A Skeptic's View of the GOP Wave Theory

As readers of this blog know we have been a bit skeptical of some of the polls showing a 7-17 point GOP likely voter advantage.  12 days out here are some reasons why:

The Shift in the Vote Is Just Too Big - If the final vote ends up being 49/41 for the Rs as the average of recent Congressional Generics have it today, it would represent a massive 15 point shift from the 53D/46R result in 2008, and a 14 point shift from the 52D/46R result in 2006.  By comparison, the shift from Bush in 2004 at 51/48 to Obama 53/46 in 2008 was 10 points. Do we really believe that the shift we are seeing today in national politics is 50 percent greater than the shift from Bush to Obama and the Democrats a few years ago?

A high single digit win for the Republicans would represent the best showing for their party since 1994.  I simply have a hard time believing this could take place after the Democrats 52/53 percent performance in 2006/8, the best back to back election performance for the Democrats since the 1930s. 

This is all particularly true when the Republicans continue to have very low, lower than Democratic approval ratings, a not terribly compelling agenda and set of leaders,and the memory of their monumental governing mistakes are still very fresh in voters minds.

Initial Early Vote Totals Aren't Picking Up a Republican Wave - CBS News has a report today showing that in actual voting so far there is no evidence of a GOP advantage.   Given that Republicans traditionally are more adept at using non-election day techniques than Democrats, this finding is a bit suprising to me, and should give every pollster pause that their likely voter models may simply no longer apply to this election as it stands today.

There is No Evidence in State and Local Polling of the Election Breaking Open for the Rs - Any fair assessment of the race by race polls in recent weeks would find net Democratic improvement, and importantly, Republicans are still at very low levels in the biggest states like CA, TX, FL, NY, IL and OH.  It is almost impossible to construct a scenerio where the Republicans could make huge gains without big shifts in the largest states given the weight of their populations in a national poll sample.  And we just aren't seeing it.  There is only one race - the Ohio Senate - where the Republican candidate is over 50 in all those states above.  In almost every other case the Repubican candidate is at 46 or so below.  Remember we are talking about a proposed 15 point shift in the vote - how does that happen with the largest states showing any sign of major GOP gains, and in three states - NY, FL, TX - the Rs are arguably underperforming from their even their recent levels of support. 

The Flaws Exposed This Week in the Gallup Likely Voter Sample Raise Questions About The Assumptions In Other Polls - As I wrote yesterday, the most pro-Republican Poll, and perhaps the most influential, Gallup, has reported a fatal flaw in its likely voter model which has made its results illegitimate.  It is hard to speak to the art that has gone into the assumptions behind the various likely voter models in 2010. But what we do know is that 15-20 percent of the American electorate - Millennials and Hispanics - are not as easy to capture using traditional land line, English dominant polling techniques, and these groups have been voting 2:1 Democratic in recent elections.  A recent Pew study shows that certain more traditional (and less expensive polling techniques) can skew results by up to 5 points for Republicans. 

For more on how the changing demographics of the US is creating a new politics and a new electorate see this excellent essay by Morley Winograd and Mike Hais.

Remember that in the last two weeks we have seen a 22 point spread in the very same Congressional Generic polls clearly depending on how the sample was constructed. This is an extraordinary spread on such a simple question.  Just this week, Gallup had the registered voter number Generic at 48R/43D, WSJ/NBC 46D/44R.  And this wide spread cannot be attributed to sample construct.

Do we really have any idea where this election is today?  

I have to run now, so to conclude, quickly. my own take is that a 3 point plus GOP lead is plausible right now.  Getting to that number still requires as big a shift to the Rs as we saw occur from Bush to Obama.  This would be by any measure a huge shift of the vote.  External measures of right track/wrong track, Obama approval rating, current levels of job creation and the stock market, lack of popularity of Republicans simply do not reinforce a 15 point net swing against the Democrats, a swing 50 percent bigger than occured from Bush to Obama.  Right track/wrong track is better today than it was in 2008, and much better than in 1994 or 2006.  Obama's disapproval rating is about the same as the percentage of the people who voted against him in 2008.  By virtually every measure except unemployment - and this includes job growth - the economy is in significantly better shape than it was two years ago.  Nothing the Democrats have done is wildly unpopular (in the high 50s plus), and the on most measures of the big issues the Democrats still beat the Rs.  The country still blames Bush and the Republicans for causing the current economic mess. 

I concede this will be a Republican year, but I am deeply skeptical of any generic poll showing the Rs up more than 3-4 points at this point.   There is just too much contervailing evidence that a shift of a greater magnitude is not taking place, or that the methods used to come up with such numbers may be indeed flawed (as we have already seen with perhaps the most influential of all these polls).