Beware the politicized polls

By Tom Quiner

I’m in the marketing business. My company has done some marketing research studies over the years.

One of the biggest challenges to garnering accurate feedback for clients is to sample the right audience. It’s easy to skew the results if the sample isn’t representative of the marketplace.

The same principle holds true for polling the electorate in this election cycle. I will be the first to acknowledge the difficult task facing Gallup, Pew, Rassmussen, and the rest of their polling brethren.

For example, do you poll registered voters or likely voters? Likely voters give a more accurate response than registered voters, but it costs polling companies more money to find likely voters, because they have to make more calls.

One of the most challenging dilemmas: how many Democrats vs. Republicans vs. Independents do you poll? Should it be based on the same percentage as 2008? That probably doesn’t make sense, because Dems had a record turnout four years ago and can’t count on it again this time. In light of President Obama’s job performance, Democratic turnout is a real guessing game.

There is one key standard, though, to political polling: consistency. Whatever your sampling formula, keep it the same throughout the election cycle. Don’t change the rules midstream, or else you no longer have credible benchmarks with which to compare data over time.

That is just what the venerable Gallup Poll has done, they’ve changed the rules.

Gallup has been hammered by Democrats this cycle. Dems have blasted them for consistently producing data more unfavorable to Democrats than other polls. For the record, there is no shortage of Republican complaining about polls. It’s the nature of the beast.

Mr. Obama’s job approval numbers have been typically lower in Gallup polling than other polls. The rap against Gallup is that they over-sample whites. Whites tend to be less approving of Mr. Obama’s policies than blacks and Hispanics.

You may have noticed a dramatic bump in Barack Obama’s job approval ratings with the Gallup Poll since the Democratic National Convention. Mr. Obama’s approval rating stood at 47% before the convention. After the convention, it rose to 50%. It continues to rise and has reached 53%.

This is good news for the Obama Campaign. What voters need to realize, though, is Gallup changed their sampling methodology beginning October first. They increased their non-white sample size by 20% in an apparent cave to the pressure they have been receiving from Democrats.

You can look at the data two ways now. Either data before October 1st was skewed, or the data after October 1st is skewed. Frankly, who knows? But that’s not the point. The point is that the perception that Barack Obama’s job approval rating is on the rise has been manipulated. Gallup shouldn’t have changed their methodology this late in the game.

Now all their polling before and after October 1st is suspect.