What About the Polls?

October 4, 2000

"How can I trust the polls?" "Nobody's ever polled me. Who are they talking to?" Ever heard some of those questions? I hear them all the time. So let me take a minute to answer a few of those questions.

When the Gallup organization conducts a poll, they attempt to put together a cross section of the American people based on various demographic factors. For a typical three-day poll of 1000 people, they buy several thousand numbers which are proportioned among the nation's more than 3000 counties. They purge business phones and nonworking numbers and then go to work.

The poll is closely monitored according to the proportions of the four major regions (Northeast, South, Midwest, and West) and the proportions of people who live there. When a poll is complete, Gallup makes sure the poll accurately mirrors the racial, age, and educational makeup.

So why haven't I ever been polled? Well, the Roper Center estimates that about 400,000 people were interviewed last year. There are more than 200 million adults in the U.S. So your chances of being polled are about 500 to 1.

But how can these polls be that accurate? The answer to that depends on the questions asked, the people asked, and the sample size. Tomorrow I'll talk about the first two issues because I believe this can skew a poll. But the last issue is relatively simple. It's merely a function of statistics, or more precisely what is called "sampling error." A poll of 1000 people will represent the view of 200 million adults 19 out of 20 times within three percentage points.

So are the polls always accurate? Not at all. And next time I'll talk about why you have to view many polls with a skeptical eye.

I'm Kerby Anderson of Probe Ministries, and that's my opinion.