Don’t let bad polls rain on your parade

Members of Congress and other elected incumbents everywhere are being fooled by a phenomenon known to statisticians as ecological fallacy. After looking at enough wrong-track polls, often conducted nationally, and reading results of generic ballot and “let’s elect someone new” surveys, most incumbents falsely conclude that they themselves are vulnerable in their own local districts when, in reality, they aren’t.

The ecological fallacy occurs whenever one tries to infer from averages to individuals. Asians score highly on the math portion of the SAT exam, so maybe that Chinese kid next door can help your low-performing kid do his algebra homework. Maybe so, or maybe not. Or you might assume that someone tall and black will, on average, be the better basketball player. And you might observe that, on average, white men can’t jump. But then some pasty white guy like Pau Gasol of the Lakers or Steve Nash of the Suns comes along and challenges the stereotypes.


But stereotypes like that do have enough of a basis in reality to justify some solid conclusions (and paranoia that they are too often true). Perhaps a more appropriate way to debunk the wrong-track and generic-ballot harbingers of defeat is to consider weather predictions. Whenever your friendly weatherman says that the chance of rain is 30 percent, there is a good chance that he merely means that there is only a 30 percent probability that it might rain in any one place in the whole viewing area during the time specified. In the real world, you won’t be everywhere in the viewing area all day. But the paranoid style of many would hear that there is a 30 percent chance it will rain on you personally, wherever you happen to go, at all times. And that’s just not true. Similarly, when incumbents hear that voters don’t want to reelect incumbents or are voting generically for the other party, they think it will rain everywhere all the time. They are wrong.

I am always taken with analogies of 2010 to 1994. The bad weathermen like to highlight the 54-vote partisan swing. It makes everyone carry an umbrella. Almost no one ever highlights that only 34 incumbents were actually defeated. My point is not that 1994 didn’t reflect an important political shift, but rather that the shift was not as cataclysmic for individual Democrats as many want to remember. It was bad, on average, but not career-ending for a lot of individual Democrats. More than 200 won seats in Congress. Similarly, some incumbents, maybe as many as 34, will lose their seats in this bipartisan “throw the bums out” era of the Tea Party, but the scope and scale of the damage won’t be what many imagine. Between gerrymandered districts and genuinely good performance in office, a lot of incumbents will deserve reelection, and they will earn it.

This argument needs to be made because too many incumbents are acting like they expect it to rain on them, and it’s causing an overreaction. Instead of packing an umbrella in their briefcases, some are overdressing in rain suits while a few really paranoid types have embarrassingly donned wetsuits with full scuba gear, just in case Noah-class rains start to fall. These sadly obsessed incumbents are trying so hard to stop the rain by spouting pro-change messaging that they lose all authenticity and credibility. Rather than run on their records of accomplishment, many are goaded into trying to out-party the Tea Party. It’s like a middle-age man on the dance floor popping to some beats: an embarrassment.

Incumbents need to ignore polls and generic surveys conducted outside their own districts. The only polls that count should be the ones that have their own names on the ballot.

David Hill has been a Republican pollster since 1984. This cycle he is polling for gubernatorial campaigns in four states.