Adventures in confounding variables

by: desmoinesdem

Tue Jul 08, 2008 at 13:22:15 PM CDT

Click the link to read about an AP-Yahoo! poll that showed pet owners prefer McCain, 42 percent to 37 percent, while people who don't have a pet prefer Obama 48 percent to 34 percent.

Associated Press writer Randolph E. Schmid asserts that the "pet-owning public seems to have noticed the difference" between McCain, who has many pets, and Obama, who has none. There are some silly quotes from pet owners about the fine characteristics of people who have pets at home.

I'm not a pollster or a statistician, but without even trying hard I can think of five confounding variables that may have more to do with the results than pet owners identifying with McCain because he also has animals at home.

1. Are wealthier people more likely to have a pet? Because that group would skew more Republican than the population at large.

2. The same goes for people who own their own homes, who are probably more likely to own pets than people who are homeless or live in apartments. Remember, many landlords don't allow pets in apartments.

3. Are married people more likely to keep a pet than single people? Republicans tend to do better among married voters than single voters.

4. The AP piece mentions that dog owners are particularly slanted toward McCain. Well, most hunters own at least one dog, and people who keep a gun at home are more likely to vote Republican than people who live in a home with no guns.

5. Are men more likely to be dog owners than cat owners? The gender gap in voting behavior has been documented for decades.

The AP article doesn't bring up any most of these potential confounding variables. [CORRECTION: Buried down at the bottom of the piece, the AP article does mention that a higher proportion of dog owners are married, compared to the population at large, and that white people are more likely than black people to own dogs.]

Despite being a dog-lover myself, I didn't even know that Obama had no pets, and I'm well-informed politically. I doubt that even 1 percent of Americans will make up their minds based on whether a presidential candidate has a pet.

UPDATE: The "mystery pollster" Mark Blumenthal posted his entertaining and informative take on this story. He also links to this Google search showing how many mainstream news outlets ran with the AP's misleading but "irresistibly cute lead."