Search This Blog

Wednesday, October 15, 2008

How Rush May Be Responsible for Obama's Lead in the Polls

A famous quote attributed variously to a distinguished assortment of politicians and humorists goes like this, "There are lies, damned lies and statistics!" We're all a little suspicious of statistics, but most of us believe in them a little more than we'd like to admit.

I learned a little something about statistics during my brief tenure in two graduate schools and the big thing was that for a study to be valid you have to have an accurate sample. The closer you get to creating a sample of survey respondents that accurately reflects the same makeup as whatever group of folks you want to discover something about, the better your results.

One way to do that is to use another set of statistics to guide you in choosing your sample. Polls of registered voters, for instance, select participants that reflect the proportions of Democrats, Republicans and Independents who actually registered to vote. In other words if 49% of registered voters in South Carolina are Democrats, 38% are Republicans and 13% unaffiliated, then the pollster doing a survey of registered voters would pick a thousand or so voters and make sure he had 49% Dems, 38% Republicans and 13% Independents if he wanted his study to be accurate.

This year, that's a problem because a lot of Republicans registered as Democrats! If you go by voter registration, you're not going to get a true picture of the registered voters because a bunch of "Democrats" aren't really Democrats and have no intention of voting for Barry.

During the primaries, the infamous "Operation Chaos" skewed the registrations significantly in a lot of so-called battleground states. Rush Limbaugh's campaign to keep Hillary Clinton in the race by having Republicans cross over and vote in the Democratic primary, resulted in hundreds of thousands of Republicans registering to vote as Democrats. Since the primaries are over, most have never bothered to change their registration back. They don't need to to vote for McCain.

The only problem is that voter registration now shows an inordinately high number of registered Democrats on the books and pollsters use those numbers to determine how they sample their pre-election polls.

So, when the pollsters choose a sampling of registered voters, they select for an artificially large number of Democrats based on how voters registered for the primaries in areas where Operation Chaos took place. This could account for Obama's having such a tenacious lead. I suspect a blind sample of voters might show significantly different results.

At any rate, polls based on likely or registered voters should be suspect. They very likely have a sampling that has too many Democrats in it. The poll results are therefore going to inevitably lean toward Obama. This could mean a nasty surprise for pollsters in November, but only if the polls themselves don't discourage conservatives from voting because it looks hopeless for McCain and Palin.

This year of all years, DON'T BELIEVE THE POLLS. I'm betting they're skewed because Operation Chaos gave us a false distribution of Democrats and Republicans among registered voters.

Just one man's opinion.

Tom King

No comments: