Des Moines — Dana Van Woert strolled into this city’s downtown plaza on a September afternoon eager to help make history. Hillary Clinton was in town to headline a get-out-the-vote rally, and before it began, Van Woert scribbled her contact information on a campaign worker’s clipboard with the intention of volunteering at future events. The 27-year-old graphic designer, a graduate of Iowa State University, lives here in Des Moines and took her lunch break to cheer for a candidate who she hopes will become America’s first female president.
A few hundred feet away, just outside the event fencing, Dan Edwards was working to prevent that from happening. The 54-year-old retired painter and facilities worker, who lives 20 miles away in the tiny outpost of Van Meter, wore a camouflage NAPA Racing cap and a “Donald Trump for President” shirt. He thrust a handmade sign into the sky: Crooked Hillary. On the flip side: Honk if Hillary Lied. Edwards went to trade school, he explains in between the blare of motorists obliging him, where he learned to paint everything from iron beams to household furniture. But he never attended college.
Trump is the favorite to win this state’s six electoral votes on November 8, and for a simple reason: Iowa’s population looks more like Dan Edwards than Dana Van Woert.
They differ in their gender and geographic background, two reliable indicators of partisan preference. But the critical difference between Edwards and Van Woert — and the reason Republicans are poised for victory in Iowa — is a college degree. Political scientists and campaign tacticians have long viewed the “gender gap” — the difference in a candidate’s performance between men and women — as the most revelatory statistic inside a pre-election survey or exit poll. But the gender gap has been overshadowed this year by another demographic metric, known as the “diploma divide.”
College education has become essential to gauging political inclinations because it correlates with many other key demographic indicators: ideology, religion, geography, household income. Degree holders, on average, are more liberal, more secular, more suburban, and wealthier than voters who did not attend college. In this election, college-educated voters are far more likely to support Clinton, and non-college-educated voters are far more likely to back Trump.
In late September, Quinnipiac University surveyed 1,115 likely voters nationwide and found a dead heat: Clinton 44, Trump 43. But when sorted by educational attainment, the results were lopsided. Among college graduates, Clinton led Trump 49 percent to 36 percent; among non–college graduates, Trump led Clinton 49 percent to 39 percent. (Those margins dwarfed the gender gaps: Clinton led by five points among all women, and Trump led by four points among all men.)
The diploma divide is uniquely valuable in assessing individual battleground states because of their variance in college-educated vote share. Take Iowa. The state’s population ranks 37th in its proportion of voters with a college education, according to the financial-news service 24/7 Wall Street, which studied the Census Bureau’s 2015 American Community Survey. Its analysis found that only 27 percent of Iowa’s adults have at least a bachelor’s degree; it was the only state to see “a statistically significant decline” from the previous year. This helps to explain why Trump leads Clinton by five points in the RealClearPolitics polling average of Iowa.
Iowa’s mirror image, when it comes to education, is New Hampshire. Its population ranked eighth in the same study, with 36 percent of adults having at least a bachelor’s degree. And whereas Iowa’s proportion of non-college-educated voters in 2012 was 57 percent, four points higher than the national average of 53 percent, New Hampshire’s was four points lower, at 49 percent. Predictably, Clinton leads Trump by — you guessed it — five points in the RealClearPolitics polling average of New Hampshire.
Iowa and New Hampshire are in many ways polar opposites; their economies and cultures reflect broader regional differences. But they also have much in common. For one thing, given their positions atop the presidential nominating calendar, they got a longer look at both parties’ nominees than did any other state. In 2012, their electorates were both 93 percent white. Their gender breakdowns were similar; women made up 54 percent of Iowa’s electorate and 52 percent of New Hampshire’s. And President Obama carried both states by six points. How to explain, then, that Clinton leads by five points in one state while trailing by five points in the other?
Glen Bolger, a respected GOP pollster and co-founder of the firm Public Opinion Strategies, thinks it’s because “college-educated men are voting more and more like college-educated women, and non-college women are voting more and more like non-college men.”
What does that mean? Imagine a box divided into four squares. One belongs to college-educated women, the most liberal square. Another belongs to non-college-educated men, the most conservative square. The other two, non-college-educated women and college-educated men, are somewhere in between. Those two squares traditionally have been competitive but more recently have leaned as their gender demographics would suggest: non-college-educated women leaning slightly left, because women mostly do, and college-educated men leaning slightly right, because men mostly do. But Bolger believes that gender is no longer the determining factor in these two competitive squares; education is. Non-college-educated women are now leaning right, because non-college-educated voters mostly do; and college-educated men are now leaning left, because college-educated voters mostly do.
