POLITICS APRIL 5, 2013
When House Democrats failed to win a majority last November, despite winning the popular vote, many people recognized how difficult it would be to realize President Obama’s dream of retaking Congress in 2014. Just how difficult, though? Ian Millhiser of Think Progress argued this week that, if the election were held today, Democrats would win the House popular vote in a landslide but take only a narrow 5-seat majority. According to Millhiser, Democrats command the support that ought to give them a large majority, but gerrymandering wrongfully denies them seats they deserve. For Democrats, the interpretation is certainly tempting. But it’s too simple. The available evidence doesn’t augur a Democratic landslide, and gerrymandering is not at all the sole culprit for the GOP’s structural advantage in Congress.
Why does Millhiser think Democrats would win in a landslide? After all, it’s been just a few months since Democrats narrowly won the House popular vote by about 1 percentage point. His analysis is based on a recent Quinnipiac survey showing Democrats leading the generic ballot by 8 points, 43 to 35. But there are three reasons to be skeptical of this interpretation. First, the Quinnipiac poll is somewhat of an outlier—in fact, it's the single best survey for Democrats this year. The latest Huffington Post Pollster average shows Democrats ahead by a more modest 5 points. Second, the Quinnipiac poll is of registered voters, not likely voters. While there’s nothing wrong with using registered voters this far out, especially in a presidential election year, it’s all but bound to overstate the Democrats in an off-year election. Third, generic ballot tests have tended to overestimate Democratic support in the national House popular vote, although that hasn’t been true in the last two elections (perhaps because of the house-effects of pollsters like Gallup and Rasmussen over the same time period).
If Democrats did manage to win the House popular vote by an 8-point margin, they would win a more comfortable majority than five seats. Millhiser argues that Democrats would need to win the popular vote by 8 points because Republicans won the median district by 7 points. But just because Republicans won the median district by 7 points doesn't mean that Democrats would need to win the House popular vote by more than that to carry the House. That logic--the uniform swing--applies well to presidential elections, but it's not so useful for congressional elections. If it did, Democrats would have struggled to win control of the House in 2006 and fallen well short of their eventual 30-seat gain. Between 2004 and 2006, for instance, districts swung between 34.6 points in the direction of Democrats and 19.2 points in the direction of Republicans.
Why doesn’t the uniform swing work in congressional elections? One obvious cause is open seats: when representatives retire, they lose the advantages of incumbency and seats can swing wildly as a district returns to its partisan leanings or toward a strong candidate. Another factor is variations in candidate quality and fundraising—one year, a representative might not face a serious challenger or might not even have any opponent at all; the next year, they might face a well-funded, credible challenger who had waited for the right moment. Even the handful of scandals every election cycle create additional opportunities unforeseen by the uniform swing, as Mark Sanford may demonstrate in South Carolina.
Demographics are another reason the uniform swing isn’t a good indicator for 2014. In 2014, minority support for Democratic candidates is unlikely to increase beyond 2012 levels. Minority turnout is also likely to decline in an off-year election, which will drive the white share of the electorate higher. As a result, a Democratic landslide would require disproportionate gains in white districts, since greater Democratic support among white voters will be necessary to make-up for lower minority turnout and few opportunities for additional improvement among minority voters.
It’s wrong to place all of the blame for all this on gerrymandering. The Democrats face two bigger challenges. The first is spatial inefficiency: the Democratic areas of the country are bluer than the Republican areas are red. Since "fair," non-partisan districts promote compactness and homogeneity, they tend to help Republicans by packing Democratic-leaning groups into solidly Democratic districts. The second problem for Democrats is incumbency. According to the latest but soon-to-be obsolete data, the median congressional district had a partisan voting index of R+2—meaning it leans slightly Republican, but not by the 7-point margin of the incumbent Republican representative. Much of the difference can be attributed to the advantages of incumbency, which can make it much harder for Democrats to win a slightly Republican district. Gerrymandering makes it tougher for Democrats, especially in states like Michigan and Pennsylvania. But it’s not clear that Republicans would have lost control of the chamber with a fair districting process, making it difficult to argue that gerrymandering is the core problem.
Of course, that doesn’t mean the map isn’t tilted toward Republicans. But Democrats could probably win the House with a modest victory in the House popular vote, perhaps something like the 4 or 5-point margin suggested by a FiveThirtyEight model in 2011. Conversely, it’s not at all clear that Democrats would have taken the House with a fair districting process, even though they won the House popular vote.