The Other McCain

"One should either write ruthlessly what one believes to be the truth, or else shut up." — Arthur Koestler

Random Freaking Adults (Again)

Posted on | March 30, 2011 | 11 Comments

From time to time, I find myself obliged to give this little lecture. I get tired of repeating myself, so listen up:

LEARN TO READ POLL DATA PROPERLY OR
STOP TELLING ME ABOUT POLLS

Every two years from 1998 through 2006 (five consecutive elections), my job at The Washington Times required me to compile a weekly graphic of key election polls, in addition to whatever other campaign coverage I was writing or editing. Then, when I cut loose and went freelance in 2008, I spent the next two-and-a-half years up to my eyeballs in campaign politics. And so I became something of a connoisseur of public-opinion methodology.

Polls are sometimes wrong. I keep flashing back to Oct. 17, when a poll showed Democrat Dan Maffei leading by 12 points over Ann Marie Buerkle, who beat him on Election Day less than three weeks later.

Polls are not predictions. The further in advance of an election you take a poll, the less the poll can tell you about the outcome of the next election.

In March 2009, Obama’s approval rating was 64%. Eight months later, Republicans swept the off-year elections in Virginia and New Jersey, then in January 2010 won the Massachusetts special Senate election, and rolled on to a historic 63-seat House pickup in the November 2010 mid-terms. So, in the span of 20 months, Obama went from Mr. Can’t-Be-Beat to Mr. Can’t-Catch-a-Break — but none of that could have been predicted on the basis of the March 2009 polls.

This lack of poll predictivity is true, no matter whether you like the poll result or not. Stupid people reading political polls typically make one of two basic mistakes:

  1. Cherry-picking positive numbers. In other words, “Let’s believe the numbers that look good for Our Team, and ignore the numbers that look bad.” This is how Pelosi and the Democrats screwed up in 2010. They listened to consultants and strategists who told them to ignore poll numbers showing that ObamaCare was deeply unpopular with likely voters, because various sub-questions about health care still looked good for Democrats. And I remember in 1998 how Republicans cherry-picked polls telling them how popular impeachment was, and then got their asses kicked in the November mid-terms.
  2. Chicken Little. Doom-and-gloom panic at the first hint of negativity in the numbers is not helpful. It’s easy to get scared out of your shorts by looking at polls taken when the public is in a general mad-as-hell mood. But it’s important to remember that partisan affiliation seldom swings suddenly in one direction or the other. Even when parties put forward truly wretched candidates, they still usually get their basic share of the electorate: John Kerry got 48% in 2004; John McCain got 47% in 2010. So hysterical freakouts over a downward blip in the poll numbers is almost never justified. The sky is probably not falling.

What caused me to reiterate this lesson today is that a friend called to my attention a study released by a University of Washington academic, which claims to show a yawning gap between regular conservatives and Tea Party supporters.

Rift in the Right: Many Conservatives Reject the Tea Party’s Paranoid Views” makes for a catchy headline, but when you start scratching under the surface of this study, the stench of bovine excrement is overpowering.

Try as I might, I could find nothing there that told me how the UWash survey was identifying “conservatives” (except by asking people to label themselves), nor was their any apparent effort to screen for likely voters, or to match the “conservative”/”Tea Party” self-IDs to partisan affiliation. When you consider that 20% of self-identified “conservatives” voted for Obama in 2008, you realize that this label doesn’t necessarily mean to some people what most of us think it means. So taking a sample of random adults (i.e., 18-and-older, not screened for registered voters or likely voters) and then asking them whether they’re “conservative,” prior to surveying them on a bunch of tendentious questions, isn’t likely to provide you with any information that’s really useful, from the standpoint of electoral politics.

Much more relevant, it seems, are two separate polls — Gallup and Quinnipiac — showing Obama scraping rock-bottom. These numbers are something to keep in mind when (taking their talking points from Chuck Shumer) the media begin proclaiming that Republican “Tea Party extremists” are losing in the budget showdown.

I was born at night. But it wasn’t last night.

Comments

Comments are closed.