Let’s start with an analogy. A national presidential poll is like the scene in one of those viral-outbreak movies where the hazmat-suited scientist asks someone how many people are infected. “About 20,” comes the response, and the hazmat teams grab a bunch of supplies and heads in. But the real information comes once they start talking to patients: Which ones are sick, and how sick are they? The former gives us some sense of the outbreak; the latter gives us much better information about where this thing is headed.
In an election pitting two deeply unpopular candidates against one another, we’re going to be infected with a president that half the country hates. National polling gives us a general sense of where we’re headed — at this point, Hillary Clinton — but state polling gives us a lot more information about just how sick Donald Trump is, and where. (It’s not a perfect analogy.)
On Tuesday, we got six new polls in four likely battleground states — Florida, Iowa, Ohio and Pennsylvania — from Quinnipiac University and NBC News/Wall Street Journal/Marist. (Quinnipiac surveyed in Florida, Ohio and Pennsylvania. NBC polled in Iowa, Ohio and Pennsylvania.)
Clinton leads in all six polls. She leads by 1 point in Florida (Quinnipiac), by 4 points in Iowa (NBC/Journal/Marist), by 4 or 5 points in Ohio and 10 or 11 in Pennsylvania, with the lower figures in each case coming from the NBC/Journal/Marist poll.
Why are the NBC/Journal/Marist numbers lower? One likely reason is that Quinnipiac used likely voters, as opposed to registered voters. That matters, simply enough, because Republicans are historically more likely to vote than Democrats. “This gap is especially large in low-turnout elections,” The Post’s pollster Scott Clement replied when I bugged him to provide a technical explanation. “Washington Post-ABC News national polls, which were highly accurate in both 2008 and 2012 general elections, found Republican candidates’ support was about 2 percentage points higher among likely voters than among registered voters overall, while Democrats’ support was roughly the same among each.” In other words, a poll showing Clinton and Trump tied among registered voters would show Trump with a 2-point lead when narrowed to likely voters.
That switch by Quinnipiac from registered to likely voters probably means that the change in each state since it polled at the beginning of July — plus-4 points for Clinton in Florida and Ohio and plus-12 in Pennsylvania — probably looks smaller than it actually is. The change in the NBC/Journal/Marist polls: Clinton is plus-1 in Iowa, plus-5 in Ohio and plus-2 in Pennsylvania. In general these are smaller post-convention bumps for Clinton than we’ve seen in national polls.
Other details from the state polls mirror national polling, though. Trump voters are more likely to back him because they oppose Clinton in Quinnipiac’s survey, while Clinton voters are more likely to support her as a candidate because they like her, not simply because they dislike Trump. The largest split continues to be on gender lines. In Quinnipiac’s poll, Trump leads by 5 to 12 points with men — but trails by 13 to 23 points with women.
It’s important to note that Barack Obama won all four of these states in 2012. At the same point in the campaign, shortly after the party conventions, Obama led by about 8 points in Pennsylvania (according to the RealClearPolitics average, but had only a fraction-of-a-point lead in Iowa.
RealClearPolitics hasn’t updated its state averages with the NBC poll, but you can see that the patterns are similar to what they were four years ago. A big lead for the Democrat in Pennsylvania, smaller leads — but leads — in Ohio and Florida. (The most recent Iowa average is from between the conventions.)
The change in Pennsylvania after the Democratic convention looks like the change in Ohio after the conventions in 2012.
At some point, Trump needs to start picking up states. In fact, if nothing else changes on the 2012 map, he needs to win all of these states, save Iowa. A 1-point lead in Florida isn’t insurmountable, but a 10-point lead in Pennsylvania probably is, if nothing else changes. (Try out your own scenarios here.) If Trump can’t pick up Pennsylvania, he needs to win Ohio, Florida, Iowa and at least one other big state, like Virginia. Clinton leads by 7 in Virginia — and she leads in North Carolina, which Trump needs to hold.
Back to the “deadly virus” analogy. Clintonism has been spreading after the conventions. Trump is in danger of the outbreak spreading beyond his control — which, for his candidacy, is not survivable.