Saturday, April 22, 2006

Pollsters Day

I don’t exactly know what possessed me to dive into the polling blogs. They’re filled with dense, complex number crunching, analysis & very little humor. Dry is an understatement. That said, polling data drives the political and visa versa, yeah, a few smarties will tell us they don’t use ‘em, but they’re lying.

Political parties and their candidates need a certain level of reliable and valid data coming into the decision making process to help shape agendas, strategies and message. There’s no doubt that the primary guys over in Iowa’s first district GOP primary are aware of the generic polling numbers on Republicans’ attitudes towards immigration. They all know it moves primary voters, particularly the less educated Republican primary voter.

The polling blogs cover a wide range of issues from poll construction, to statistical methodology (snore), to baseline analysis on the tabs; these blogs take a comprehensive approach to political poll dissection. My favorite polling topic of the moment is the emerging dialogue on changes in party identification.

The Mystery Pollster & Political Arithmetik have recent posts up on the shift in voter identification. Mystery Pollster Mark Blumenthal takes the Gallup longitudinal numbers on party id and makes a case that Republicans are in a real slide. Sure, some folks have suggested that the Gallup organization didn’t cull enough of the less likely voters out of their samples for their tastes. But I think the magnitude of the change, almost four percentage points in less than two years, suggests there’s probably some truth in the Gallup numbers. In addition, the Gallup numbers are not unique. Professor Franklin's Political Arithmetik finds a similar result weighting and aggregating questions on party id from multiple polls and a variety of pollsters.

Between January 1, 2005 and March 12, 2006, Republican partisan identification declined by an estimated 3.6%. The percentage of the adult population calling themselves Independent rose by 4.6%, and the percentage of Democrats declined by a statistically insignificant 0.4%. These changes are important for polling methodology and also present a politically important shift in the partisan balance.

I’m thinking that the statistical evidence of voters moving away from the Republican Party and into no party land is not a fluke, but indicates a problematic trend for the Republicans, particularly since the Democrats have not realized the same level of party identification decline.

Anytime you see a statistically significant change in a set of variables, you should think real change. This is not some sort of quirk in the numbers that can be explained away by attacking the methodology, although political spinners might find it useful to just chew up the methods and hope the GOP can turn this party id problem around…quickly. The real question is why a shift out of the GOP and not the Democrats and are these potential voters going to come back to the party at some time down the road, preferably in the next six months.

Comments: Post a Comment

Links to this post:

Create a Link

<< Home

This page is powered by Blogger. Isn't yours?

Who Links Here