[9/4/10] A Nightmare on Capitol Street

What Freddy Krueger stalks Democrats' dreams this year?  It seems likely that their nightmare is a Republican wave election that will make their 2008 tsunami look like a ripple in a goldfish bowl.

Sean Trende has recently written that Dem prospects for the House are bad and getting worse:

Right now, the idea of gains in excess of 60 seats for the GOP is unthinkable to many. Gains of that magnitude haven't happened in over 80 years. But unthinkability is not evidence. What actual evidence we have reminds us that no political party has hit the trifecta of a lousy economy, an opposition at its nadir (in terms of [previous] seat loss), and an overly ambitious Presidential agenda in over 80 years.

Eighty years ago, the Republicans under Herbert Hoover were dealt a crushing defeat, when they went from a slim 218 seat majority -- though Democrats controlled the House owing to Republican deaths -- to a tiny minority of 117 seats, losing 101 seats.  Sherry points out to me that this was followed by five consecutive elections of Democrat presidents.   Furthermore, only go-along-to-get-along almost-Democrat national-defense Republicans -- Eisenhower and Nixon -- were elected until Reagan 50 years later.

But what about this year and what about the Senate?  For our answer, we have that very insightful analysis by Trende.  He begins by comparing exit polls from the 2009 Virginia and New Jersey gubernatorial races to those from the 2008 Presidential election, and concludes this:

In Virginia, the Republicans' share of the electorate increased by 12% from 2008 to 2009; in New Jersey it was 10%. In Virginia, Democrats were at about 84% of their 2009 level; in New Jersey it was 93%.

In both states, the Democrats' share of the Republican vote dropped by about 50% (50% in Virginia, 56% in New Jersey), and their share of the Independent vote dropped about 66%. The Democrats' share of the Democratic vote was pretty stable; up 1% in Virginia and up 2% in New Jersey.

How valid is an analysis based on such turnout shifts?  When he applied these shifts to last winter's Massachusetts senatorial contest, he concluded that Scott Brown had a real shot.  You may recall that after that incredible result, on January 18, in a piece titled "Yes, Democats, the Sky IS Indeed Falling," I wrote this:

It's truly amazing that, in just one year, the Democrats have so angered the electorate that not even Ted Kennedy's seat is safe for them.  It's hard to imagine what could prevent November's elections from being a total debacle for their party.

The critical shifts are the turnout changes -- up for Reps and down for Dems -- and the shift in Independents from Dem to Rep.

When analyzing this election by examining polls, it is critical to remember that pollsters don't just sample and report.  They tinker with the data by weighting the responses of the Reps and Dems differently, based on some assumption about the composition of the electorate in the actual election. 

Effectively, they sample Democrats and Democrat-leaning independents in one poll, and then sample Republicans and Republican-leaning independents in another, and then reassemble their two polls into one weighted to give them the electorate they expect.  One effect of this is to raise their sampling error by about 40% ( actually sqrt(~2) - 1, for those who care).  But the most significant effect is to make their result critically dependent on their assumption of the composition of the electorate in the actual election.  In other words, the turnout question is key.  That's what makes Trende's trends so critical.

Now his shifts illuminate Dem prospects in the Senate.

...the exact same methodology used to forecast Massachusetts in late December was applied to the various competitive Senate races. The composition of the electorate was altered in each state so that the electorates would be 11 percent more Republican, 5 percent more Independent and 11 percent less Democratic than they were in the 2008 presidential race. Additionally, the Democratic candidates' share of the Republican vote was decreased by 47 percent (which usually worked out to only a couple of points), and by a third among Independents. The share of the Democratic vote was increased by one and a half percent.

The results suggest that Republicans would pick up 12 Senate seats, before accounting for any candidate effects.

Trende lists the states with competitive races grouped by incumbency: Dem, Rep, or Open.  His results ordered by predicted Democrat vote are here:

Stt     RepPred DemPred DemRCP  Incumb
AR      67.2%   32.8%   30.8%   Dem
LA      66.2%   33.8%   36.0%   Rep
KY      64.5%   35.5%   42.0%   Rep
WV      63.5%   36.5%   51.0%   Open
IN      59.3%   40.7%   50.0%   Open
MO      58.5%   41.5%   40.0%   Rep
CO      58.1%   41.9%   43.3%   Dem
FL      58.1%   41.9%           Rep
NC      57.8%   42.2%   38.0%   Rep
IA      57.7%   42.3%   36.0%   Rep
ND      57.2%   42.8%   25.0%   Open
OH      56.4%   43.6%   39.0%   Rep
NH      55.8%   44.2%   39.0%   Rep
NV      54.2%   45.8%   47.5%   Dem
PA      54.2%   45.8%   40.0%   Open
OR      53.5%   46.5%   52.0%   Dem
WA      52.9%   47.1%   47.8%   Dem
WI      52.9%   47.1%   45.3%   Dem
CT      50.4%   49.6%   49.0%   Open
CA      50.2%   49.8%   43.8%   Dem
IL      49.0%   51.0%   39.0%   Open
DE      48.4%   51.6%   35.0%   Open
NY      46.9%   53.1%   51.3%   Dem

Three points need to be made here.  First, note that Trende's predictions for the Dem vote are similar to the RCP averages.  The RCP averages aren't corrected to two-party percentages, but the point remains.

Second, if we decide that the battleground states are those where his predicted Rep vote is 45% to 55%, then those states are all Dem or Open seats: NV, PA, OR, WA, WI, CT, CA, IL, DE, and NY, a truly remarkable list.

Third, there are lots of individual candidate effects, on which Charlie Cook, Stuart Rothenberg, and Larry Sabato are famous for focusing.  Trende discusses those at some length.  For example, West Virginia has the fourth highest predicted Republican percentage. What in the world is the commentariate doing saying that Manchin is a shoo-in winner there?  Here's Trende again: "Manchin is a popular, socially conservative Democrat who fits the state well. Still, the West Virginia numbers should make the Manchin campaign nervous, as they are clearly fighting a steep headwind."  Trende wrote and published that before the WV Rasmussen poll showed Manchin up only 6 points on the not quite Previously Unknown Republican* John Raese.  Trende's analysis leads me to believe that Manchin is walking his last mile, but that remains to be seen.

There are certainly individual candidate effects, but this Freddy Krueger nightmare is stalking someone, and it's not the Republicans!