Morning Jay: How To Read The Polls, the "Northeast Firewall," And Quantifying The Dems' Midwestern Malaise!

1. How To Read The Polls, Part 2 of a Series. Here’s another tip for consuming the polls: From this point forward, likely voter polls are really a requirement. 

In 2004, 111.8 million people voted in the nationwide House elections; in 2006, just 80.1 million people voted in nationwide House elections.  That’s a nearly 30 percent drop-off.  Generally, the most active and engaged people turn out to vote in midterm elections.  The best polling is going to filter out the non-voters to capture the actual electorate via likely voter models.  Registered voter models won't do that.

A great example of the problems of registered voting polling can actually be seen in Gallup’s most recent write-up of its registered voter generic ballot, published yesterday evening. Gallup finds an even split in the generic ballot, but it doesn't seem to have great confidence that its numbers give an accurate sense of where the voting public is:

Gallup's generic ballot for Congress for the week of Sept. 20-26 shows the race tied among all registered voters. However, Republicans' continuing higher enthusiasm coupled with the usual GOP turnout advantage suggest a significant Republican edge in the nationwide vote for the U.S. House, and, in turn, significant Republican House seat gains…

This is the biggest reason why registered voter polls are so problematic this cycle: Republicans are much more enthusiastic to vote than Democrats, and in relatively non-stimulating midterm elections such as this, enthusiasm counts for a lot.  More Gallup:

Given the usual Democratic advantages in party identification among the general public, it is rare for Republicans to lead on the generic ballot among registered voters. This was the case even when Republicans were the majority congressional party from the mid-1990s to the mid-2000s. Turnout is crucial in midterm elections. With at least 80% of Americans registered to vote but only about half that number likely to vote in the midterm elections, registered voter and actual voter preferences can differ significantly.

All things considered, I’m not entirely sure why Gallup would even put this number of registered voters out there.  Clearly, it thinks the GOP has a lead in voter preferences for the House.  Why publish a poll that shows a dead heat?  On top of that, in the previous six cycles (1998, 2000, 2002, 2004, 2006, and 2008), Gallup had already released its likely voter generic ballot by this point in the cycle.  Its first likely voter model of this cycle is not due out until early October.  That’s pretty peculiar.

I do not know what explains this delay, but regardless: be wary of registered voter polls from this point forward.

2. A Northeastern Firewall for Democrats?  That’s E.J. Dionne’s argument in a widely-read column from yesterday.  He writes:

In 2010, Republicans run away in horror at the prospect of being called moderate, let alone progressive, and that is an obstacle in the GOP's path to a congressional majority. It will be very hard for Republicans to take the House if they don't break the Democrats' power in the Northeast -- and they still have to prove they can do that.

His main source for the long-term problem Republicans will have if they cannot win in the Northeast is … Northeastern House Democrat Dan Maffei, who I am sure offers a totally objective and fair analysis of the Republican party!

This is a very difficult argument to make, especially in light of the fact that Charlie Cook finds eighteen current Democratic seats in the Northeast as Lean Democratic or better for the Republicans.  That means that the GOP stands a good chance of returning to the same position in the Northeast as it was in 2004.