Sunday, February 17, 2013

Polling: Name This Fallacy

Sorry, I don't know whether this fallacy has a name (it's not the ecological fallacy, by the way) but it seems to be quite common among political pundits. I'd call it the fallacy of homogeneity if that weren't such a snooze.

The mistake is in thinking that all members of a sample are alike. For example, exit polls after the 2004 election revealed that "moral values" was one of the most important issues to voters, and 79 percent of those voters voted for Bush over Kerry. "Oh, no!" said many Democratic talking heads. "Better not talk about gay rights, or abortion, or drug legalization, or..."

The problem is that all voters are not alike. So in fact, we don't know if a single vote was changed because of Bush's and Kerry's positions on moral values. In particular, evangelical Christians are (I imagine) inclined to say that moral values are important to them. But they are not swing voters; they were going to vote overwhelmingly for Bush anyway. Were there any undecided voters who were swayed by a candidate's position on moral values? We don't know.

More recently, I heard Melissa Harris-Perry on MSNBC announce that the myth of NRA influence was dead, because a new poll showed that while 26 percent of respondents said they would be more likely to vote for a candidate who had an NRA endorsement, 39 percent say they would be less likely. That's a striking result, but hang on a minute. If you were Heidi Heitkamp, the newly elected senator from North Dakota (Democrat, pro-gun) would this make you change your position?

Probably not by much. Most likely, most of that anti-NRA 39 percent is people who were pro-gun-control to begin with. The question that interests Senator Heitkamp is, what do the percentages look like in North Dakota? And we can't tell.

No comments:

Post a Comment