This morning I was speaking (virtually) to a group and went into some detail about what has quickly become the most dominating factor in American politics—partisanship, or more accurately, extreme partisanship. But when time came for Q&A, one of the first questions got into the weeds of the 2022 midterm elections, 635 days from now.

So what do—or can—we know about the midterms? Generally speaking, there are four major factors: history, exposure, environment, and circumstances.

History. Since the end of World War II, elected presidents’ parties have suffered an average loss of three Senate and 22 House seats in midterms. First-term midterm elections more specifically have cost their parties one Senate seat and 23 seats in the House. But given Democrats’ wafer-thin margins in both the House and Senate, any loss is catastrophic.

Exposure. In other words, how many Senate seats does each side have to defend? Are they in friendly or enemy territory, or very competitive areas with a history of swinging back and forth? You cannot lose a seat you don’t have, and the more seats you have to defend—particularly above your ‘normal’ level—the more places where things can go wrong. How many of those seats are open is also important, as it’s usually easier for a party to hold onto an occupied seat. (One exception to that rule is that if there is an incumbent who starts out really damaged, perhaps by a scandal.) Next year, Republicans have 20 Senate seats up versus the Democrats’ 14. Even more importantly, four Republicans thus far have announced their retirements. Three of them are in potentially swingy states—Richard Burr in North Carolina, Pat Toomey in Pennsylvania, and Rob Portman in Ohio. No Democrats have retired yet, but the incumbent most likely to draw a tough race seems to be Maggie Hassan in New Hampshire.

Environment. Midterm elections are usually referenda on the sitting president or the party in control, so whether a president and his party have a tailwind or a headwind is very important.

Since Biden has only been in office for 22 days, it’s a bit premature to be making many judgements. Quinnipiac has his approval rating at 49 percent, NPR/PBS/Marist puts it at 50 percent, Monmouth slightly higher at 54 percent and Gallup higher still at 57 percent. He’s running a bit lower than Obama’s early ratings, but higher than Trump’s.

But the Republican brand is taking a beating these days. Another Gallup poll had some ugly numbers for the GOP, with the party’s approval sinking to 37 percent. But this far ahead of an election, these aren’t strong indicators.

Circumstances. Are there any relatively unique attributes or situations that may make an election deviate from historical trends? In 1998, for example, in President Clinton’s second-term midterm election, Democrats picked up four House seats while the Senate saw no net change.

But Democrats went into 1998 with far fewer seats than normal, having suffered their worst election in generations four years earlier. More importantly, the election was a referendum not on the president, but on Speaker Newt Gingrich and the Republican Congress, more specifically on their decision to impeach and try Clinton.

The very next midterm election was also an exception to the rule: The 2002 election was held just over 14 months after the Sept. 11 attacks. George W. Bush still enjoyed a 63 percent approval rating in the last Gallup poll before that election.

Yet another exception came in 2018. While the political climate was awful for the GOP, losing 40 seats and control of the House, they managed to pick up two Senate seats. Why? Because the one-third of seats up for election that year were mostly in very red states.

We’ll have to wait a few more months to see how the specific circumstances develop. In the meantime, political analyst Seth Moskowitz, writing for Sabato’s Crystal Ball at the University of Virginia’s Center for Politics, notes that we can watch the generic-congressional-ballot poll question as a key leading indicator. I would only add the caveat that the more gerrymandering that occurs nationwide during redistricting, the less of a relationship the map might have with the generic ballot.

Yet given how Democrats are hanging onto both houses of Congress by their fingernails, at this point, every factor is conceivably important.

This article was originally published for the National Journal on February 12, 2021.

More from the Cook Political Report

Trump Biden
2024 Swing State Project
First Person
Cook Politcal Logo
Larger logo
2024 Swing State Project
Cook Politcal Logo