These trends have been visible for decades. Blue-collar whites have been steadily defecting from the Democratic party since the 1960s and white-collar whites have been leaving Republican ranks since the late 1980s, culminating in what The Atlantic’s Ron Brownstein has called “the class inversion.” It has accelerated beyond expectation in 2016 thanks to two nominees with limited appeal to the groups from which their respective parties were already bleeding support.
Trump performed twelve points better with non-college-educated voters during the GOP primary than he did with degree holders.
Trump, running hard to the right on immigration and preaching anti-trade populism, performed twelve points better with non-college-educated voters during the GOP primary than he did with degree holders. (“I love the poorly educated!” he memorably declared during his Nevada victory speech.) Clinton, who enjoyed strong support among working-class whites against Obama in the 2008 Democratic primary, saw much of it disappear eight years later as Bernie Sanders dragged her leftward on virtually every policy front. (Particularly unhelpful in places such as Ohio and Pennsylvania, Democrats say, was her pledge to “put a lot of coal miners and coal companies out of business.”)
The reason non-college-educated voters tend to support Trump, pollsters emphasize, isn’t simply their lack of schooling — or intelligence. Rather, competing in the 21st-century job market without a college degree probably means they “have had a really tough run over the past decade, have a much bleaker perception of their economic situation,” and are therefore more receptive to anti-Washington rhetoric and promises of sweeping change, says Steve Koczela, president of the MassINC Polling Group, which has partnered with Boston’s public-radio affiliate to poll New Hampshire throughout the primary and general elections.
Indeed, when analyzing the exit polling from the 2016 primaries, renowned pollster Gary Langer of ABC News concluded: “Perhaps most important is the long-term trend in lagging incomes. Americans with a college degree have seen their real median weekly earnings rise by nearly 23 percent in the past 35 years. All well and good — but those who lack a four-year college degree have seen their real incomes decline by 9 percent.”
Here again, Iowa and New Hampshire are a useful case study. In 2012, according to exit polls, 33 percent of New Hampshire voters reported an annual household income over $100,000, ranking it among the wealthiest states. In Iowa, however, that number was just 19 percent, the lowest of any state that was polled. The correlation — states with more educated populations also have more affluent populations — was predictable but not politically predictive: Obama won Iowa and New Hampshire by an identical six-point margin; there was no reason to expect that a college diploma would be the brightest dividing line in the electorate four years later.
“I would love to tell you that I was brilliant and forward-looking and knew that college education would be the dominant indicator,” says Chris Wilson, the polling and data guru for Ted Cruz’s presidential campaign. “We were weighing and analyzing more than 500 indicators. But we didn’t really start to see education emerge as dominant until South Carolina. And Nevada was the first time we zeroed in on it.”
In fact, Trump and Cruz performed comparably with non-college-educated voters in Iowa, exit polls showed. And New Hampshire was such a landslide — Trump topped the second-place finisher by nearly 20 points and swept every education demographic — that conclusions were difficult to draw. But in South Carolina, a heavily Evangelical state where it expected to perform well, Cruz’s team ran into trouble. Many of the Evangelicals they targeted ultimately broke for Trump, “and they were the Evangelicals without a college degree,” Wilson says. By the time Nevada voted, Cruz and his fellow Republicans saw a trend developing: Any other demographic characteristic of a particular voter that might have made him or her a supporter — age, occupation, marital status, preferred grocery store, favorite TV show — was overwhelmed by the single question of whether he or she had graduated from college. If the answer was no, the voter was much more likely to support Trump.
This dynamic delivered Trump the nomination; if amplified, it could carry him to the White House. But that seems unlikely: College-educated voters account for a significantly higher portion of the November electorate than of the GOP primary electorate, and they are heavily concentrated in the suburbs of the nation’s most competitive battleground states. Even if Trump mobilizes millions of working-class whites who sat out previous elections, it probably would not compensate for the defections of college-educated Romney supporters to the Democratic party. And to be clear, any such surge in non-college-educated voters appears highly improbable. In each of the past three general elections, the share of voters with a bachelor’s degree has climbed steadily; so too has the share of voters with a post-graduate degree. Those trends appear inexorable, and they pose yet another demographic problem for a party that has failed to solve several others.
“The Republicans have this wacky plan where we try to do better with groups that are getting smaller in the electorate, and we try to piss off groups that are getting larger in the electorate,” Bolger says. “Whether it’s women, Latinos, college-educated — it’s like betting on the slowest horse in the race.”