After all, here is a national poll showing a two-point edge for the Republicans in the generic ballot (44-42) over the Democrats. And here is a poll that showed a pretty outsized negative-14 split on job approval for the President (43-57, they did not allow undecided as an option, evidently).
On further review, however, there is a lot less than meets the eye here. If anything, this poll might be yet another sign of a status quo election in November.
The reason why lies just after the jump...
The simple answer to why I think this poll does not speak to inherent Democratic peril in the fall is simply a matter of the sample. Politico (through their polling partners GfK and Social Sphere) decided to, rather than conduct a simple national poll, to conduct a poll of what they deemed as the "competitive" House and Senate seats up for grabs in November.
Now, assuming that the readers can appreciate the distinction, there is nothing wrong with doing that. Unless you lump these results in with other national polling, which is (again, from a mere scan of the Twitter chatter on this poll) what a lot of folks were doing.
Herein lies the problem: Politico's decisions on what races rate as "competitive" are decent (maybe a bit broad: 16 Senate races and 66 House races). But it is also inherently biased, however incrementally, to the GOP. Look at the numbers: even though Barack Obama carried more of the over five dozen House districts included in the sample (he carried 36, while Romney carried 30), the average Obama performance in those districts (as culled from our own database of presidential performances by congressional district) stood at 49.0 percent. This runs almost precisely two points behind Obama's national performance.
Indeed, in these 66 House districts that made up the Politico poll sample, Mitt Romney was at near perfect parity with the president, despite losing by nearly five million votes nationally (or, to keep apples to apples, 3.86 percent).
Of course, on the Senate side, the differences were even more stark. Anyone who follows elections knows that the reason why the 2014 battle for the balance of power in the Senate is so tenuous for the Democrats is that they are playing incumbent defense in some states that are, quite simply, pretty butt-ugly for them. And, as it happens, the Politico sample reflects that. To their credit, they tried to ameliorate that by including some states where the Senate races are, at best, marginally competitive to date (Minnesota, Virginia, Oregon, and New Hampshire immediately come to mind). Even at that, however, the average Obama share of the vote in those states was hovering around 45 percent, which is a full six points below Obama's national 2012 performance.
So, taken together (the House districts made up slightly more of the sample than the Senate districts), it wouldn't be unreasonable to estimate that the electorate for this poll would come
in around 51-47 in favor of Mitt Romney, in terms of their 2012 electoral preferences (See addendum at the end of this piece).
Viewed through that lens, the polling outcomes in this Politico poll actually aren't too bad for the Democrats. Obama's 43 percent job approval looks lousy, for sure. But, while it isn't something you'd necessarily pin on the refrigerator, it is roughly four points below how this sample presumably voted in 2012. In recent national polling, the president has run anywhere from five to 11 points behind his 2012 performance nationally. So this falls right in line.
And a two-point generic ballot edge for the GOP looks less impressive when you consider that, in all probability, this was a group of respondents that had more Romney voters than Obama voters, so the Republican edge here looks less impressive.
What this poll does tell us (besides a fairly fascinating set of data saying that, regardless of party, high-profile endorsements are a net drag on a candidate) is that the Senate map sucks for Democrats. Which, of course, is nothing that close followers of electoral politics haven't known for a very long time.
Addendum: (5:13 PM PST) An important note from the team at team at HuffPollster:
Keep in mind that Politico's sample consisted only of voters in 16 states and 66 U.S. House districts with competitive races in 2014 (voters were selected proportionately for the initial sample, so that a large state like Michigan would contribute roughly six or seven times as many registered voters as a smaller state like New Hampshire).So, the 2012 numbers would be a little better for Obama in that respect, when retrofitted for size. Still, HuffPo's Rachel Lienesch estimates that the sample did vote for Romney in 2012, but by a smaller margin (she estimates it at 0.3 percent). So, while the larger point is still valid, the margin separating the "competitive races" sample from a larger national sample are a few points less broad than previously thought.