Assessing the Generic Ballot, Part II

In yesterday's column, I assessed the generic ballot - and concluded that we should avoid it. The reason for this assessment is that the statistical process we use to make a prediction based upon the generic ballot requires assumptions that it does not meet. Therefore, it is an invalid indicator of electoral results.

This was my argument yesterday. On the way to making these points, I noted that - correcting for the historic pro-Democratic skew in the generic ballot - indicates that the generic ballot is today predicting a final result of 51.75% D to 48.25% R. Later on, of course, I argued that this prediction is not valid. But this figure does offer a nice opportunity for a thought experiment: is a Democratic majority of this size in the two-party vote enough to yield a Democratic majority of House seats?

Important to note, right off the bat, is that ours is not a parliamentary system - our House is composed of plurality-elected representatives from districts that are geographically distinct. The national vote, in itself, does not matter; what matters is its distribution. So, the answer might be no. There are reasons to believe that it will be. The biggest is that the electoral alignment that was slowly developing since the 1970s, but that dramatically altered our political configuration in 1994, has ossified the House of Representatives. Republican voters are in what are today Republican-held districts. Democratic voters are in what are today Democrat-held districts. This has meant that the number of districts decided by a small margin has decreased, and decreased quite dramatically.

On top of this, most incumbent members of Congress know their districts very well - and know how to get to half-plus one. The incumbent reelection rate has increased in the last fifty years to an astonishingly high 99%. Simply stated, incumbents have become very good at staying in office. They have developed incredible informational and campaign capacities: they know the ins-and-outs of their districts, and so are able to reflect their constituents' interests in Congress on salient issues; come campaign season, they are able to advertise this reflection very efficiently. Think of it as an evolutionary biology model - as we move from a citizen class of legislators who intend to stay in Congress for a limited time to a professional class of legislators who intend to make a career out of Congress, and as we give these legislators better and better tools for studying and communicating with the voter, we expect the legislators who are the most skilled to survive election after election. We expect the less skilled to lose. This implies that, over time, as long as all else remains equal, legislators will get tougher and tougher to beat. And so they have.

So, we have realignment and we have a very skilled set of professional legislators. What has this meant? It has meant ossification - the process has been very dramatic in the last 12 years. From 1950 to 1994, every 1% increase in the Democrats' share of the two-party vote meant an 8-member increase in their House delegation. Since 1994, however, a 1% increase in the Democrats' share of the two-party vote has barely yielded a 2-member increase in their House delegation. From 1952 to 1992, the Democratic caucus tended to vary from its mean by an average of 21 members. Since 1994, the variation from the mean has been a scant 4 members. In other words, the House has become less responsive to changes in voter mood, and the partisan caucuses have become more stable in their composition. This is why, in 1952, the Democrats increased their share of the 2-party vote by 2.6% and netted 19 seats - but why, in 1996, they increased their share of the 2-party vote by 3% and only netted 2 seats.

The conventional wisdom on this election is that the "national conditions" are so anti-Republican that they will upend this alignment and endanger the strongest of Republican incumbents. This alignment is not impregnable. No alignment ever is. Strong incumbents are not invincible. A Democratic victory in the two party vote akin to their victory in 1982 - 56% to 44% - would terminate the GOP's majority and disrupt the current alignment. But, the generic is actually calling for a Democratic edge that is about a quarter of this size. What will this mean?

Here is how 51.75D - 48.75R might break down in 2006. In 2004, the Democrats won 48.6% of the two-party House vote. Thus, we are talking about an increase of 3.15% in the Democrats' share of the two-party vote. If that share were distributed evenly across all House districts - such that all Democratic candidates increase their share of the two-party vote by 3.15% over the last general election - the Democrats would net 5 seats: IN 09, PA 06, CT 04, WA 08, CO 04. The reason for this is that 227 House seats are held by Republicans who won the two-party vote by more than 3.15%. If you take this amount from every House Republican and Republican challenger, and add the same amount to every House Democrat and Democratic challenger - you would see a 110th Congress with a 25-seat Republican majority, 227 to 208. Great news for Baron Hill, Lois Murphy, Diane Farrell, Darcy Burner, and Angie Paccione; but otherwise a ho-hum affair.

The task for Democrats, then, will be to distribute this 3.15% in such a way that enough Republican incumbents lose sufficiently more than this amount. This is where the ossification of the House makes this task difficult. Most Republican seats are from districts that feature incumbents running for reelection who (a) consistently get at least 60% of the two-party vote, and/or (b) represent districts that are conservative.

This makes an important difference. We expect strong incumbents to beat their party's national performance with relative ease. That is why we call them "strong." As for weak incumbents in conservative districts - we expect the anti-Republican mood to be less strong in those places, and therefore the danger to them that much less. The sorts of places where we expect Republicans to do worse, sufficiently worse to bring in a Democratic replacement, would be in Republican-held open swing districts, or Democratic-leaning districts with weak Republican incumbents. The problem for Democrats, in a nutshell, is that there are just not enough of those types of districts. By my count, there are only about 13 of these districts in all, and the Democrats should not expect to win all of them.

Most in the press examine the quantity of seats on the Democrats' list - and are impressed. My interest is more in the quality of seats - and I remain unimpressed. There are many seats on it that, to me, just do not look as promising for the Democrats as pundits believe. At the top of the Democratic list are a number of incumbents who roughly match the share of the two-party vote the average House Republican enjoys: Steve Chabot, Nancy Johnson, Deborah Pryce, Curt Weldon. We generally expect these strong incumbents to beat the party's national trend - and therefore to survive a 3.15% shift of the national two-party vote. Also at the top are a number of weaker incumbents in conservative districts: Chris Chocola, Geoff Davis, Thelma Drake, John Hostettler, Randy Kuhl, Mike Sodrel, Charles Taylor. We should generally expect an anti-Republican mood of 3.15% to be relatively muted in these districts, since the average voter in that district, even if he is still feeling a little more anti-Republican than normal, is still conservative. If the anti-GOP mood is muted, incumbents like these - even though they are relatively weak - will probably survive.

I would be more bullish about Democratic prospects for taking the House if the list did not have so many seats like these - seats that feature Republican incumbents who are either strong or in conservative districts. If there were more open seats on the Democrats' list, and/or more incumbents like Jim Gerlach, Chris Shays, and Heather Wilson - I would be right there with all the pundits who are talking up a Democratic capture of the House. Thirty years ago, there would be enough districts like these - enough Republican open seats and Republican incumbents in Democratic-leaning districts - to induce a 15-seat swing in the House. Today, though, these sorts of seats are simply not there. In 2006, netting 15 seats from 3.15% is like squeezing blood from a stone.