Answer Man: Putting the Cars Before the Fact

By John Kelly
Monday, September 5, 2005

T he Labor Day weekend is upon us. We hear authoritative pronouncements from AAA Mid-Atlantic confidently predicting the number of persons in the metropolitan area who will be traveling over 50 miles outside the area during the holiday weekend. We hear similar predictions for Memorial Day and Thanksgiving. How do they know that? What methodology do they use?

Vincent McDonald, Fairfax Station

Answer Man has often thought that it would be useful to possess a time machine. He would use it to go into the past to rectify certain bad decisions -- an unfortunate quip at a social event, a poor choice of haircut, the more ill-advised portions of his senior prom -- and, occasionally, to peer into the future.

He also would lend it to AAA so it could leap forward, count some cars, then come back and tell us exactly what's in store for us, traffic-wise.

Sadly, such technology does not exist. And so AAA has to rely on a different method for its predictions of holiday traffic volume. Here's what it does: It asks people if they're going to travel.

Actually, the Travel Industry Association of America asks that question. To arrive at this year's Labor Day forecasts, TIA pollsters called at random 1,300 U.S. adults between July 15 and 25 and surveyed them about their travel intentions, said AAA spokesman Justin McNaull . The results of that survey were then massaged, with the massage oil being a proprietary forecast model developed by TIA's economist. Its ingredients are the results of the surveys the association is doing continually, along with such indicators as unemployment rates, consumer spending and gross domestic product. The final figure was then broken down for various geographic locations.

Last week the results were announced: According to AAA, 34.5 million Americans, including 856,000 Virginians and 550,000 Marylanders, were expected to travel 50 miles or more this Labor Day holiday.

How accurate are surveys such as this one? One problem with any survey that asks a question about a personal habit or decision is the possibility of a social bias. If a researcher asks how many alcoholic beverages you consume in a week, you might be tempted to lowball your answer, for fear that you'll look like a lush. If a researcher asks whether you intend to vote, you might say yes, just to seem as if you're civically engaged.

That sort of bias shouldn't apply here. Travel is not very controversial. Also, it's usually an activity that is knowable. It's not as if survey respondents were asked to predict what was going to happen in Iraq. People are able to control their own travel decisions.

Except when they change their minds, which might happen over this holiday. When TIA conducted its survey in mid-July, gas was a lot cheaper than it is now. Drivers, unwilling or unable to pay higher prices, might change plans.

There's another way to predict traffic flow, and that's basically to count vehicles. Governments do that in different ways. Sensors embedded in the pavement can count the cars and trucks that roll along the highway. There are also tubes that stretch across the pavement.

CONTINUED     1        >

© 2005 The Washington Post Company