonwhite. Exit polls say the share of whites in the electorate has fallen from 77 percent in 2004 to 74 percent in 2008 and 72 percent in 2012. This year, they said, 13 percent of voters were black, 10 percent Latino, 3 percent Asian.
Since then, those groups have voted mostly for Democrats, though their support has varied, and analysts say it is very challenging for a Republican to win, especially one who, like Donald Trump, alienates people of color.
But exit polls can be imperfect, Cohn said. As he points out, they are not designed to accurately represent the proportions of each demographic group, but to indicate the actual outcome, within a statistical margin of error, and to show the different responses of significant subgroups.
Moreover, and Cohn does not mention it, the results are manipulated by exit pollsters and media analysts who realize that some voters (juvenile voters, for example) are more likely than others to complete exit polls.
So Cohn considered two other sources of information about how people voted: the Census Bureau’s Current Population Survey and the Democratic Party’s Catalist voter data, which is compiled from public voting records.
Both sources, like exit polls, have their weaknesses. Many nonvoters tell the Census Bureau they voted. Voter registrations are often flawed or out of date. Still, both sources paint a significantly different and probably more right picture of the electorate than exit polls.
The CPS and Catalist say that the 2012 electorate was 74 per cent and 76 per cent white, up from the 72 per cent in the exit poll. They say that only 15 per cent of voters were under 30, rather than the 19 per cent in the exit poll, and 61 per cent and 62 per cent were 45 or over, rather than the 54 per cent in the exit poll.
Most importantly, they estimate that the percentage of white people over the age of 45 who do not have a college degree – the main group of Donald Trump’s supporters – is 30 percent and 29 percent of voters, respectively, which is significantly higher than the 23 percent reported in the exit poll.
Assuming these figures are correct, Cohn estimates that 34 percent of Barack Obama’s 2012 voters were white noncollege graduates, not the 25 percent that exit polls indicate. Obama’s share of this demographic group declined in 2008, but mostly in the South, and held steady in much of the North, including the target states of Ohio, Wisconsin, Iowa, and Colorado.
In other words, Democrats are more dependent on whites without college degrees than most analysts assume. They have been losing ground with this group for years, but they could lose more, in which case Obama’s 51 percent majority in 2012 might not be repeated.
Current polls show Trump doing better than Romney among whites without college degrees. But he also does worse among white college graduates, so he doesn’t gain among all whites.
But notice that one thing pollsters have a tough time gauging is turnout. Contrary to popular belief, turnout—and especially Democratic turnout—was down during the Obama years. Obama received 3.5 million fewer votes in 2012 than he did in 2008. Cohn presents data showing degenerating turnout, especially among juvenile white Democrats.
It’s reminiscent of something evident in this year’s primary results. Overall, Republican and Democratic turnout was roughly equal. But by April 5, when the results were uncertain, Republican turnout was 5 million higher than Democratic. In the swing states, the Republican advantage was 1.6 million.
Republicans also had a 5 million turnout advantage in states with open primaries, where voters could choose the party they wanted to vote for. The Democrats’ 4 million-person lead in the post-April 5 primaries, almost all of which were closed, was absorbed by pre-existing Democratic registration advantages.
This is in stark contrast to 2008, when Republican turnout exceeded Democratic turnout in only five states, including two whose primaries were canceled by Democrats nationally (Florida, Michigan), two predominantly Mormon states (Utah and Idaho), and Alabama, Arizona and Nebraska.
The turnout in the primaries was a harbinger of the high turnout favoring Democrats in November 2008. It is not clear whether that will be the case in November, or whether the exit polls are wrong. But both signals suggest that Trump’s chances may — may — be better than current polls suggest.
