Midterms: Anything Could Happen

Even more than usual, I mean.

It’s September, and I know longtime readers are wondering: where’s De Civ’s usual election nonsense? The political horse race maybe shouldn’t really be within this blogletter’s ambit, since this is a blogletter about civilization and the rule of law. Election outcomes are a symptom of our civilization, not its cause. I would do more to illuminate the causes of our political dysfunction by writing a review of Spider-Man: No Way Home (“sweet-smelling, soul-killing poison”)¹ than by writing about the forthcoming elections.

But here’s the thing: I really like elections. They are the one and only sport I avidly follow. So I write about them anyway. I hope you enjoy reading about them.

Image from FiveThirtyEight’s best-on-Earth election forecasting model.

Here are three election scenarios that strike me as about equally likely.

I’m going to repeat that before I get started: I think these are equally likely election outcomes, given the information available today.

SCENARIO 1: Polls Hit

The polls are dead-on. Republicans lose the Senate but take the House—narrowly. Both sides win some of the tossup seats in both houses. For example, Adam Laxalt (R-NV) wins his tossup Senate race, but Raphael Warnock (D-GA) manages to hold on in his. This is what conventional wisdom says to expect, and, for once, conventional wisdom and polls align pretty nicely. After all, the polls are right sometimes, including in 2018, 2010, and 1998.²

Having won the House, the GOP cannot move partisan legislation, but can block Dem partisan legislation, its cooperation becomes essential to passing a budget, and it is able to oversee and investigate the Biden Administration. However, President Biden still has full control of appointments. Practically speaking, very little changes. “American national governance” continues to be, primarily, an argument between the administrative state and the judicial branch, but Republicans get to be gadflies.

SCENARIO 2: Polls Miss Red

The polls are missing certain Republican voters, as they did in 2020, 2016, 2014, 2010, and 2004. Whether that’s because they don’t answer polls, because they lie in polls (the “shy Tory” theory), or because undecideds broke mostly GOP in the end… well, the pollsters can argue later about “why”. They’ve been arguing for years about why they keep missing Republicans, and, while they do, Republicans keep performing unexpectedly well in elections, and that guy from Trafalgar Group gets to keep running victory laps.

The upshot is that, once again, the polls underestimate GOP support, this time by a modest 3 points (on average). They miss by even more in white working class battleground states. The GOP walks away with a House majority large enough to rival its 2010 triumph and wins all the tossup Senate seats.³ Republicans end up with 52 Senate seats, after even Blake Masters scrapes out a win (but Dr. Oz still loses).

Despite winning a popular majority, the GOP still cannot move partisan legislation (due to the presidential veto, which I oppose), but can block appointments, particularly judicial appointments, that it finds unsuitable. Republicans go into 2024 in a very comfortable position… not least because the GOP will have now confirmed that it doesn’t need Donald Trump on the ballot to get his coalition out to the voting booths.

SCENARIO 3: Polls Miss Blue

Another possibility is that the polls are missing certain Democratic voters, as they did in 1996, 2008, and 2012.⁴

There's clearly a lot of Blue Energy right now. We can see it vividly in recent special elections. Polls are seeing that energy, but polls may not be capturing all of it (see those previous elections). Moreover, there’s a lot that’s weird about this election. Midterms are usually referenda on the President, but the former president, Mr. Trump, has refused to retire from the spotlight as past presidents have done. This changes the game. Indeed, Mr. Trump has done a great deal to make this election about him instead—which turned out to be a losing proposition at the 2018 and 2020 elections.

Democrats also recently helped redraw Congressional maps for the first time since 2010. Both parties gerrymandered like crazy, and the result is a pretty balanced House map. The small built-in GOP House advantage from geography alone has been erased, and the House now tilts verrrrrrry slightly blue. Thanks, gerrymandering! (The Senate, of course, uses static state boundaries and cannot be gerrymandered.)

Meanwhile, Roe v. Wade was overturned. Americans still overwhelmingly have no idea what Roe said or did, but they are clearly upset that it got overturned. Lurid news coverage hasn’t helped. Nor have flat-footed red state governments ineptly defending certain state abortion bans that are good and humane, but indisputably unpopular. The last time the Supreme Court handed down a decision with this much immediate political impact was likely Brown v. Board of Education. That could turn this election from a referendum on the President into a referendum on the Supreme Court. Polls often struggle to capture the full extent of reaction to unprecedented events… and the fall of Roe has already helped Dems more than I expected it would.⁵

The upshot is, the anti-Trump/pro-abortion wealthy suburbanite coalition gets even stronger than it was in 2018. Meanwhile, without Trump himself on the ballot, the Trump coalition doesn’t show up in enough force to counter the Blue Wave. Polls end up underestimating Democrats by 3 points, on average. Democrats slightly grow their House majority, win all the close Senate seats, and even manage to knock out J.D. Vance in Ohio. Dems end up with 54 Senators.

52 of those senators are eager to abolish the legislative filibuster. They do so immediately.

Without the legislative filibuster in their way, Democrats have free rein. They swiftly pass…

• national voting legislation (abolishing photo ID requirements nationwide),

• abortion legislation (abolishing all regulations of abortion at any stage of pregnancy nationwide; this includes striking down popular state regulations like parental notification and waiting-period laws, gutting conscience protections, and restoring taxpayer funding for Medicaid abortions that has been suspended since 1980),

• the Equality Act (suspending the bipartisan Religious Freedom Restoration Act with respect to LGBTQ issues, among other things),

• various gun regulation bills,

• a likely bill making Washington, D.C. a state (maybe Puerto Rico, too).

