The Corner

The one and only.

Why the Generic Ballot May Overestimate Democrats


Nate Silver, founder of the election-prognostication website and now a New York Times blogger, posted this analysis today, in which he argues that the GOP’s advantage in the generic polls might overstate the number seats they will gain in the House. Many district polls show Democratic incumbents running ahead of the generic margin, and Nate hypothesizes that this may show that the Democrats’ strategy of trying to localize races in each House district might be working, and hence the GOP could win fewer seats than might otherwise be expected.

Nate knows more about regression analysis than I do, but it seems that he has overlooked the most obvious explanation for this discrepancy: candidate name ID. Early in a campaign, voters often have heard of the incumbent but not the challenger, and hence are likelier to say they support the incumbent simply because that is the only name with which they are familiar. As Eddie Murphy showed in his wonderful comedy The Distinguished Gentleman, sometimes voters just support “the name you know.”

As proof of his thesis, Nate provided a list of 31 House races polled in August by the American Action Forum — Republicans led the generic ballot by an average of 6.3 percent, but only led in ballots with actual names by 2 percent. But he did not look to see if the Democratic incumbent was better known than his GOP challenger.

I did, and here is what I found. Only 5.7 percent of the voters in the average CD polled had not heard of the Democratic incumbent, but 36.9 percent had never heard of the Republican challenger. As anyone with campaign experience knows, if a challenger can successfully close the name-ID gap with his incumbent rival, the challenger will see his percentage of the vote rise.

Indeed, that is what the AAF data show. In four seats (MI-7, OH-1, PA-11, and PA-12), the Republican candidate is as well known as the Democratic incumbent. In each case, the GOP nominee did as well or better in the named ballot as in the generic. In contrast, in four seats (CT-4, FL-24, MO-3, and NY-20), the ID gap between the Democrat and the Republican exceeded 50 percent. In three of the four, the Democrat did between 10 and 13 percent better on the named ballot than on the generic, significantly better than the four-point average for all districts.

There’s plenty of time for the candidates to campaign, and each race will be determined in part on personal appeals. In every wave election, some endangered incumbents pull through because of their strengths or their opponents’ flaws. But, as one noted prognosticator observed earlier this year, “On average the generic ballot has overestimated the Democrats’ performance in the popular vote by 3.4 points since 1992.” When this is applied to the AAF data Nate cites, it appears that the Democratic problem is not better than it appears; it’s worse.

Henry Olsen is director of the National Research Initiative at the American Enterprise Institute.