The Corner

The Tightening — Polling, RIP

So three major polls released this weekend show Republicans cutting the Democratic advantage on the so-called “generic” question to single digits — with Dems leading by 4, 6 or 7 points rather than the double digits they’ve been running consistently in recent months. I think what this tells you is almost nothing. Here’s why.

If the Democrats achieve a blowout on Tuesday night, then a momentary spasm of life in a GOP corpse is meaningless — in November 1992, there was a poll showing Bush the Elder within two points of Clinton, which caused a flurry of hope in GOP breasts until around noontime on Election Day when word quietly went forth from the RNC that all hope was lost. 

If Republicans hang on to the House and Senate, it won’t mean that these new polls measured any real change. It will mean, instead, that all the polling in this midterm election was garbage — that, indeed, polling is in a crisis because it can no longer measure anything specific owing to the increasing sophistication and annoyance of Americans dealing with telephone solicitors.

What’s more, pollsters know most of their polls this year were garbage, as did those who sponsored them, and yet they went ahead and did them and reported on them anyway. They were garbage because pollsters are getting response rates on their calls maybe two out of ten tries — which means that simply for them to get enough respondents to poll, they have had to dig for every last voice they could.

In addition, the vast majority of polls this year were of “adult Americans,” a category that is absolutely useless and meaningless in political polling. Adult American polls are the kinds that end up with responses like 40 percent of people think people have been abducted by UFOs, or that Elvis is alive, that kind of thing. Not to mention the fact that voters form a relatively small subset of “adult Americans” — in a year like this, maybe 35 percent in all.

In October, most news organizations and pollsters started trying to screen for voters and likely voters. But how could they possibly be trusted to achieve that aim when they can’t even get most of their phone calls answered? Do you really think a news organization is going to be willing to pay for everything that goes into making 10,000 phone calls — the phone banks, the trained personnel, the analysts, the data processors — in order to find 1,000 genuine “likely voters” who can be appropriately placed into demographic and ideological categories reflecting the nation as a whole?

If Republicans hang on and don’t really lose significantly, it will be because things were never that bad for them in the first place. And if Democrats win big, it will be because things have been bad for Republicans all year and nothing could have changed it. The only way, in fact, that the polling this weekend will be worth citing is if Republicans do vastly better than anybody believes they’re going to – thus indicating a significant turn in their fortunes upward that has been measured by the poll shift in the last week.


The Latest