You can’t repeat the elections past

So we beat on, electoral boats against the current, borne back ceaselessly into the past.

Scott Fitzgerald wrote that “the test of a first-rate intelligence is the ability to hold two opposed ideas in the mind at the same time and still retain the ability to function.” I would be among the last to know what goes through a first-class mind, but in analyzing elections we are frequently confronted with forces pushing in opposite directions and pointing toward completely different outcomes.

It comes down to how much weight to place on each argument.

In this case, one side of the equation centers on President Trump’s unpopularity. It’s a challenge for the GOP given that midterms tend to be referenda on the sitting president. Two national polling averages put Trump’s approval rating at around 38 percent and 39 percent. The Pew Research Center recently put Trump’s current standing in a historical context, finding it lower than any of his six predecessors at the same point in their terms, as well as lower than Trump’s rating at this point in his first. For perspective, President Carter had a Gallup job-approval rating of 39 percent in August 1978.

The common thread in midterms is that those who identify with or lean toward the president’s party are usually less enthusiastic about voting and turn out at lower rates than those who identify or lean toward the opposition party. Opposition voters are still angry over losing two years earlier, thus voting in higher numbers. Independents in the middle tend to get buyer’s remorse by midterm election time, breaking toward the opposition party in substantial numbers. That makes midterm elections a devil’s triangle for a presidential party.

Proponents of the Democratic “blue wave” case generally stop here, ignoring the other side of the equation. There is an ironclad law of politics: A party cannot lose seats it doesn’t have. Republicans are heading into this general election with the smallest House majority that either party has had in 94 years—Republicans had a single-seat advantage in the 72nd Congress from 1931 to 1933.

We have now had three elections in a row with little turnover in either direction. The victorious party emerged from each with between 219 and 222 seats; 218 are needed for a majority. The losing party finished with between 211 and 215. The country is remarkably evenly divided, and the House largely reflects that split. There is very little in the way of low-hanging fruit for either party to pick up, resulting in little electoral elasticity. Put another way, an enormous force would be required to push the outcome outside its narrow recent trading range. In the Senate, with the GOP defending only two seats outside reliably Republican-leaning red states, their vulnerability, at least on paper, seems small.

While historical averages are often cited as predictors of the next election, in citing examples from the 1970s, 1980s, and as recently as the early 1990s, one might as well be citing examples of when dinosaurs roamed the earth. Have politics and voting behavior changed that much? The shape and composition of the two major parties were dramatically different in those Jurassic times than today.

Those who were once conservative Democrats today are either dead or no longer Democrats. The same is largely true with liberal Republicans. Conservative and small-town/rural Democrats and liberal Republicans and those with urban constituencies were among the first to abandon their party in a bad year. Now, they no longer make up a meaningful segment of either party. Each side is more ideologically cohesive, reducing the downside risk for each party in a poor election cycle.

Now partisans mostly march in lockstep, and if unhappy with their own party or nominees, simply staying home is the most they will do. Defecting–that is,voting for the opposite party–is no longer something even to be considered. It is important to remember the arithmetic difference between a party identifier or leaner abstaining by simply not voting and a defector, actually casting a ballot for the opposite party. The former represents the loss of one vote, the latter means one side loses a vote while the other side gains one.

This dynamic will most likely play out in the Maine Senate race. It’s too early to accurately assess the impact of the Democrats’ candidate switch from Graham Platner to former state Senate President Troy Jackson, who will face GOP incumbent Susan Collins.

Jackson does not appear to have the personal baggage that Platner brought to the race, but neither does he possess the candidate skills that Platner had. A high-quality poll that was released last week showed Jackson leading Collins by four points, 49 to 45 percent. Hart Research, the polling company, noted that at this point during the Collins-Sarah Gideon race six years ago, the race was tied, and that while the new poll showed Collins’s job approval rating underwater–46 percent approving to 52 percent disapproving–their three polls from the summer of 2020 showed Collins approval and disapproval numbers at 48 percent. This is noteworthy because the public polling during that race six years ago consistently had Gideon well ahead, but private polling by firms on both sides for the campaigns, party committees, or super PACs showed the race considerably closer, thus the public perception of the race was quite different from what the pros were seeing.

An oft-repeated nostrum from pseudo-psephologists is that public polls–which are polls not commissioned by the campaigns, party committees, or super PACs–are more accurate and should be given more credibility than those “internal polls” done for interested parties. This could hardly be more misguided. Most of the public polling is done on a shoestring, while the institutions with skin in the game are mostly seeing data of much higher quality. It’s true that the internal polls that are released or leaked almost invariably show favorable numbers, but only a tiny sliver of private polling ever sees the light of day, and what is seen is only released or leaked because it serves some interest.

All of this is to say that we are now finally seeing high-quality public polling. Surveys released over the past two months from New York Times/Siena, Fox News and the AARP, are comparable in quality to the polling generally available to the candidates, parties, and super PACs. It will be interesting to see whether Republicans release their own data and when additional neutral polling data becomes reliable. Whether Jackson is four points ahead or whether this is simply a margin-of-error race, watchers would be well advised to avoid obsessing over the mediocre public polls that were so wrong in Maine six years ago and, more recently, so wrong in the Michigan primary.

None of this is to argue that Republicans will escape losses in November. But despite Trump’s horrific approval numbers, the GOP losses may well fall short of the post-World War II average: a net loss of 26 House seats and four Senate and gubernatorial seats.

Overall, my gut tells me that few Republicans will actually defect, but many will stay home, resulting in the GOP losing votes without Democrats gaining many.

There is no question in my mind that Democrats in a lot of red states will be overperforming, but when playing a tough schedule of road games, beating the point spread is a lot harder than actually winning the contest.

This is a tough election to read, and it reminds me of the story about President Harry S Truman getting exasperated by hearing economic experts say, “On the one hand…” And then, “On the other hand.” He eventually told an advisor to go find him a one-armed economist.