There’s A Better Case For A Top 2 Than A Top 3

Before last week’s debate, I argued that there was a lot of ambiguity as to who belonged in the top tier in the Democratic primary. Depending on which factors you emphasized, the top group could plausibly consist of any number of candidates from one (Joe Biden) to five (Biden, Elizabeth Warren, Bernie Sanders, Kamala Harris and Pete Buttigieg).

The Houston debate didn’t upend the campaign overnight; polls since the debate don’t show huge movement. You can, of course, find big swings for individual candidates in individual polls if you’re willing to cherry-pick, but you shouldn’t do that.

The post-debate polls do, however, reinforce some trends that were already in the works before:

They tend to argue for the presence of a top two (Biden and Warren) as opposed to a top three (Biden, Warren and Sanders), especially if you look at polls of Iowa.
They make it awfully hard to argue that Harris belongs in the top tier. (I recently argued that Harris did belong in the top tier, so put that take in the “didn’t age well” bucket, at least for now.)

Meanwhile, a new trend since the debate is that several of the lower-tier candidates, such as Amy Klobuchar and Beto O’Rourke, are showing slightly livelier numbers, although we’re still only talking the low-to-mid single digits.

Let’s touch on the new national and early-state polls, and then we’ll revisit each of these themes. There are quite a few of these: seven new national polls conducted entirely since the Houston debate, which variously contain good news for Warren, Biden and Sanders. For the gory details of which polls are included and what they say, see the footnote.1

We also have two new polls of Iowa since the debate, which look good for Warren and Biden but not particularly for Sanders; instead, those polls show Buttigieg and (in one of the two polls) Klobuchar outperforming their national numbers.2

As I said, that’s plenty of data to choose from — or to cherry-pick from — so this is precisely the sort of time when it’s useful to look at the polling average instead of individual polls. Since the new polls vary a lot in sample size and pollster quality, I’ve weighted the average using the same algorithm we use for our Trump approval ratings calculation:

Biden, Warren look strong in post-debate polls

Weighted average of national and Iowa polls since the Houston debate

National
Iowa


Candidate
Average
Trend
Average
Trend




Biden
30.2%
0.2
20.6%
-0.2


Warren
20.4
0.7
23.5
2.8


Sanders
16.6
-0.6
12.5
-2.2


Buttigieg
5.8
-0.1
12.5
2.8


Harris
5.8
-2.2
5.0
-8.2


O’Rourke
3.1
1.2
1.5
0.2


Yang
2.8
0.8
2.5
1.3


Booker
2.6
0.7
2.0
0.0


Klobuchar
1.4
0.3
5.5
2.4


Castro
1.0
0.1
0.5
-0.3

The trend reflects the average change as compared to the previous, pre-debate edition of the same poll, except in the case of the Civiqs Iowa poll, where it reflects a comparison to the average of all Iowa polls in August. Only candidates who participated in the third debate are listed in the table.

There’s not a ton of movement in the national polls. On average, Biden held steady at 30 percent. Warren and Sanders’s position isn’t much changed, but she does appear to be about 4 points ahead of him nationally and has continued to gain ground slowly but steadily, having now reached 20 percent in the polling average. The biggest gainer since the debate is actually O’Rourke, who added an average of 1.2 percentage points, although that still puts him only in the low-to-mid single digits. The biggest loser is Harris, who is down to only about 6 percent and lost another 2 points or so in the most recent round of polls, although some of that movement predates the debate itself.

In Iowa, the major headline is that Warren and Biden form a clear top two in the post-debate polls — a clearer top two there than in national polls. Iowa has always looked like a relatively strong state for Warren and a weak one for Biden, so this isn’t a huge surprise, but the trends are nonetheless favorable for Warren. The rest of the field also looks a bit different in Iowa than it does nationally; as I mentioned, the Midwestern candidates — Klobuchar and especially Buttigieg — are doing quite a bit better in the Hawkeye State. Sanders, meanwhile, is surprisingly weak in Iowa given that he nearly won the state last time around, although other Iowa polls have shown better numbers for him than the two released since the debate. The new edition of the Selzer/Des Moines Register poll is due out soon, which should help us to confirm whether the new polls are a trend or a fluke.

For me, that all adds up to something like this:

Nate’s not-to-be-taken-too-seriously presidential tiers

For the Democratic nomination, as revised on Sept. 19, 2019

Tier
Sub-tier
Candidates




1
a
Biden



b
Warren


2
a
Sanders ↓



b
Harris ↓, Buttigieg


3
a
Yang, O’Rourke, Klobuchar, Booker



b
Castro


4
a
Steyer ↑, Gabbard ↑



b
Everyone else

None of this looks dramatically different than before the debate, but to revisit those themes I touched on before …

Theme No. 1: There’s a better case for a Top 2 (Biden, Warren) than a Top 3 (Biden Warren Sanders). I’ve done something slightly sneaky with my tiers, which reflects the fact that Warren’s position continues to improve incrementally — rather than meteorically — relative to Sanders’s. As before, I have Sanders as the third most likely Democrat to win the nomination. (Keep in mind that these rankings aren’t driven by any sort of statistical model;3 they’re my subjective opinion, which sticks fairly closely to be polls but doesn’t omit other factors entirely.) But I’m demoting him to tier 2a from tier 1c. (I’m also demoting Harris, who has bigger problems than Sanders does — see the next item.)

You could argue that Sanders and Warren belong in the same tier since they’re almost tied in national polls. But they aren’t quite tied anymore; Warren is ahead in the average of national polls, as you can see above.

