Search

It’s election year: Generic Ballot and the House

Happiest of starts to 2022!


It is officially the start to a midterm election year… We’ve got primaries and debates and redistricting… We’ve got narrow margins in the House, and narrower margins in the Senate… We’ve got a lot of things to cover this year, but the first thing I’ll tackle is how to best answer a question which I’m getting asked a lot lately: Who gonna’ win that House?


To answer this question, I think it’s important to show House election outcomes in the midterm after a partisan transition of the White House. That sounds convoluted but it’s to say if the White House just switched parties, what can you expect in the following midterm?


Ok, the below bar chart shows historically what has happened in the House after a transitional presidential election.

*2004/2006 may be a stronger example of transitional elections due to 9/11/2001


So, in the past 4-‘ish’ transitional elections, on average, the current President’s party losses about 47 seats. Is that to say that this year Democratic members can expect to lose around 47-seats in the house? Absolutely not. It’s super early, and our historic ‘sample size’ is only 4 elections. Anything could happen.


BUT, what I’ll be looking at this year is how the generic ballot is tracking with that of previous transitional midterm election year’s generic ballots to best understand how ‘average’ this election may be in terms of a historical context.


For reference, the ‘generic ballot’ is a question that survey researchers ask to best understand who voters would support between a ‘generic’ (meaning no name given) Republican or Democratic candidate. For example, Gallup usually askes the question “If the elections for Congress were being held today, which party’s candidate would you vote for in your congressional district (the Democratic Party’s candidate or the Republican Party’s candidate)?


Let’s say the response to that question was that 50% of voters were to say ‘Democratic’ and 50% would say ‘Republican’, then the generic ballot would be ‘even’. If 55% said they’d support a Democrat, and 45% said Republican, then the generic ballot would be read as ‘D+10’, indicating that the Democrats have a 10-point advantage.


Looking back at Gallup’s generic ballots in historical transitional midterm elections, which they usually ask in late October or early November:


1994: R+7

2006: D+11

2010: R+9

2018: D+7


Now we can see how tracking generic ballot may help tell us what we can expect come election day.


1994: R+7 -> Republicans pick up 54 seats

2006: D+11 -> Democrats pick up 31 seats

2010: R+9 -> Republicans pick up 63 seats

2018: D+7 -> Democrats pick up 41 seats


Right now RealClear’s average has it at around R+2. FiveThirtyEight has it around R+1. If I were asked today who gonna’ win that House? I’d guess the Republicans would pick up the required seats to flip the House. BUT IT’S ONLY JANUARY PEOPLE! What if in a couple weeks from now it's D+8? The generic ballot will continue to fluctuate, and with it, a strong part of my election forecast.


One more thing to take notice of with this data: When generic ballot was R+7, Republicans picked up 54 seats. When generic ballot was D+7, Democrats only picked up 41 seats. There’s a slight disparity there. If we were to look back at the 2012 election (not a midterm or transitional election, but trying to make a point), Gallup had the generic ballot at ‘EVEN’, and in that election, Democrats picked up 7 seats in the house, and 2 in the Senate… This is all to say that if in an election night no seats were to change hands, the generic ballot would probably favor the Democratic candidate by maybe 2 or 3 points (there’s a couple reasons behind this, if anyone wants to get into the weeds, email, we’ll chat).


… I should also mention that all these historical points are far better used for understanding what will happen in the House rather than the Senate. The reason is fairly simple: the House has 435 races in a given midterm, vs. the Senate where you may only see around 33 races (that’s why it’s called the Law of Large Numbers!!!).


In coming posts I’ll point out other things to keep an eye on to best forecast an election, including one on the Senate races I’ll be watching… I friggin’ love election years.