• possible major new spending. Although I like to think they're sane enough not to do that in the face of inflation, Sen. Elizabeth Warren has already indicated her support for universal child care and additional student debt reduction.

If the Supreme Court gets in the way of this new Democratic majority, the Democrats would just pack the Supreme Court. They will have the votes, so why not? Who’s going to stop them? They won, fair and square!

***

These three outcomes are, again, about equally likely, in my mind.⁶ The race is very close, modest polling errors (including the modest errors suggested in this post) are normal, and polarization means small polling errors have large national reverberations.

Nevertheless, if either Scenario 2 or Scenario 3 came to pass, it would come as a huge shock to a lot of people, and media would represent it as historic. (Well… if the Democrats win. If the Republicans win, media will represent it as a historic catastrophe for America, replaying the 2016 greatest hits.) Why is that? The “Republicans and Democrats split Congress” outcome is technically the most likely outcome, yes… but that “most likely” still only happens about one-third of the time! We have about a two-thirds chance of waking up the day after Election Day to full control of Congress by one party or the other!

Between the three outcomes, it’s really a coin flip. It will seem inevitable the day after, and we’ll all construct narratives about how it was always bound to go the way it did, but, here and now, we just don’t know.

I'm finding this stressful. All I want right now is for the pro-life movement to be given time and space to consolidate itself in a number of pro-unborn “free states.” We need to show voters that, yes, a pro-unborn legal system can work (and even has some nice side benefits for grownups) before we’ll be able to change hearts and minds in, say, California. I’d rather have control of the Senate, too, as “Clarence Thomas Death Insurance,” but I’m not greedy.

There's a two-thirds chance I get my wish. Republicans just need to win enough of Congress to block Democratic court-packing legislation and make filibuster abolition unattractive. Then I breathe a small sigh of relief and nothing really changes. That would be peachy, in my book.

But then there’s a one-third chance that, instead of “nothing changes,” the world ends.

Despite that, I think I’d still rather be a right-winger than a left-winger right now. If left-wingers don’t win this election outright, right now—Scenario 3—they are very likely to lose the Senate badly in 2024 (it’s a good map for Republicans), and Republicans stand a good chance of a filibuster-proof Senate majority in 2024, with an agenda of National DeSantisism. At that point, left-wingers would perceive the world as ending. So the stakes in this election are very high for progressives, nearly as high as they are for conservatives… and they only have a one-in-three chance of getting through it in one piece. Those are much better odds than they had six months ago, but still long odds.

I’m largely just stress-typing here, but, if you take away anything from this article, make it this:

We have amassed an enormous amount of data about the upcoming midterm election—arguably, more data than has ever been assembled for a non-presidential election anywhere on Earth ever in history. That data, taken together, is confidently flashing a single, clear message, which we can be very confident in:

"Anything could happen in the midterms, from a Red Wave to a Blue Wave to the status quo, and anyone who predicts the outcome with any confidence is selling you snake oil.⁷"

This is frustrating and unhelpful. But, sometimes, the only thing the data can tell you is that the question you desperately want answered isn’t merely unknown, but unknowable.

1. I feel kinda bad about this Spider-Man take. Firstly, Spider-Man is my favorite comic-book-movie series, I’ve seen every single one in theatres, including both Andrew Garfield movies, and No Way Home is clearly a valentine to fans like me. Secondly, after years of progressives and conservatives inexplicably polarizing along political lines to hate each other over various sequels, from Ghostbusters to The Last Jedi to Star Trek Discovery to The Rings of Power, everyone finally got together and agreed that Spider-Man: No Way Home was great. They are, unfortunately, wrong!

2. Polls also did okay in 2006 and 2000, but results were mixed. For example, in 2006, the generic ballot overestimated Congressional Democrats by a wide margin, but underestimated Democrats in the Senate, while gubernatorial polling was accurate.

3. People always act like it’s a big shock when one party wins all (or nearly all) the tossup seats in a given election, but it’s actually the more common outcome. Parties only occasionally “split” the tossup seats.

Why do people act like it’s a big shock, then? Three and a half reasons:

1. Political parties like to trumpet their massive successes (and even, sometimes, massive defeats), because they can fundraise off that. Also, politicians like to believe they are living at the crossroads of history. So they inflate their wins.

2. The news media likes to exaggerate the drama of an election because it sells papers and scores clicks.

3. People fail to notice this pattern because people are very bad at thinking about probability in general, and haven’t spent much time studying election forecasting specifically.

4. Conventional wisdom favors moderate outcomes, and people will believe conventional wisdom over data nine times out of ten. (*Not a real statistic.)

5. In defense of all of the above, splitting the tossups used to be a bit more common, because lower polarization in the past meant polling error was less correlated across different elections.

4. Did you never notice the polls missed in those years? That’s probably because Democrats were already expected to win all three races; they just won by larger margins than expected. No actual outcomes changed.

This is different from 2016 and 2020, which made waves not because of the size of the polling error, but because the error changed the outcomes in a number of key races—including delivering the presidency to Donald Trump in ‘16 and turning his re-election campaign into a photo finish in ‘20. That’s why most people remember 2016 as a uniquely big polling miss, even though 2012 was almost as bad, and 2014 objectively worse.

5. Yes, this is a blogletter that deliberately calls attention to its author’s past incorrect predictions!

6. There are quite a few other outcomes that are still plausible, but less likely. It would be mildly surprising, but not shocking, for Republicans to have such a good night that Bolduc wins in New Hampshire… or for Democrats to have such a good night that Ron DeSantis loses in Florida.

7. Unfortunately, the day after Election Day, exactly one-third of those snake oil salesmen will have been proved correct, and they will all get treated as prophets for an election cycle or so. But we can worry about that later.