Insider vs. Outsider Matchup Finds Clinton, Trump Near Even

Hillary Clinton and Donald Trump run essentially evenly among registered voters in a head-to-head matchup for president in a new ABC News/Washington Post poll, testament to the strength of party loyalty as well as to Trump's anti-establishment profile and anti-immigration views.

The hypothetical contest stands at 46-43 percent, Clinton-Trump, a gap that's within the survey's margin of sampling error. That compares to a clear Clinton lead among all adults, 51-39 percent, indicating her broad support in groups that are less apt to be registered to vote, such as young adults and racial and ethnic minorities.

See PDF with full results, charts and tables here.

The close result in this poll, produced for ABC by Langer Research Associates, says as much about partisanship as it does about the candidates. Registered voters divide 45-40 percent between identifying themselves as Democrats, or leaning that way, vs. Republicans or GOP leaners. And 82 percent of leaned Democrats say they'd support Clinton, while 76 percent of leaned Republicans say they'd back Trump, were they the party nominees.

That said, Trump also is tapping factors including discontent with the political system, anti-immigration attitudes and dissatisfaction with the Obama administration. He leads Clinton by a broad 64-25 percent among registered voters who prefer a candidate from outside the political establishment and by 49-38 percent among those who strongly distrust politicians.

Trump also leads Clinton by 73-14 percent among those who favor his controversial views on immigration, 74-13 percent among those who disapprove of President Obama's job performance, 68-22 percent among political conservatives and 52-36 percent among whites, a broadly pro-GOP group in recent years. (They favored Mitt Romney over Obama by 20 percentage points in 2012.) Among evangelical white Protestants, a core GOP group, Trump leads Clinton by 67-22 percent.

This analysis is the first slice of a new ABC/Post poll on the 2016 election. More detailed results on the primaries, views of candidate attributes and attitudes about the political system overall will be released Monday morning.

There are some important provisos in evaluating these results. Early polls are not predictive. They seek to measure preferences if the election were today, but the election is not today, and if it were, voters would have had a full campaign's worth of information on which to base their choices - including whether to vote in the first place. Campaigns clearly do matter; front-runners have failed in past elections and single-digit candidates have surged to victory. Polls at this stage, then, are best used to understand attitude formation, not eventual election choices.

Statistical analysis shows which factors best predict Clinton vs. Trump preferences, holding all else equal. The biggest by far is whether or not registered voters support Trump's positions on immigration. That's followed by partisanship, preferring experience vs. a political outsider, ideology, race and gender.

GroupsNotably, in the general election matchup, Trump leads by 52-37 percent among men, while Clinton leads by 55-34 percent among women. Fifty-three percent of women in this survey say they're Democrats or lean that way, compared with 36 percent of men.

The results produce a vast 36-point gender gap -- Trump +15 points among men, Clinton +21 among women. The average in general election exit polls since 1976 has been 13 points; the biggest was 22 points in the Gore-Bush contest of 2000. The Clinton-Trump gender gap is more than twice as big as the Clinton-Jeb Bush gender gap in an ABC/Post poll in July, presumably reflecting Trump's controversial remarks about women.

Clinton's support among women is based on her overwhelming backing from college-educated women, 68-20 percent. By contrast, Trump leads Clinton by a broad 55-34 percent among men who aren't college graduates. He runs about evenly with Clinton among women without a college degree and among men who've graduated from college.

The education gap, like the gender gap, is outsized. In exit polls since 1980, there has been little difference in candidate support among those with a college degree vs. non-graduates, an average of just 2 points; the biggest gap was 11 points in 1996, when Bill Clinton's support was higher among non-grads (+14 points) than among college graduates (+3 points). In the Clinton-Trump matchup, there's a vast 35-point gap; it's 57-31 percent, Clinton-Trump, among those with a college degree, vs. 49-40 percent, Trump-Clinton, among those without one. Indeed, even among college-educated leaned Republicans, Trump's support slips to 67 percent, vs. 80 percent among those without a degree.

This reflects a challenge in Trump's support profile; while he does much better with less-educated registered voters, they're less likely actually to vote.

At the same time, Trump has a 20-point lead over Clinton among senior citizens, 55-35 percent, while Clinton has an even broader advantage among adults under 30, 63-27 percent. In this case it's Trump's group that has a higher propensity to vote.

Clinton also does vastly better than Trump among nonwhites, 72-19 percent; they're a core Democratic group and a growing share of the electorate. And there's a strong regional effect, with much better results for Clinton in the Northeast and West, with Trump ahead in the Midwest and South. Again, it's largely partisanship that leads the way.

MethodologyThis ABC News/Washington Post poll was conducted by landline and cellular telephone Sept. 7-10, 2015, in English and Spanish, among a random national sample of 1,003 adults, including 821 registered voters. Results have a margin of sampling error of 3.5 points for the full sample and 4.0 for registered voters, including the design effect. Partisan divisions are 33-22-35 percent, Democrats-Republicans-independents, among all adults, and 34-25-33 percent among registered voters.

The survey was produced for ABC News by Langer Research Associates of New York, N.Y., with sampling, data collection and tabulation by Abt-SRBI of New York, N.Y. See details on the survey's methodology here.