November 10, 2019 📊 The 2020 election will be close (until it's not)
Plus, should we take Michael Bloomberg's (likely) entry into the Democratic primary seriously? What happened in KY last Tuesday? And what do swing voters want?
Welcome! I’m G. Elliott Morris, data journalist for The Economist and blogger of polls, elections, and political science. Happy Sunday! This is my weekly email where I write about politics using data and share links to what I’ve been reading and writing. Thoughts? Drop me a line (or just respond to this email). Like what you’re reading? Tap the ❤️ below the title and share with your friends!
Dear Reader,
This week’s main read: It is a year out from the 2020 election, so take this with a grain of salt, but new polling shows Trump and several Democrats tied in hypothetical matchups in the swing states. Biden is up 3 in Wisconsin. Sanders is up 1 in Pennsylvania. Warren is down 6 in Michigan. All within the margin of error—and while all are up ~7-10 nationally? Let me try to make sense of this for you.
Plus, I’ve got links to some data on urban-rural polarization in last week’s state-wide elections in Kentucky (and elsewhere), swing voters in swing states and Michael Bloomberg’s presidential run.
Thanks all for reading my weekly email. Please consider sharing online and/or forwarding to a friend. The more readers, the merrier! If you’re shy, the best way you can support my newsletter is to press the heart button below the title (this makes it rank higher in Substack’s curation). If you’d like to read more of my blogging I publish subscriber-only content 1-3x a week on this platform. Click the button below to learn more!
—Elliott
This Week's Main Read
The 2020 election will be close (until it's not)
A variety of data suggest a large split between the national popular vote and outcomes in the key swing states
OK, I’m going to go over a lot of data in a short amount of time, so strap in…
The New York Times released new polling last Monday that showed the leading Democratic candidates for 2020 essentially tied with President Trump in the swing states. Here’s the main graphic from their piece:

The NYT polls show Joe Biden running even with Trump in Michigan and beating him—albeit within the margin of error—in Pennsylvania, Wisconsin, Florida and Arizona. Bernie Sanders fares slightly worse, polling ahead of Trump in 3 of the 6 swing states, and Warren is down everywhere except Arizona (again, within the margin of error)
But I have something to note and then I’ll jump into commentary.
I think the differences between vote shares for Joe Biden and Bernie Sanders are notable, if not statistically significant. But I’m worried that Elizabeth Warren is not known enough to make these comparisons all that useful. In a poll from The Economist and YouGov last week, 16% of Democrats said they “don’t know” their opinion about Warren. Only 7% of Democrats said that about Sanders and 11%said so about Biden. These differences sound trivial, but if we’re removing ~10-15% more votes from the pool when comparing Warren and Trump versus Biden and Trump, suddenly the 3 percentage point gap in their support in Wisconsin seems more volatile. For this reason I think the Sanders versus Biden numbers are more stable than the Warren ones. This is not an argument to ignore the data, however—just to read with caution.
So here’s what I’m stuck on right now: The 2020 presidential election looks close y’all. And it’s not just state-level polling that suggest this is the case. National polls suggest an election in which the Democrats probably only have ~4 points to spare in Wisconsin (my #1 draft pick for the tipping-point state). The “fundamental” economic and political predictors of national presidential election point to a close race, too.
Let’s talk polls first.
Here are the latest averages of 2020 general election polls from RealClearPolitics for each of a Biden, Sanders and Warren v Trump matchup:



Although the most optimistic scenario for Democrats right now (a Biden versus Trump general election) gives them a ten percentage point advantage in the national popular vote, that likely doesn’t translate to a 100%-likely electoral college victory. Given that Wisconsin, Pennsylvania and Michigan lean 4 points to the right of the nation, these massive national leads only translate to ~5-or-so-point margins for Joe Biden in the swing states. And such leads in the polls would not predict a certain win for Biden, as we saw in the disconnect between state-level polls and results last time around. Sanders and Warren would face an even tighter battle there.
Importantly, this math is only valid if the relationship between national and state level outcomes next year matches the link between them last time around. In other words, Democrats’ large popular vote margins in the polls could be distorting the picture of the electoral college even more than we think if they have gained more voters elsewhere. Maybe, for example, in large population centers and, I don’t know, Texas? Point being, national polls largely suggest that 2020 will be close (though maybe not quite as close as the NYT/Siena polls suggest).
And then there’s forecasts. Oh, the forecasts…
Though I’ve talked on here before about how the “fundamentals” predictors for US presidential elections might not generate as reliable of forecasts today as they used to, they are nevertheless a useful benchmark. (Maybe they shouldn’t make up 100% of your final prediction, but even juicing 10% predictive power out of them would be handy.) Nathan Kalmoe, a professor at LSU, posted some data showing how GDP growth and Trump’s approval ratings today point to a razor-thin 2020 general election:

According to Kalmoe’s figures, Trump polling at -13 on job approval means predicted two-party vote share of 49% (with some uncertainty around the projection). As far as the economy goes, GDP growth of 2% suggests Trump would win 51% of the popular vote (again, with some uncertainty around the projection). I wouldn’t take either side of that bet if I were you…
Other forecasts also point to a close election (if you look at them the right way). One of such forecasts is the one generated by Rachel Bitecofer, a professor at Christopher Newport University. If you take her state-level predictions and simulate uncertainty around them, they only show that the Democratic candidate is favored to win the contest 64-75% of the time. This slide is from a presentation I gave at George Washington University last month:

