Right now most of the estimates are for the Democrats to win about 30 seats. Some estimates are higher (FiveThirtyEight has a median estimate of 38). My own estimate is for a Democratic gain of 30 seats, but with more close GOP-held seats in danger.
But two signs suggest that something bigger is in the offing. The first is shown in the table below. There has been a consistent gap between what the generic ballot polling is telling us and what district-level polling is telling us. Note something else: these leads are stable. There is no evidence in the House polling that Kavanaugh has much effect one way or another.
| Democratic Estimated Lead | ||||
| Number of Polls (last 30 days) | Since 9/1 | Last 31 days | Last 10 Days | |
| Generic Ballot | 58 | 7.37 | 7.24 | 7.50 |
| Non-Partisan House Polls | 140 | 12.18 | 11.50 | 11.67 |
| Siena | 48 | 11.69 | 11.27 | 12.23 |
If you compare state polling with national polling in presidential elections, you find that the swing in the state polling is actually a better predictor of the final result than the national polling is. I discussed this phenomenon here shortly before the 2016 election. In 2012 I suggested the day before the election that state polling was pointing to a larger popular Obama margin that was indicated in National polling.
The other data point–more compelling for considering a larger wave than people are expecting–is to look at the individual seats in play. In the UK, parties plan their strategy based on target lists. They take the other party’s seats and list them from the most likely to be vulnerable to the least likely.
I have performed that exercise for the U.S. House using a common measure of partisan performance (the PVI). The table below is a list of some of those seats and the current polling margin.
For example, let’s consider two seats on this list: the second Congressional districts of Kansas and Maine. KS-02 ranks 121st on the list of Democratic seats to target. Put another way, there are 121 other GOP seats that you would expect to find that are more vulnerable. Similarly, ME-02 ranks 67th on that list. And yet BOTH seats are close. In KS-02 the GOP leads by only 0.7 percent in polling. In ME-02 the Republican is 0.8 percent ahead.
What is important to understand here is that 54 seats between KS-02 and ME-02 were we have no polling. Some oddsmakers are using things like fundraising or the generic ballot to try and predict the results in these seats.
But look at the table — the polling in these seats is very inconsistent. Any method that tries to predict the results in these seats is going to miss badly in some of them.
And that is the entire point here: the House polling is saying something very different that than the national polls are.
One of the best signs of a wave is races that should not be close, but are close. This list suggests that most predictions about the size of the Democratic victory may be far too small.
None of this analysis is certain. A 3-point shift in polling in the House is the difference between the Democrats taking 50 seats and falling short of the number they need to win. Nevertheless, the two signs I identified suggest that a wave is far more likely than a GOP surprise.
A wave most people, even the odds makers, are not going to see coming.