Democrats Are Pulling Away in the Generic Ballot. What Does That Mean?

Despite all the unusual political events of the last few years, one thing in electoral politics is proceeding by the book: the growing Democratic advantage on the generic congressional ballot.

The generic ballot is a poll question that asks voters whether they will vote for Democrats or Republicans for Congress, and historically it has been a decent predictor of the House popular vote. Democrats now lead by more than 13 points, according to the FiveThirtyEight generic ballot tracker, after a CNN/SSRS poll on Wednesday gave Democrats an 18-point edge, 56 percent to 38 percent.

A 13-point edge would be similar to or larger than their advantage in 2006, when Democrats won control of both the House and the Senate.

Although the size of the advantage seems startling, a Democratic edge on the generic ballot is not that much of a surprise. All of the conditions for a big Democratic victory seem to be in place. The president’s approval rating is in the mid-30s. Republicans have full control of government heading into a midterm election. And an unpopular tax bill may be at least partly responsible for the increased Democratic edge over the last month or so.

It is a bad sign for the Republicans next November. It is probably enough to make the Democrats fairly clear if modest favorites to retake the House, despite their structural disadvantages in the chamber.

Here is what you need to know about the generic ballot:

Yes, the midterm elections are still 11 months away. But among the various measures of American public opinion, the generic ballot is about as stable as it gets.

As FiveThirtyEight’s Harry Enten wrote months ago, even the very earliest generic ballot polls tend to correlate with the final generic ballot polls. There’s a tendency for the polls to drift toward the party out of power — and that’s exactly what has happened over the last few months.

At this later stage, the generic ballot is even more stable than it was when he wrote about it in April. Since 1970, there has been surprisingly little movement between generic ballot polls taken between 221 and 421 days ahead of the midterms — we were at 321 Wednesday — and those taken over the final 30 days. The average shift is just 2.4 points.

I can’t think of any other measure of political sentiment that tends to be so stable. This is probably because the generic congressional ballot isn’t about a specific race. It’s a lot easier for news or events to move views of a specific race than to move the overall national political environment.

The tax debate may be the sort of thing that could yield a temporary shift in the generic ballot, especially since polls show that the Republican tax plan is unpopular. It is possible that Republicans could gain some ground back if the tax debate is responsible for the recent decline in their support.

Sometimes there are larger shifts. In 2010, for instance, the generic ballot moved decisively toward the Republicans between December 2009 and Election Day. This is in keeping with the tendency for the generic ballot to shift toward the party out of power; it is also a result of the switch from registered to likely voters in polling, which generally helps Republicans. But this is a reminder that there’s time for the generic ballot to move a lot between now and Election Day — even if it usually does not.

The final generic ballot polls — and, for that matter, generic ballot polls taken 221 to 421 days ahead of the election — tend to do a solid job of predicting the final national popular vote.

It’s not a perfect relationship, of course, and the record of generic ballot polling gets fairly sparse before 1990 or so. But even including the years of sparse data, the average of generic ballot polls over the final 30 days of the race misses the popular vote by only an average of 4.1 points. That’s not perfect, but it makes a 13-point lead look pretty strong.

Gerrymandering, incumbency and the tendency for Democrats to win urban areas by lopsided margins combine to give Republicans a considerable advantage in the fight for control of the House. It gives the party a chance to survive a so-called wave election, like the one that brought Democrats to power in 2006 or the one that swept Republicans to House control in 2010.

My view is that Republicans have a realistic chance to survive up to around an 8-point deficit in the popular vote. But it is very doubtful that Republicans would survive a 13-point loss in the popular vote.

There’s one potential bit of good news for Republicans: Historically, parties have struggled to match landslide margins on the generic ballot with landslide margins in the popular vote.

For instance, Democrats in 2006 had an 11.5-point lead in the final RealClearPolitics average on the generic ballot but won the popular vote by 8 points. Similarly, Republicans in 2010 had a 9-point lead and won the popular vote by 7 points.

Typically, there would be no reason to pay attention to this kind of difference between the polls and the outcome of so few elections. But the generic ballot is unusual: It’s generic. It doesn’t directly measure attitudes about the actual ballot that voters will cast. And some generic polls ask which party voters would prefer to see in control of Congress, not whether they’ll vote for Democrats or Republicans.

You can think of it as somewhere between a presidential horse race poll and presidential approval ratings. Both are highly correlated with presidential election results, but the latter is just a correlate of presidential vote choice, while the former is a direct measurement. It is well acknowledged that a 25 percent approval rating may signal only, say, a 12-point loss rather than a 50-point loss.

The empirical evidence does hint at bias in the relationship between the generic ballot and the popular vote, but it’s not conclusive. On the other hand, there’s a decent theoretical case that it’s close enough to a direct measurement of congressional vote choice that it can be treated as such.

Whether there’s such a bias makes a difference. If you assume that, over the long run, the generic ballot yields an unbiased estimate of the popular vote, the Democrats might be poised to do better than they did in 2006, for example, since Democrats won the popular vote by only 8 points. If it is biased, an 11.5-point Democratic generic ballot lead might yield only an 8-point popular vote edge, like in 2006, and the House might still be something closer to a tossup.

But today’s generic ballot polls give Democrats an even larger lead than the one they had in 2006 or that the Republicans had in 2010. The president’s approval rating is lower as well. That doesn’t guarantee that the Democrats will do better than they did in 2006 — or than the Republicans did in 2010. But it does mean that it’s the best estimate based on the available indicators of the national political environment. If it holds up, Democrats are favored to retake the House.