Today we have some outlier statements made by other political analysts. As rounded up by Andrew Sullivan, we have Nate Cohn at the New Republic quoting Alan Abramowitz:
Democrats need a 13 point Democratic edge on September 1 to win the 17 seats necessary to retake the chamber in November.
For any wave election, that would be a very soundly built levee. However, recent data suggest this statement is rather overstated, for two reasons:
- In the last 6 elections, the drift from previous-year generic-opinion polls to November outcomes is a median of 3 points (4.6 points for midterms only).
- Republican-favoring gerrymandered districts are weaker than they look. Because they seem to have more independents than other Republican districts, the necessary national popular-vote margin for a Democratic takeover is lower than in 2012, and I estimate it at about 4 points.
I have analyzed this all recently. I think the levee could be breached by a 7-point generic polling margin next October. Also, here’s a rather good overview by David Wasserman at the Cook Political Report. Let me review a few points about what appears to be in past and present polls.
To revive an old promise I made last year: if the race moves that far toward Republicans during the month of October 2014, I will eat a bug.
Does the generic opinion predict actual votes? To answer that, we have to quantify Delta, the 12-month difference between opinion and the actual national vote. Using generic Congressional ballot data from RealClearPolitics, I get the following medians:
|Year||1 year before||Outcome||Delta|
|2002||D+0.0%||R+4.6%||Toward R 4.6%|
|2004||D+1.5%||R+2.6%||Toward R 4.1%|
|2006||D+8.0%||D+7.9%||Toward R 0.1%|
|2008||D+12.0%||D+10.7%||Toward R 1.3%|
|2010||D+1.0%||R+9.4%||Toward R 10.4%|
|2012||D+1.0%||D+1.3%||Toward D 0.3%|
|Average||D+3.9%||D+0.5%||Toward R 3.4 +/- 4.0%|
The correlation between polls now and eventual outcome is r=+0.89, which is pretty good! Indeed Democrats are likely to lose ground, but by less than implied by other analysts. The average movement is Delta=3.4% toward Republicans, with a standard deviation of 4.0%. (For total nerds: my discussion of the nice work of Bafumi et al. is here.) I note that the drift over the final two months of the campaign, from September to November, should be smaller.
In regard to point #2, here is a mathematical fact about redistricting: when you pack your opponents tightly into a few districts, the swingable voters are more likely to be in your districts. Call it Gerrymandering Lemma #1.
(An aside: The lemma is even mathematically provable! I estimate that as a general rule, win margins in a gerrymandered district should be discounted by as much as one-third, i.e. a 15-point win is only about as safe as a 10-point win elsewhere. I’ll write that up at some point.)
To summarize: Democrats face an uphill battle for House control, but the levee built by gerrymandering is not so high as you have been led to believe. A popular-vote win of 4-5 points might be enough. In terms of opinion polls, which usually overestimate Democratic turnout, if the generic ballot is at D+7% next October, it might be time for Republicans to start looking for high ground.
Update: Nate Cohn and I find points of agreement. See comments.