Yesterday fans of Hillary Clinton were probably pleased to see her doing pretty well in head-to-head match-ups with a number of Republican candidates in a new national CNN/ORC poll. Among registered voters, she led Jeb Bush 52/43; Donald Trump 51/45; Scott Walker 52/46; and Carly Fiorina 53/43.

But then today a new hailstorm of bad news for HRC arrived from Quinnipiac (which began the whole “Clinton is in trouble!” meme last month with polls showing her support dropping sharply in Colorado and Iowa) via new surveys of Florida, Ohio and Pennsylvania.

In Florida, Q-Pac has Clinton losing by double digits to Bush and Rubio, and even trailing Donald Trump. Her favorability ratio there is 37/55, even worse than Trump’s.

She’s doing a bit better in PA, trailing Bush and Rubio by 3 and 7 points, respectively, and actually leading Trump by 5. But her favorability ratio is a terrible 38/55. Keep in mind that the Keystone State hasn’t gone Republican in a presidential election since 1988.

The silver lining for Clinton in this battery of “swing state polls” is in OH, where she leads Bush and Trump and only trails Rubio by 2. But again, her favorability numbers are wretched (36/54).

For dessert, Q-PAC shows Biden doing better than Clinton in virtually every match-up, and having significantly better favorability ratios.

So who are we to believe? CNN/ORC or Q-PAC?

The CNN/ORC numbers are pretty consistent with overall polling trends: HuffPost Pollster‘s polling averages have Clinton leading every named Republican nationally by a minimum of five points. There’s not enough “swing state” polling to provide much of a comparative test of Q-PACs numbers. It’s possible, I suppose, that something’s going on in Colorado, Florida and Iowa (the states where Q-PAC shows Clinton doing so poorly) that makes them less battle-groundy than in the past; Republicans did do unusually well in all three states in 2014, not that this should necessarily matter in a presidential year.

At this point it makes sense to pay more attention to the more abundant national polls in which HRC is still in good shape than in one pollster’s take on specific states. Some observers think national polls are inherently superior, though in 2012 FiveThirtyEight’s judgment was otherwise (in part because some big national firms, notably Gallup and Rasmussen, missed the mark by quite a bit).

I’m still not with the “ignore ’em all” camp that refuses to look at presidential polling data until some arbitrary date when suddenly they matter a lot. But when they conflict, there’s not much recourse until there’s a lot more data in the can. You can safely ignore, however, the drumbeat of spin we’ll hear today suggesting that Q-PAC has proven HRC is tanking fatally and/or Joe Biden should jump into the race.

UPDATE: I received an email from Alan Abramowitz of Emory University that likely explains the Q-PAC mystery: their samples in FL, OH, and PA don’t much resemble the general electorate as it appeared in 2012. For example, the PA sample is 86% white; the 2012 exit poll showed an electorate that was 78% white. More egregiously, Q-PAC’s PA sample shows a tie on party identification whereas in 2012 Democrats had a ten point advantage. All three state samples in the Q-Pac poll include larger percentages of voters 65 and older than the exit polls. All these variances could likely explain better Republican performances than in other polls.

Our ideas can save democracy... But we need your help! Donate Now!

Ed Kilgore is a political columnist for New York and managing editor at the Democratic Strategist website. He was a contributing writer at the Washington Monthly from January 2012 until November 2015, and was the principal contributor to the Political Animal blog.