There's a mystery, or riddle, in the latest national Washington Post-ABC News poll.
A majority of registered voters say Donald Trump is "not qualified" to be president — 58 percent said as much in the latest survey, compared with 42 percent who said Hillary Clinton is unqualified. In all, 49 percent now feel "strongly" that Trump is not qualified to be president. Yet Trump trails Hillary Clinton by a mere 4 percentage points in the latest survey (47 to 43), a smaller margin than one might imagine, given that so many reject his basic qualifications. How is he overcoming (most of) the deficit?
Digging one level deeper in the survey offers some clues, looking at voters who have contrasting views of Clinton and Trump's qualifications. The decision for these voters seems like it should be fairly simple — one candidate is qualified and the other is not. It isn't. In the poll, 46 percent of all registered voters said Clinton is qualified and Trump is not, while a smaller 30 percent said only Trump is qualified.
These groups of voters are acting largely as expected — Trump receives a near-universal 98 percent support from voters who say that only Trump is qualified, while Clinton receives slightly less support among those who say only Clinton is qualified (92 percent).
Greater unity among the only-Trump crowd is notable, but only nets Trump about 1 percentage point more support among the electorate overall. Clinton still nets a 13-point advantage among all registered voters based on her advantage with those who see Clinton or Trump as more qualified than the other.
By deduction, Trump's strength must come from the remaining fifth of voters who see no difference between Clinton and Trump's qualifications — either saying both Trump and Clinton are qualified (9 percent) or "neither" of them are qualified (11 percent). These voters see Clinton and Trump as roughly equal on this basic question of qualification, and we might expect they'd split about evenly in support for the two major-party candidates.
But it turns out that that expectation would be wrong. Pooling June and July surveys to boost sample size for these smaller groups, Trump leads by 67 percent to 25 percent over Clinton among voters who say both candidates are qualified, and Trump has a 41 percent to 17 percent lead among those who say neither is qualified. A sizable 41 percent of the "neither" group say they would support neither candidate or have no current preference.
Trump's margins with "both" and "neither" qualified groups have a big impact. Trump nets a 7 percentage-point gain in support from voters who see Clinton and Trump as equally qualified or not qualified, overcoming more than half of Clinton's advantage among voters who say only one of the candidates is qualified.
Why is Trump faring so well with voters who see Clinton and Trump as equally qualified or unqualified? Their partisanship offers clues — over the past two surveys, voters who said "both" or "neither" candidates were qualified were 14 points more likely to lean toward the Republican Party than voters overall. Those who said both are qualified were also 11 points more likely to be men, a group that tilts more Republican.
This latter dynamic makes some sense — many Republican leaders have vocally questioned Trump's fitness for the presidency, and some of those concerns are seen among Republicans nationwide. These voters were always going to be hard gets for Clinton, and many may adopt Paul Ryan's reasoning that stopping Clinton is enough motivation to put aside reservations about Trump and vote the party line.
The poll also shows that Clinton has a lot riding on her ability to differentiate herself from Trump on a basic level of qualifications, one of his biggest weaknesses. Trump's choice of Indiana Governor Mike Pence as his running mate could help him along these lines, as could taking on Clinton's record as secretary of state, in which capacity she was seen as an effective and strong leader.
But the more important takeaway is that Trump doesn't need to beat Clinton on the qualifications measure — a draw with the right amount of voters would be enough.
The Post-ABC poll was conducted July 11-14 among a random national sample of 1,003 adults reached on cellular and landline phones. Overall results have a margin of sampling error of plus or minus 3.5 percentage points; the error margin is 4 points among the sample of 816 registered voters.