The election is no sure thing. In many ways, it actually looks exactly like 2016—at least for now.
Maybe this is the new normal
As Wu-Tang-Clan said in their 1994 hit “C.R.E.A.M”, “Partisanship rules everything around me”. (OK fine, they didn’t actually say that, they were talking about money, but I’m saying the partisanship thing now so simmer down.) In an age where most Americans’ vote choices are locked-in by the party label that a candidate choses, it shouldn’t be surprising to us that opinion about the president has been relatively stable over the past 3 years. One’s partisan identity is now becoming one of their most influential social characteristics—on par with education or race and age. The near-immutability of that partisanship will come to force election outcomes closer to the partisan equilibrium of the country. Landslide elections—which result from voters (in key states) shirking off their partisan (again, social) identities and voting for the candidate on the other side of the aisle—might thus be a thing of the past.
And here are some selected links to the work I read and wrote last week:
Posts for subscribers:
November 8: Bloomberg schmoomberg. The former Republican mayor of NYC starts late and behind in the Democratic primary. Can he make up the ground?
Political Data
Nate Cohn (The Upshot): “A Sliver of the Electorate Could Decide 2020. Here’s What These Voters Want.”
Today’s America is so deeply polarized that it can be hard to imagine there are people who are really not sure whether they want to vote for President Trump or his Democratic rival.
But these “mythic,” “quasi-talismanic,” “unicorn” swing voters are very real, and there are enough of them to decide the next presidential election.
They are similar in holding ideologically inconsistent views, but they otherwise span all walks of life, based on an analysis of 569 respondents to recent New York Times Upshot/Siena College surveys in the six closest states carried by the president in the 2016 presidential election.
These voters represent 15 percent of the electorate in the battleground states, and they say there’s a chance they’ll vote for either Mr. Trump or the Democrat.

Nathaniel Rakich (FiveThirtyEight): “How Seriously Should We Take Michael Bloomberg’s Potential 2020 Run?”
In a field this crowded, entering the race in the high single digits wouldn’t even necessarily be a bad thing, but the problem is that it might be harder for Bloomberg to build on that support than it would be for other candidates. In an average of polls from January and early February, I found that 62 percent of Democrats knew enough about Bloomberg to form an opinion (which was pretty high), but his net favorability (favorable rating minus unfavorable rating) was only +11 (which was pretty low). As you can see in the chart below, Bloomberg was a real outlier — for as well known as he is, we would have expected him to be much better-liked, with a net favorability of about +35, not +11.
And history suggests Bloomberg’s low favorability ratings would be a major obstacle to winning the nomination. Our past research indicates that people who win presidential primaries tend to either be (a) already well known and well liked or (b) relative unknowns to start off the campaign. Only one nominee since at least 1980 has been in Bloomberg’s position (well known but not well liked), and that’s Trump himself.

Note: a recent survey from Morning Consult showed that a quarter of Democratic primary voters have an unfavorable view of Michael Bloomberg, the worst rating of any of the 15 current contenders that they polled.
Yphtach Lelkes on Twitter: The relationship between density and vote share has doubled in Kentucky since 2015:


Other Data
Noah Buhayar and Christopher Cannon (Bloomberg): “How California Became America’s Housing Market Nightmare”
This piece is filled with great reporting and even better graphics


Political Science, Survey Research, and Other Nerdy Things
Bradley Jones (Pew Research Center): “Democrats far more likely than Republicans to see discrimination against blacks, not whites”
Overall, the survey finds little change in attitudes about discrimination against most groups since March. Large majorities continue to say that Muslims (83%), gays and lesbians (79%), blacks and Hispanics (77% each), women (68%) and Jews (66%) face a lot or some discrimination in the U.S.
As in the past, there are wide partisan divides in perceptions of discrimination against most of the groups included in the survey. Republicans are less likely than Democrats to say there is a lot of or some discrimination against Muslims, gays and lesbians, blacks, Hispanics, women and Jews.
Democrats are less likely than Republicans to say there is at least some discrimination against evangelical Christians, whites and men.

What I'm Reading and Working On
I’m writing next week on politics and musical preferences, the education divide in American politics, Andrew Yang, and impeachment… It’s going to be a busy week. I’m reading a bunch of nerdy papers on variational bayesian methods (oooooh) and getting a bunch of data/programming stuff ready for the upcoming election in the UK.
Something Fun
This is a fun interactive on what people google about their pets.
Thanks for reading!
Thanks for reading. I’ll be back in your inbox next Sunday. In the meantime, follow me online or reach out via email. I’d love to hear from you!
If you want more content, I publish subscribers-only posts on Substack 1-3 times each week. Sign up today for $5/month (or $50/year) by clicking on the following button. Even if you don't want the extra posts, the funds go toward supporting the time spent writing this free, weekly letter. Your support makes this all possible!