It is still true that the Democrats could hold both chambers. But optimism is relative, and much of the optimism over the summer was that things wouldn’t be as bad as other indicators (like presidential approval) might suggest. Over the past week or two, however, the slow leftward shift seen in the polls since the start of the summer has begun to stall or reverse.
While the Democrats will benefit from a surge of enthusiasm in November, that enthusiasm is no longer showing in the dynamics of the polls.
Sign up for How To Read This Chart, a weekly data bulletin from Philip Bump
Before going too far here, I recognize that the poll is an imperfect measure. I wrote about it; here’s how polling experts recommend thinking about polling and here’s a look at recent trends in polling errors. Polls are loaded – but what we measure below is not who will win but how things change.
So how do they change? Below is a graph of net margin in the average of generic FiveThirtyEight polls (the average of the percentage of people telling pollsters they will support the Republican on their House ballot versus the percentage saying they will support the Democrat) over the past three months. You can see the line start to drop in mid-July and continue to drop — that is, toward the Democrats — over the next two months. The black line entered blue territory, a Democratic advantage, in early August.
Since last week, however – and, really, since last month – it’s been flat.
This is an average of the polls. If we look at a specific poll, conducted regularly by YouGov, we can see a more pronounced shift towards the GOP in recent weeks. Across a number of demographic groups, there has been movement in the Republican leadership, including among men, women and seniors.
Note that survey experts I’ve spoken with generally prefer averages like FiveThirtyEight’s. The usefulness of the YouGov poll here is that it easily breaks down these demographic groups. It’s also worth noting that YouGov polls have generally been better for Democrats than the average itself.
The generic ballot average gives us a good idea of how the election might turn out, although in 2006 and 2014 it underestimated what Republicans would do. (In 2010 and 2018, it matched the national margin in all House races.) Individual race averages may be more informative — especially when considering the races that will determine Senate control.
Republicans only need to flip one Democratic seat to gain control. In the 10 closest races identified by FiveThirtyEight, each party holds five — but the Democrats lead in six.
Notice what has happened in Nevada, Pennsylvania and Wisconsin in recent weeks. These averages can vary a lot depending on new polls, but there has been an upward shift in each case.
There are other suggestions that the national image isn’t as good for Democrats as many Democrats have come to think. CNN’s Harry Enten points to a broad Republican advantage in handling the issue voters identify as most important to them. The New York Times’ Nate Cohn wrote last week about the growing importance to voters of issues that don’t stimulate the left – a point reinforced by a new poll this week from Monmouth University.
Again, relative to where the party was in the spring, Democrats are in a decent position. They could hold the Senate (if only 50-50, as they currently do) and the Republicans are unlikely to arrive in 2023 with a massive majority in the House. But what the polls are showing now is not what many Democrats had hoped for. Polls have spent a few months defying the downward pull of historical indicators. But this tug turned out to be strong enough.