SO doomed.
Those who have read my poll musings over the past six years on Daily Kos know that I tend to cling to a simple maxim about polls with "holy shit" headlines, not unlike the one visited upon us by Quinnipiac on Wednesday morning.
That maxim is: "if you have one set of results, and everyone else has a different set of results, the odds are that 'everyone else' isn't the one that is wrong."
It is a credo I have employed when numbers were good for Republicans, but also one I have employed when numbers were good for Democrats. It is not an infallible motto—pollsters can certainly function as a "canary in a coal mine," planting themselves on the leading edge of a race that is undergoing dramatic change.
But sometimes, you just get a wrong'un. And despite the efforts of a number of voices in the political Twitter pundit community to take this result seriously, it just seems like this one doesn't stand up to scrutiny very well.
Follow me past the jump for the reasons why I suspect the Q team sliced this one into the woods.
REASON #1: The Favorability Issue
The Quinnipiac poll gives Hillary Clinton some bloody awful reviews on the classic favorable/unfavorable question.
Favorable/Unfavorable ratings, per state, for Hillary Clinton:
Colorado: 35/56
Iowa: 33/56
Virginia: 41/50
Now, as the presidential cycle has begun to develop, there is no question that Hillary Clinton's standing has taken a beating. The
Pollster average for Clinton's "fav/unfav" has dropped considerably since this time last year. In July of 2014, her numbers were at a "plus-7" (50 percent favorable, 43 percent unfavorable). The most recent Pollster average has the former senator and secretary of state at a "minus-5" (44/49).
But, the numbers Clinton gets in these three "swing states" would be roughly the numbers that President Obama gets in places like Arkansas, Kentucky, and Kansas. It strains credulity to consider that the Democratic frontrunner is polling that low in what are known to be competitive states.
What's more—in national surveys, Clinton's fav/unfav is competitive with the large GOP field. However, in this poll, her numbers are in the neighborhood of Chris Christie and Donald Trump. Republicans whose national numbers mirror, or are worse than hers (think: Bush, Huckabee, etc) blow her away in this survey. Again, that doesn't seem very plausible.
Quinnipiac also sees a deep collapse in her numbers that has not been matched nationally. According to their Colorado trend lines, Clinton's net fav/unfav has collapsed twenty points since February. The Pollster average has dropped, as well ... but only seven points.
What gives here? Well, they've either found a sample that is unusually filled with conservative Independents (the methodology pages seem to indicate a fairly plausible partisan split), or just one that has an odd level of animus for Hillary Clinton. Because, let's face it, there's no way she could be at Trump levels in Colorado and Iowa and not be cratering badly nationally, as well.
REASON #2: The Quinnipiac Precedent
Quinnipiac did okay in our pollster ratings for 2014. That said, a longstanding (to say nothing legitimate) critique of their work is that they have numbers that tend to bounce wildly without much in the way of external explanation for the movement.
In fact, Quinnipiac's tendency to occasionally produce "holy shit" numbers that get contradicted pretty quickly has been a topic on Daily Kos Elections before. Our own David Nir addressed it in September of 2014, when the Q poll had Republican Bob Beauprez up ten points with six weeks to go. No other pollster in the race's final months gave Beauprez anything better than a four-point edge. He eventually lost by three points.
Of course, the Q poll was also the one that served as a bit of a harbinger of Iowa Democratic doom, by being the first to show Joni Ernst with an outsized lead over Democrat Bruce Braley (in that same September battery of surveys). But, even then, they contradicted themselves, with their final poll showing the race tied (Ernst went on to win by eight points).
This is nothing new. The Q poll also had Republican challenger John Kasich up nearly 20 points on incumbent Democratic Governor Ted Strickland in 2010. The final margin wound up being just two points.
There are some other structural issues. Some of the crosstab numbers seem wonky—look, for example, at the Colorado crosstabs. How can Clinton sweep Democrats, Rubio sweep Republicans, Rubio lead Indies by 7, but lead overall by 8? Especially when the methodology document indicates a GOP +3 pool of voters? One would think that Rubio would have to be whomping Clinton with Independents in order to get to an eight point lead overall.
But, the bottom line is that these numbers simply don't jibe with the multiplicity of polls we have seen nationally. If Clinton was really down by mid-to-high single digits in Iowa and Colorado, and down a few points in Virginia, she wouldn't be leading the three leading Republicans—she'd be getting smoked.
The numbers are only possible to reconciled if, and it is a huge if, Clinton is doing dramatically better in some nontraditional places. We're talking "leading in Texas" kind of stuff.
More likely, someone is wrong. Either Quinnipiac missed on these three state polls, or the smattering of recent national polls are seriously overstating Clinton's strength. As they stand, however, the differing results simply don't add up.