One constant feature of political analysis is the tendency for both pundits and prognosticators to overstate changes in the political environment. There are various incentives for this. One is that it makes an expert look stale and out-of-touch to lag the conventional wisdom. This leads people to overreact to trends in news coverage and polling data such that they can later say “See? I told you so!” if the trend continues. There is also a journalistic demand for news, driving reporters to over-correct to shifts in the prevailing political winds in their organization’s pursuit of new content to publish.
A different, psychological incentive is tied to what meteorologists and probabilistic forecasters call the “wrong-side-of-maybe” fallacy, which is when someone assumes a probability just over or under the 50-50 line equates to a 100% or 0% probability. In electoral handicapping, this translates to a tendency for small leads in polling to get exaggerated — something tied to our collective misunderstanding of the inherent uncertainty of political surveys.
Because political pundits and journalists are so prone to falling for these incentives and fallacies, more empirically driven writers and forecasters often end up playing a contrarian role in reflecting on the probable relative stasis in the horse race. In this way, we are fighting two fronts of the battle: On one side, we have to justify to supporters of trailing candidates why our forecast does not have them in the lead; on the other, we often have to tell supporters of the leader why they cannot take the lead for granted.
Today, I think we are at this juncture yet again. The Democrats have “rallied” out of a tough midterms position, saving themselves — as the now-conventional wisdom goes — from what was previously accepted as assured electoral defeat in November. They now have a chance to gain a seat in the Senate and may even hold the House, bucking historical trends.
Or will they? How much are the Democrats’ allegedly rosy prospects due to overreaction to new data by the usual suspects, versus real change in the race?
First, let me say that I do think there is a bit too much optimism among some Democratic commentators right now. For instance, claims that the party now has a solid lead in the national environment, ushering in a rare incumbent-party “wave” election, seem frankly divorced from both the history of midterms and the reality of current polling, which shows the party roughly at parity with Republicans on the House generic ballot and with shrinking leads in key Senate races, such as Ohio and Georgia. And while, as the Cook Political Report’s Amy Walter said to New York Magazine this week, “ there’s nothing wrong with ‘taking a hit of the hopium,’” it is possible to take things too far.
The conventional wisdom about the midterms has also been contaminated by selective back-forecasting from the perspective of early 2022’s political environment, too. Miscalibrated and misleading claims that the Republicans were guaranteed to win the House and Senate are now used as proof that the race has changed — instead of us pointing out that they were simply overconfident at the time.
The data do, however, show a clear story of Democrats gaining ground over the last two months — it’s just not as much as the Blue Wave Truthers would have us believe. The party has gained about 1-2 percentage points in vote margin in my average of generic ballot polls (which is a bit less aggressive than 538’s jumpy aggregate). The Democrats are also up compared to early July in Senate polls, by about the same margin — and even if I don’t believe, for example, that Tim Ryan is up by 4 points in Ohio, the overall trend in the data is clearly positive.
I should note that things get a bit trickier when we consider the so-called “fundamentals” of midterms. Wave elections tend to break against the incumbent party later in the year; in 2014 and 2010, Democrats lost 4 and 6 percentage points, respectively, of vote margin on the generic ballot between June and November. Republicans lost about one over the same time frame in 2018. While Democrats are defying gravity right now, it will get harder and harder to do so as time goes on.
But none of this is enough to dismiss the clear quantitative patterns describing the electoral horse race today. Democrats are certainly in a better position now than they were earlier in the summer. Forecasters now find ourselves in the rare position of agreeing with the typically hyperresponsive Washington pundit class. The question now is whether they can sustain their momentum.
2. Cook Political, Fivethirtyeight, and Larry Sabato are in general agreement - Democrats keep the Senate, may even pick up 1-2 seats, and probably lose the House.
3. It is equally as important or more important to focus on the races in the states for Governor, Attorneys General, Secretaries of State, and state legislatures. Keep in mind that gerrymandering and rules changes have made the states the battlegrounds for 2024.
Just in brief, Secretaries of States can halt the vote count - remember Florida Secretary of State Kathleen Harris in 2000? Did you know she was chairperson of Bush's Florida campaign?
The GOP has spent decades focusing on the states and in the past 20 years, the Democratic Party has lost 60% of state legislative chambers.
I think expectation management is key. I think there are a range of outcomes for the midterms. The Senate always looked competitive, but now that Democrats appear to be gaining in the polls, I've seen some predictions that are a bit out there. For example, do I really think Rubio is going to lose to Demings since Demings has gotten a few good polls during the last few weeks? No, I don't. I am also concerned about partisan non-response bias. If there is one, it could be overestimating support for Democrats. Overall, things could change, I am interested in seeing the post-Labor Day polls.
BTW, is the Economist going to release an election model for the midterms?
Hello! I enjoy a good Simon Rosenberg post as much as the next Democratic person who is hoping against hope, but one thing is odd to me, his insistence that Republicans are under 50%. Is that actually meaningful?
For basic party division of HR seats, I use the average seats after the past six midterm elections, 211 D - 223 R, which is what we might expect knowing nothing about the 2022 situation. Democrats and Republicans each had two surge years in this period, telegraphed in advance. In this period, only 2008 produced a surge linked to presidential victory; otherwise, parties have not held seats especially vulnerable in the subsequent midterm, true again in 2022. These basics would suggest a modest Republican gain, enough to control the House. Redistricting, presidential approval, candidate quality, fundraising, nature of the times, and the rest might usefully be considered in terms of how likely they are to move results from a partisan base.
1. Thanks for this perspective, Elliott.
2. Cook Political, Fivethirtyeight, and Larry Sabato are in general agreement - Democrats keep the Senate, may even pick up 1-2 seats, and probably lose the House.
3. It is equally as important or more important to focus on the races in the states for Governor, Attorneys General, Secretaries of State, and state legislatures. Keep in mind that gerrymandering and rules changes have made the states the battlegrounds for 2024.
Just in brief, Secretaries of States can halt the vote count - remember Florida Secretary of State Kathleen Harris in 2000? Did you know she was chairperson of Bush's Florida campaign?
The GOP has spent decades focusing on the states and in the past 20 years, the Democratic Party has lost 60% of state legislative chambers.
Cheers.
Hi Elliott,
I think expectation management is key. I think there are a range of outcomes for the midterms. The Senate always looked competitive, but now that Democrats appear to be gaining in the polls, I've seen some predictions that are a bit out there. For example, do I really think Rubio is going to lose to Demings since Demings has gotten a few good polls during the last few weeks? No, I don't. I am also concerned about partisan non-response bias. If there is one, it could be overestimating support for Democrats. Overall, things could change, I am interested in seeing the post-Labor Day polls.
BTW, is the Economist going to release an election model for the midterms?
I hope all is well!
Elliot
Hello! I enjoy a good Simon Rosenberg post as much as the next Democratic person who is hoping against hope, but one thing is odd to me, his insistence that Republicans are under 50%. Is that actually meaningful?
For basic party division of HR seats, I use the average seats after the past six midterm elections, 211 D - 223 R, which is what we might expect knowing nothing about the 2022 situation. Democrats and Republicans each had two surge years in this period, telegraphed in advance. In this period, only 2008 produced a surge linked to presidential victory; otherwise, parties have not held seats especially vulnerable in the subsequent midterm, true again in 2022. These basics would suggest a modest Republican gain, enough to control the House. Redistricting, presidential approval, candidate quality, fundraising, nature of the times, and the rest might usefully be considered in terms of how likely they are to move results from a partisan base.