Either Hillary Clinton has a comfortable lead over Donald Trump, or Trump and Clinton are virtually tied and the momentum is with the Republican.
Other recent surveys present differing pictures of where the general election stands in key swing states.
A June 30-July 11 Quinnipiac poll found Trump up by two points in Pennsylvania, while a July 7-10 NBC News-WSJ-Marist poll showed Clinton up by nine points in the state. Similarly, while Quinnipiac found Trump leading by three points in Florida, the NBC-WSJ-Marist survey showed Clinton ahead by seven points in the Sunshine State.
If Trump is even or ahead nationally, then the Quinnipiac state numbers seem possible. But if Clinton is ahead by five points or more nationally, the NBC-Journal-Marist state numbers are much more reasonable.
So what should you believe?
At this point, you can do one of two things. You can pick the poll that you want to believe (as many do), or you can look below the top-line responses to understand each survey’s results. And, of course, you should look at the past few elections to see how various groups have voted in the past and to assess which of the current numbers seem most reasonable.
I usually look to the cross-tabs for some help in thinking about the data, and I almost always start with race, since it produces such a dramatic division in the electorate.
The June CBS-New York Times poll had Clinton leading by six points and Trump winning whites by only six points. The July CBS-Times poll had the ballot test tied and Trump winning whites by 13 points.
Trump had to improve among white voters between June and July to perform better on the overall ballot, so that makes sense. He got a higher percentage of whites, so his overall number against Clinton improved.
But Trump’s showing among whites in each of the major polls suggests big general-election problems for him, not a dead heat.
Mitt Romney won whites by 20 points in 2012, John McCain won them by 12 points in 2008, and George W. Bush won them by 25 points in 2000. Of the three, only Bush won the election, and that was when whites constituted 77 percent of the electorate (or at least of the exit poll).
It seems very unlikely that Trump can defeat Clinton — or even run close to even against her — if he carries whites by only 13 points, so the CBS-Times ballot test looks suspicious.
The other three recent surveys show Trump winning whites by 15 points (Post-ABC News) and 13 points (NBC-WSJ and CNN) but losing the ballot test. That is a much more reasonable outcome based on recent performances of Republican nominees.
If Trump is winning whites by only a dozen points or so, he would need to be out-performing McCain and Romney with African Americans or Latinos to have any chance of winning in November. (There is no evidence that whites will constitute a dramatically larger percent of the electorate in November than the 72 percent they did in 2012.)
That is not happening, at least not yet. A July 9-13 NBC News-WSJ-Telemundo poll of registered voters showed Clinton leading Trump 76 percent to 14 percent among Latinos, far worse than Romney’s 27 percent showing among those voters in 2012.
The Post-ABC, NBC-WSJ and CNN polls contain plenty of other data, and almost all of it is bad for Trump.
Trump’s negatives remain sky-high — and higher than Clinton’s — and the GOP brand is horrible (and much worse that the Democratic brand). In addition, President Obama’s poll numbers have rebounded, and a majority of Americans now say they approve of his performance in office. And even with the public’s doubts about Clinton’s honesty, a plurality of registered voters tell pollsters she has the better qualifications and judgment to be president.
So what about those Quinnipiac polls that show Trump leading in Pennsylvania and Florida? They are hard to swallow — if you believe that the Post-ABC, NBC-WSJ and CNN polls are an accurate snapshot of the race nationally.
If Clinton is holding a middle-single-digit lead nationally now, it’s very likely that the NBC-WSJ-Marist polls in both states, which show Clinton comfortably ahead, are a better reflection than Quinnipiac’s of where the race stands.
Of course, if that makes you unhappy, you can believe any poll you want, even Rasmussen.
Or you can wait for the next poll and hope that it will show something different.