Yesterday the NYT (with Siena College) released a group of head-to-head polls in key swing states. Perhaps you saw this tweet?
Then today, the Washington Post released a series of national head-to-head polls showing significant leads for the top five Democratic candidates over Trump. Biden, +17! Warren, +15! Sanders, +14! Buttigieg, +11! Harris, +9!
OMG So Confused, Tell Me What To Think!
Well, there’s two things here. Let’s get the easy one out of the way first: single polls are single polls. Don't blow them out of proportion. As evidence, here are all the head-to-head polls that are a) from October/November and where b) the pollster has some sort of rating with 538:
Iowa Biden Warren Sanders
NYT (A+) -1 -6 -3 registered voters
Emerson (A-) -1 -2 +1 registered voters
Wisconsin Biden Warren Sanders
NYT (A+) +2 -2 0 likely voters
NYT (A+) +3 0 +2 registered voters
Marquette (A/B) +6 +1 +2 registered voters
Fox News (A-) +9 +4 +5 registered voters
N. Carolina Biden Warren Sanders
NYT (A+) -2 -4 -4 likely voters
NYT (A+) -2 -3 -3 registered voters
PPP (B) +5 +3 +3 "voters"
Florida Biden Warren Sanders
NYT (A+) +2 -4 -2 likely voters
NYT (A+) +2 -4 -1 registered voters
UNF (A/B) +5 +3 n/d registered voters
Pennsylvania Biden Warren Sanders
NYT (A+) +1 -2 -1 likely voters
NYT (A+) +3 0 +1 registered voters
Michigan Biden Warren Sanders
NYT (A+) +1 -4 +3 likely voters
NYT (A+) 0 -6 +2 registered voters
Emerson (A-) +12 +7 +14 registered voters
Arizona Biden Warren Sanders
NYT (A+) +2 0 -4 likely voters
NYT (A+) +5 +2 -1 registered voters
Emerson (A-) 0 0 -3 registered voters
OK, I Feel A Little Better. What’s The Second Thing?
Sorry, but the second thing won’t make you feel better. For starters, it isn’t necessarily contradictory for one candidate to lead national polling but still lose individual states and the Electoral College. Clinton beat Trump in the popular vote by 2%, after all.
To that point, though, the WaPo poll shows head-to-head advantages far in excess of 2%. Why would we not see this reflected in state-level head-to-heads? The NYT’s explanation is that white, working-class voters are not abandoning Trump in these states.
Meanwhile, Trump’s approval rating is historically low. Could it be that anti-Trump sentiment is driving up Democrats’ numbers against Trump in states that are already blue? Could a Democrat win the popular vote by an even larger margin than Clinton yet still lose in the Electoral College to Trump?
Let’s take a look at some states that Clinton won and how the top Democratic candidates are doing against Trump:
Minnesota: Clinton over Trump by 1.5%
Oct 21 Star-Tribune/Mason-Dixon (538: B+):
Biden +12
Warren +11
Sanders +9
Nevada: Clinton over Trump by 4.5%
Nov 3 Emerson (A-):
Biden -1
Warren -2
Sanders EVEN
Washington: Clinton over Trump by 16%
Oct 31 PPP (B):
Biden +22
Warren +23
Sanders +21
California: Clinton over Trump by 30%
Oct 22 SurveyUSA (A):
Biden +27
Warren +21
Sanders +26
Maine: Clinton won popular vote by 3%, though Trump got 1 EC from ME's 2nd District
Oct 15 PPP (B):
Biden +12
Warren +10
Sanders +10
Virginia: Clinton over Trump by 5%
Oct 15 Virginia Commonwealth University (B/C):
Likely Voters:
Biden +5
Warren +3
Sanders -3
Connecticut: Clinton over Trump by 14%
Oct 9 Sacred Heart University (B/C):
Biden +19
Warren +14
Sanders +16
While Minnesota, Connecticut, Maine & Washington support our theory, Nevada, California & Virginia do not. Personally, I’d discount California’s numbers since the margin there is so high anyway.
This, however, begs this question: what about states that Trump won by similar margins to those polled by the NYT, but which also have potentially more diverse voter bases? States like Georgia, Ohio and even Texas (Trump won by 9%) all fall within the same range of victory as the NYT states (e.g. Trump won Iowa by 9.5%).
Texas: Trump over Clinton by 9%
Nov 4 YouGov (538: B-):
Biden -7
Warren -7
Sanders -5
Ohio: Trump over Clinton by 8%
Oct 14 PPP (B):
Biden +2
Warren EVEN
Sanders EVEN
Oct 3 Emerson (A-):
Biden +5
Warren +3
Sanders +6
(Sadly, there’s nothing particularly recent for Georgia.)
Great, I Feel Worse. What’s The Silver Lining?
I take away a few things from this analysis:
1. The General Election is still a year away. Don’t blow individual polls out of proportion.
2. There’s room for optimism in a number of these polls, on the basis that states Trump won in 2016 look like they’ll at least remain in play, or possibly be difficult for him to defend.
3. National polls can be misleading, and Democrats need to be very careful not to get a false sense of security. We’re fond of thinking that a national lead of X shifts any number of states below the “blue line”, but the NYT polling, at least, is a data point against this thinking.
And of course, there’s always a chance Trump will be impeached and all of this will mean nothing. :D