Home / Dave Brockington / Thursday Afternoon Poll Blogging

Thursday Afternoon Poll Blogging

Comments
/
/
/
2369 Views

As Erik points out, Silver has positive news for Obama in Ohio.  Much of this optimism hangs on the house effect of the recent Gravis Marketing poll; measuring and adjusting for house effects can be a bit of an art.  House effects can also vary from election to election, which to my mind makes reliable inferences drawn from them somewhat risky.  House effects perhaps aren’t as unreliable as BABIP or Catcher ERA, but it isn’t impossible that, in this electoral context, Gravis Marketing has a more valid combination of a likely voter model, demographic weighting of the sample, method of measuring leaners, etc., than do the other houses.  Unlikely, I’ll allow, but not beyond possible.

Incidentally, two days ago UK Polling Report put up a thorough post on house effects.  Granted it’s from the British context, but the basic principles involved are context-independent.

One good illustration of how variance in house effects matters is on August 25, Silver has a discussion of a recent CNN poll and the difference between it’s registered voter estimate (Obama +9%) and the estimate after their likely voter filter is applied (Obama +2%).  Another, perhaps more optimistic (from a Democratic POV) illustration is an article by Jonathan Chait discussed over at The Democratic Strategist.  Briefly, most houses are apparently assuming an electorate that is whiter than reality.  The example cited is ABC, which assumes a 78% white electorate.  This would be going against a trend consistent since 1992 when the white share of the electorate fell from 87% (1992) to 74% in 2008.  It was 77% in 2004.  Even if the enthusiasm gap favors Republicans in 2012, reversing this trend ten years seems highly unlikely.

TDS nails it in the end, of course:

None of which changes the priority challenge facing Democrats — to launch the most extensive and intensive GOTV mobilization of the base constituencies in the history of the party.

Unrelated to polling (and I trust that Lemieux might have something to say about this) a district court panel unanimously ruled the Texas voter suppression law unconstitutional, and the three judge panel included a Bush II appointee (along with Clinton and Obama appointees).  Before we get too excited, the chance of me having those drinks in New Orleans as scheduled for right about now is marginally better than Texas flipping blue any time soon, and it has to meet a more stringent test for changes in electoral law as Section 5 of the Voting Rights Act applies to the entire state, burdens that Pennsylvania, Indiana, and Wisconsin can ignore.

Of course, Texas is also challenging the constitutionality of Section 5, a challenge the same panel allowed to proceed (and good luck with that).

  • Facebook
  • Twitter
  • Google+
  • Linkedin
  • Pinterest
It is main inner container footer text