10 Reasons Trump is Not Getting Reelected

The conventional wisdom seems to be that despite his awful approval numbers, Trump is likely to win reelection in November.  This seems absurd on its face, but some folks seem awfully sure about it.  And it's not just blustering on the right; there are plenty on the left who've decided that we're doomed as well.

This may be some sort of defense mechanism. It's easier to be the cynical cool kid who insists that we're going to lose, and then be pleasantly surprised when and if it doesn't work out that way.  And a lot of people are still shell-shocked from 2016.

Well, I'm not buying it.  I've done a deep dive on a lot of different platforms (like and follow all of them), and I'm here to tell you that while it's not impossible, Trump's chances in November are pretty awful.

No big math calculations.  Just a few takeaway points, a figure or two, and some general, numerically justified, reasons for optimism.

1. Being the incumbent isn't as helpful as you think.

I did a whole thing about this over at Daily Kos, but in short, only about 60% of presidential incumbents win reelection, and that's if they make it to the general.  On the face of it, that's barely better than the 50-50 odds you get by just being in the race in the first place.

2. A lousy first election is not a great sign for reelection.

Lest we forget, Trump lost the popular vote by 2.1%, and only won the Electoral College by one of the narrowest margin in history.  Well, it turns out that you don't just get a clean slate.  Presidents who won by a wide margin, generally do better than presidents who just squeaked by.  The correlation is low (about 0.2), but not insignificant.

3. He did worse than typical Republican candidates in 2016.

There's a bizarre misconception floating around that somehow Trump has an electoral magnetism that can't be resisted, but the simple fact is that not only did Trump lose the popular vote in 2016, but district-by-district he underperformed Republican Congressional candidates (incumbents and challengers alike) by an average of about 2 points.

Less quantitatively, there's also the fact that when Trump has tried to nationalize races – Roy Moore (AL-Sen), Rick Saccone (PA-18), Eddie Rispone (LA-Gov), Matt Bevin (KY-Gov) – his endorsements don't exactly seal the deal. And those are just the high profile races in very, very red districts that he's responsible for losing.

4. Even in 2016, he got lucky.

We can spend eternity debating all of the things that went wrong in 2016 (A 30 year campaign against the Clintons! Hillary skipping the rust belt! Comey! Complacence!) but the fact remains that the timing of Election Day couldn't have been worse for Hillary, and even then, Trump got lucky.  Here are all the states within a point and a half – essentially coin-flips on Election Day:
New Hampshire446.846.50.3
Much is made about MI, PA, and WI, but there's Florida as well.  There were 4 coin flips with high consequence states (flipping NH wouldn't have mattered) and Trump won all them. 38 electoral votes would have made the difference.  That's Florida, and any of the others or MI, PA, WI.

Run through all the permutations, and even including the "wrongness" of the polls, Trump only had a 50-50 shot of pulling it off.

5. Republicans are on their way out.

Look, people have been predicting the end of the Republican Party for decades, but in a real sense, it's happening. More liberal and browner populations are replacing older, whiter, more conservative ones.  Following 2016, I wrote a post in which I looked at demographics and voting patterns and found that based on nothing more than aging, dying, and acquiring new voters, Dems gain a net of about 2.4% nationally every 4 years!

Turns out that's a little optimistic, but it's still close.  In a quick facebook post, I looked at the last half-century of presidential elections (following the realignment of the Voting Rights Act), and found that on average, Dems gain 1.4%/cycle.

That means that following that pattern, if 2020 were a typical year, Dems would win nationally by about 8 points (with a scatter of 8 points).  Even if Trump were a "typical" Republican, he'd be unlikely to win the popular vote.  And lest you think I'm exaggerating, remember that Republicans have only won the popular vote once (!) after 1988.

