Like a fresh visitation of a recurring nightmare, a new state-by-state New York Times/Siena poll informs us that Donald Trump’s lookin’ pretty good for reelection, particularly if Elizabeth Warren is the nominee. All the old slurs against Warren that Democrats were beginning to shrug off — she’s too left wing, she’s a girl, she’s a cold and unlikable girl, etc. — are heard once again in the voices of the white working-class swing voters we are told hold the nation’s destiny in their calloused hands.
The polling data isn’t especially positive for Joe Biden or Bernie Sanders, either, but the story comes across as a big red flashing sign to Democrats that Warren is too risky in a race that’s likely to be very close no matter who the nominee is. And across the left-of-center chattering classes, angst if not panic over these numbers is spreading:
As a public service, I’m going to push back against this gloom and suggest that it is premature if not misplaced, while acknowledging what we should have already understood and internalized about Trump’s underlying strength, at least in the Electoral College:
Yes, Trump Can Win Reelection, and Can Win Again While Losing the National Popular Vote
If some people reading the Times story were astonished to hear that Trump hasn’t already lost, they must not have been paying full attention. Yes, his job approval ratings, the most reliable indicator of his 2020 prospects, has been stuck in a relatively narrow high-30s/low-40s range for nearly all of his presidency. But most of the polls documenting his unpopularity do not (as is appropriate at this early juncture in the cycle) screen for likelihood to vote, which traditionally gives an edge to Republicans. If his approval rating rises just a bit, and add in minor party voting, you could easily see Trump finishing pretty much where he did in the popular vote in 2016. And as multiple analysts have explained, the poor distribution of Democratic votes (particularly unnecessary votes in California and narrowly insufficient votes in Texas) do indeed mean that Trump can register his party’s third Electoral College “coup” (to use the abrasive term Republicans deploy for another entirely constitutional but unusual mechanism, impeachment) in the last six presidential elections.
Nothing about Trump’s path to reelection is new, and the Times/Siena survey just paints a particularly vivid road map of another very close election.
No, Head-to-Head Polls a Year Out Are Not Terribly Reliable
One of the ironies of Cohn’s article is that he seeks to rebut a blizzard of head-to-head general-election polls showing multiple Democrats trouncing Trump (at least nationally, but also in some of the same Rust Belt states he thinks are again leaning to the GOP) with head-to-head general-election polls providing different results. Maybe the Times-Siena methodology is simply superior to everyone else’s put together, but that is at best a rebuttable presumption. The bigger issue is that this sort of research has a spotty record of accuracy this far away from a general election. Cohn anticipates this argument:
[O]n average over the last three cycles, head-to-head polls a year ahead of the election have been as close to the final result as those taken the day before. The stability of the president’s approval rating is a reason to think this pattern might hold again for a fourth cycle, at least for the three leading and already well-known Democrats tested in these polls.
Three cycles is not a very large sample. Here’s what Perry Bacon Jr. found with a broader lens:
In the runup to the 2016 presidential election, this same question came up, and FiveThirtyEight analyzed general election polls from 1944 to 2012 that tested the eventual nominees and were conducted in the last two months of the year before the election (so for 2012, that would be November and December of 2011). On average, these polls missed the final result by 11 percentage points.
Yes, partisan polarization probably means the ceiling and floor of support for either major party’s candidate is more limited than it was, say, in 1992. But the variable results different pollsters are finding for head-to-heads ought to create significant doubt that this one nails the results, particularly when it produces such questionable numbers as those showing Trump making significant gains among African-American and Latino voters post-2016.
Get back to us with fresh data in January when Democratic voters are actually voting, doomsayers.
Warren’s Standing Relative to Biden Is Marginally Different, and Has Been Steadily Improving
Any interpretation of the Times/Siena data that projects Biden as a clear winner and Warren as a clear loser against Trump ignores margins of error and the role name ID can play in polls at this point in the cycle. Among registered voters in the battleground states these polls focus on, Biden’s margin over Warren against Trump is one point in North Carolina and three points in Arizona, Wisconsin, and Pennsylvania. Yes, it’s six points in Michigan and Florida, but given margins or error and how far out we are, that’s not a lot. In addition, anyone attributing Biden’s better performance to ideological moderation should look at Bernie Sanders’s numbers: better than Warren’s, and in Michigan, better than Biden’s. Is Bernie perceived as more “moderate” than Warren, or is it a matter of gender, or of superior familiarity?
What we do know is that Warren’s “electability” credentials have steadily improved along with her name ID and her rise in the percentage of Democrats favoring her for president. The last national poll showing her trailing Trump was a Fox News survey in July. In August, Fox had Warren up by seven, in September by six, in early October by ten, and in late October by five. You can find a similar upward drift in her performance against the incumbent in virtually every outlet conducting regular polling.
None of this is to say that the Times/Siena numbers, or Cohn’s analysis, should be waved away. Data is data, and the best corrective for contradictions in the available data is more of it. But even if you buy the claim that Warren’s too risky a proposition at this point a year before Election Day, it’s not clear whether the problem is her ideology (and/or policy positions) or her gender or simply that she’s not sufficiently well-known. Of course she is coming across as more “left wing” now than she would in a general election; she’s not in a general-election contest at this point. It would be shameful if Democrats denied her the nomination because of her gender. Time alone will tell how well she wears on voters of all sorts — and in all sorts of states.