November landslide?

Lately something curious has been happening with polling for the Presidential race.  Gallup and Rasmussen, which release daily tracking polls, continue to show a tight race -- with sometimes Obama ahead and sometimes Romney, but rarely by more than a point or two.  Other pollsters have begun showing a large Obama lead.  Reuters-Ipsos has him ahead 49%-to-42%, CNN/ORC 52%-to-45%, Pew 51%-to-41%, NBC/WSJ 49%-to-43%, Democracy Corps 50%-to-46%. Even Fox News, not noted for a liberal bias, has Obama leading 49%-to-40%!

Why the discrepancy?  Some Republicans think they've spotted the explanation.  Those pollsters showing a big Obama lead also show a suspiciously-high level of Democratic party ID in their samples.  That Pew poll, for example, shows its sample 38% Democratic-leaning, 25% Republican-leaning.  Even Fox's sample shows 44% Democratic, 35% Republican.  Not all pollsters release the party ID breakdown of their samples, but those that do are mostly showing similarly Democrat-heavy leans.  Surely, Republicans say, that's why these polls are showing such big Obama leads -- they're just polling samples which include too many Democrats relative to the general population.

It's odd, though, that this should be the case.  Most pollsters don't weight their samples to match an assumed party ID breakdown.  They weight by other factors such as age, race, gender, etc., but the party ID breakdown is just another number that emerges from the results, if the pollster even bothers to ask about it.  So if skewed party ID explains the big Obama leads, then an awful lot of pollsters are using samples that just happen to turn out to have too many Democratic-leaning respondents in them.

The same is true of the polls in the various swing states, many of which have shown large Obama leads while averages of national polls, such as the RCP average, continue to show a close race.  Those swing state polls also show oddly-high levels of Democratic party ID among their samples, and again, I've seen Republicans dismiss them on that basis -- the polls are wrong because the samples are unrealistically Democratic-leaning.

But what if something else is going on?  What if, instead of poll after poll (both nationally and in the swing states) using samples that just happen to have too many Democrats and too few Republicans, the samples are accurate and there's actually a huge shift in party ID under way?

It makes intuitive sense.  We tend to analyze upcoming elections on the assumption that current trends, including party ID, will resemble those which determined past elections, but lately the Republicans have been trashing their own brand in an unprecedented fashion.  Forced vaginal ultrasounds, "personhood" laws, nutjob teabagger House and Senate candidates, the Ryan budget, vote-suppression laws, debt-ceiling brinkmanship, government-shutdown threats, doubling down on global-warming denialism even as record heat scorches the country, the spectacle of freaks like Santorum and Perry being taken seriously as Presidential candidates -- this headlong descent into madness is unprecedented, even during the Bush years.  We're constantly baffled that the great sane pragmatic middle isn't deserting the Republican party in droves over this stuff.  But what if they are?

But if that's happening, why aren't Gallup and Rasmussen showing it?  Well, I'm not sure about Gallup, but Rasmussen is one of the few pollsters that does weight its samples by party ID -- that is, by what they expect the party ID breakdown to be.  And Rasmussen is noted for being somewhat Republican-leaning, as pollsters go.  If there's a massive shift in party ID going on, but Rasmussen doesn't recognize that it's happening, then their system for weighting their samples could be filtering the very real effects of that shift out of their results.

This would also explain why national averages such as the RCP average still show a close race.  Those averages skew strongly toward Rasmussen and Gallup since they release a new poll every day, while other pollsters are less frequent.

If I'm right about this, then what's happening is straightforward:  Democrats will win in a landslide this November because the public rightly sees that the Republicans have gone bananas.

I'm hesitant to even suggest this because we can't afford complacency.  We need to fight like hell because we need a landslide.  We need Obama back in office with such a margin that his mandate is beyond challenge.  We need to take back the House and expand the Senate majority.  And the other side's LIV base will turn out.  The Obama's-a-Marxist, Earth-is-6,000-years-old, Obamacare-means-death-panels crowd will vote.  The last thing we want is for our side to think we have this thing in the bag and we don't need to bother.  I myself am in a position where it may take some inconvenience and expense to make sure I'm correctly registered to vote at the date of the election, but I'm still fixedly determined to make sure it happens.  Never in my lifetime has there been an election with stakes this high and a contrast this stark.

But as the sun rises over our country on November 7th, we may find that we have more to celebrate than we knew.