That alone might not be enough to put Warren in a higher tier than Sanders. But virtually every other polling-based metric other than the topline numbers in national polls tends to favor Warren over Sanders:

As I mentioned, she’s doing a fair bit better than Sanders in Iowa, especially in post-debate polls. (New Hampshire, where there hasn’t been any post-debate polling, is more of a mess.)
She has equal favorable ratings to Sanders and Biden, but lower unfavorable ratings, meaning that she’s more broadly acceptable to Democratic voters; she also tends to do better than Sanders when polls ask about voters’ second choices or which candidates they’re considering.
Although past movement in the polls doesn’t necessarily predict future movement, she obviously has the most momentum, having slowly but steadily gained in the polls for several months now.
In Iowa, and increasingly in national polls, she generates more enthusiasm than other candidates, including Sanders.
She does considerably better than Sanders among Democrats who are paying the most attention to the race, possibly a leading indicator of what will happen as more voters tune into the campaign later on.
And although the gap is closing, she’s still slightly less familiar to voters than Sanders, which implies that she has more room to grow.

The one factor working in Sanders’s favor is that his coalition appears to be a bit more diverse in terms of race and educational status; Warren is eventually going to have to expand her coalition beyond liberal, college-educated whites. On the other hand, Sanders has historically had problems with several major Democratic constituencies, including blacks over the age of 40, whites over the age of 55, and moderate voters of all races and ages. Overall, even from a “polls only” perspective, Warren is now in a meaningfully better position than Sanders.

And if you look at non-polling factors, the gap between Warren and Sanders increases. Empirically, the best way to look at the primaries apart from polls is through a “Party Decides” perspective, wherein so-called “party elites” — influential Democrats such as elected officials — are instrumental in picking the nominee, or at least are a leading indicator of voter preferences. Sanders is a problematic and even implausible nominee by this metric; he rather proudly doesn’t get along with the party establishment (and technically isn’t even a Democrat) and instead is running more of a factional campaign where he hopes to win the largest plurality of voters rather than necessarily uniting the party. Warren also doesn’t do terrifically well in this framework. (Party elites seem to prefer Biden and Harris.) But she has more endorsement points than Sanders does, and there’s lots of reporting to suggest that she has better relationships with party insiders than him. Particularly in the event of a contested convention, when superdelegates are allowed to vote on the second and subsequent ballots, this could be a big advantage for her. She’s also winning more support from progressive groups and organizations such as the Working Families Party and MoveOn.org which may or may not get along with the Democratic Party establishment, but can help to turn out voters in primaries and caucuses.

So while I don’t know that I’d go as far as prediction markets, which have Warren with a 35 to 40 percent chance of winning the nomination and Sanders at only 10 to 15 percent, there’s quite a bit to differentiate them, with most of the indicators in Warren’s favor.

As to whether Warren is more likely to win the nomination than Biden, I’m not quite ready to go there yet. (I still have Biden in tier 1a and Warren in tier 1b.) As my colleague Clare Malone likes to say on our podcast, Biden is still the “statistical frontrunner”: If you built a model consisting solely of objective indicators — e.g. national polls, Iowa polls, endorsements — he’d probably still be ahead. That isn’t necessarily the only way to look at the nomination. But I’d like to see Warren win a few more endorsements, and I’d like to see a couple more polls showing her leading in Iowa and New Hampshire, before removing Biden’s front-runner status. Warren is also near an empirical inflection point in national polls; once you get to 20 percent or so, you go from a factional candidate to a real threat to win the nomination. To the extent she can continue gaining ground — going from 20 percent to 25 percent, say — that would be a pretty huge deal for her, especially if those gains came from voters other than college-educated white Democrats.

Theme No. 2: Harris is in trouble. As I said earlier, I’ve mostly been bullish on Harris throughout the campaign. She’s a strong “Party Decides” candidate and she’s still second in endorsements behind Biden, although her pace of endorsements has slowed since a flurry of them after the Miami debate. But she’s in real danger now. Rather than uniting the party, she’s losing college educated liberals to Warren and black voters to Biden. She’s also well behind in polls of Latino voters.

There’s still plenty of time to turn it around, but the risk is that donors, voters and potential endorsers abandon her for a candidate who appears to have more momentum. (Harris’s fundraising totals for the third quarter, which closes on Sept. 30, will reportedly be weak.) Buttigieg has caught up with her to tie for fourth place nationally and is in a stronger position than her in Iowa and New Hampshire. And O’Rourke, Cory Booker and even Andrew Yang are right on her heels nationally. Media coverage is also a factor here: If media coverage that was once devoted to Harris begins to go to candidates such as O’Rourke and Booker instead, that could deepen her problems. And I’m not sure that refocusing her campaign on Iowa, her latest strategic gambit, is going to help.

Theme No. 3: There’s a little bit of life in the third tier. I don’t want to overstate this, since going from 2 percent to 3 percent means you’re still … at 3 percent. But (with the exception of Julián Castro) this is the best set of polling for candidates outside of the top 5 in a while. That’s not entirely surprising, since voters gave several of these candidates — notably Klobuacher, O’Rourke, Yang and Booker — relatively good marks for their debate performances. If anything, the problem is that a number of these candidates had strong evenings, so voters who were shopping around for new alternatives were divided several ways between them.

Perhaps the most interesting number of all in the new set of polling was Klobuchar’s 8 percent in the David Binder poll of Iowa; among other things, a Klobuchar mini-surge in Iowa could cause big problems for Biden by crowding the moderate/electability lane there. But that number hasn’t been replicated in other surveys yet.

All of this does leave one to wonder whether candidates such as Booker and Klobuchar were hurt by the large field and getting mixed up with the Marianne Williamsons and Eric Swalwells of the world. I’ll stick with what I said last week: If you’re polling in the low single digits, you’re probably toast. But let’s hold off on declaring their campaigns dead until at least the next debate.