Thread by @johnmsides

In advance of election night, I think it's useful to calibrate expectations to the fundamentals, not to polls or vibes.

So, a good calibration:

House: Dems lose ~40-45 seats
Senate: Dems lose 1-3 seats

Anything significantly less = Dem over-performance. Here's why.

A standard model for House elections is Gary Jacobson's: presidential approval, income trends, and # seats controlled by the president's party.

Forecast: Democrats lose 45 seats.

dropbox.com/s/95s7hip8bzg5…

A similar model by Lewis-Beck and Tien: -44 D seats in House and -5 in Senate.

If they factor in expert ratings by @nathanlgonzales, they estimate -37 D House and -3 D Senate.

centerforpolitics.org/crystalball/ar…

A model by James Campbell based off of @CookPolitical expert ratings: -36 or -42 Democratic House seats, depending on how you code the ratings.

For Senate, -1 Democratic seat.

The best case for Democratic overperformance is that the polls show them beating the fundamentals.

Per @AlanIAbramowitz, the current generic ballot (basically a tie) forecasts that Democrats would only lose ~16 House seats and maybe 0 Senate seats.

centerforpolitics.org/crystalball/ar…

The same thing is true if you look at @FiveThirtyEight's polls-only ("Lite") forecast: -15 Dem House seats and +1 Dem Senate seat.

projects.fivethirtyeight.com/2022-election-…

I don't have a strong prior about the "right" forecast. And even a small seat shift will obviously change party control of Congress and that's huge.

My only point: basic indicators suggest that 2022 should be bad for Democrats and you should calibrate your expectations to that.

So if the Democrats lose 20 seats in the House, that is bad for the party -- no question. But it is also better than they might have expected. Both things are true.