6. Polls have been very accurate!

We get it. The polls were off in 2016 and to some people, this means that polls are meaningless and should be ignored for the rest of time.  I did a detailed response to this over at Daily Kos, but here's the short version:

  1. Even in 2016, the state polls were only off by 3 points.  The were off by a similar amount and in the opposite direction in 2012, at least nationally.  You just may not have noticed because Obama was expected to win, and then did win.
  2. In 2018 (which is more recent than 2016, so maybe it's time to update your criticisms), there were lots of polls and lots of results, and they were all very accurate, with a net bias of less than half a point:
  3. There have been lots of other high profile elections (specials, 2017 and 2019), and polls have also been doing great:
In a close race, sure, be nervous, but in general, if the polls say you're winning by 4 or 5 points or more, you should believe it.
7. The midterms, 2019, and the specials have been really good for the Dems.

The Dems flipped the House in 2018, and held on to senate seats in very red states like West Virginia, Montana, and purple "Trump" states like Pennsylvania and Ohio, very handily. We also flipped Nevada and Arizona. Meanwhile, despite gaining seats, Republicans didn't win in a single state that Trump didn't carry in 2016, most by very large margins.

There are now Democratic governors in Kansas, Wisconsin, Louisiana and Kentucky -- and if it hadn't been stolen outright, there would now be a Democratic governor in Georgia. There's a Democratic senator in Alabama. There's a Democratic congressman in Western PA, and one in Georgia's 6th. Maybe Republicans should come to grips with the fact that these elections aren't one-offs or flukes.

Tracking the special elections in 2019, for instance, there were dozens of races, and an average swing of 7 points toward the Dems compared to the presidential results in 2016 (when, remember, we won by 2 points nationally).  Not as large as in 2017 or 2018, but easily enough to carry the presidency.

8. Trump's approval is in the terlet, and that's a great predictor of outcomes.

Here is a current snapshot of Trump's approval from 538:

It's terrible – and very flat.  He has a net approval of -10.5.  In one of my DKos write-ups, I found that not only is approval a very good predictor for electoral success (with the exception of Truman, where the polling biases of not realizing that only rich republicans had phones in 1948 played a big role).

No candidate who was underwater has gone on to win the election, and on the whole, the more they are underwater, the more they tend to lose by.

This is not a difficult concept.  Unpopular presidents don't get reelected.

9. Trump is losing head-to-head against the frontrunner.

Like it or not (and I'm "meh" about it), Joe Biden is the current frontrunner. He's winning national polls in the primary by 10 or more points against Sanders and Warren.  That's not an endorsement, but that said, it makes sense to compare the person most likely to challenge Trump in the general election.  And even though there's been a slight dip in the last month or so, Uncle Joe is still beating Trump solidly and decisively.  Here's the current map in my high-quality tracker:

I've colored the states as "winner-take-all," but many are purpler than drawn. The upshot is that even with a recent downtick (and I suspect this is because of a dearth of polling), Biden has a 76% of winning if the election were held tomorrow.  Less than I would like, but still much higher than the 50-50 that the prediction markets are posting.

10. The Dems haven't consolidated, but the Republicans have.

We are in the Thunderdome portion of the Democratic primary.  The Iowa and New Hampshire primaries are in about a month, and in the meanwhile, the staunchest supporters of all of the frontrunners are trying to show that their candidate is the only one who could beat Trump.  The result: there is a small, but likely substantial fraction of poll respondents who are doing so strategically.

How hard is it to imagine a Biden supporter who would obviously vote for any Dem in the general saying that he'd vote for Trump (or is undecided) if Warren were the nominee?  You don't need very many such respondents to seriously cut into electoral margins.

After we have a nominee, the overwhelming majority of those strategic respondents will come home – unlike the Republicans who already have a nominee. This is why, in the higher quality polls, Trump is only pulling ~42% against all challengers, with a very high number of "undecideds."  They're decided, but want their nominee to win.

Think I'm just making this up?  As I mentioned above, incumbent net approval is a great predictor of electoral success. And if you look at the last couple of reelections (2004, 2012), you'll see that the head-to-head polling converges on the approval polling once the candidates are more or less finalized:

Be skeptical, sure.  But at this stage, I would much rather be the Dems than Trump.  And if you think I'm ignoring reality, stay tuned for: Responses to 8 Arguments that Trump is Going to Win.


  1. Thanks for this, Dave! It's good to step back from the day-to-day of the news and get a broader picture!


Post a Comment