My pet peeve (for today)

By Phil Lawler ( bio - articles - email ) | Sep 23, 2003

The 2004 campaign season is already beginning, and we can expect to see plenty of stories about public-opinion surveys. In an extraordinary number of cases, reporters will mangle the meaning of the polls.

This particular story is taken from the Boston Herald, but you can find similar errors in newspapers every day.

The retired four-star Army Gen. Clark, showing surprising strength after joining the race just last week, led Bush 49-46 percent-- a statistical dead heat given the poll's three-point margin of error.

Wrong!

The poll showed Clark leading Bush by 3%. Assuming that the poll is accurate to within 3% (a very big assumption), that means the margin could be off by 3% in either direction. Maybe it's a dead heat. Then again, maybe Clark is actually leading by 6%!

Now here's the next point: What does that "margin of error" really mean? The presidential election is almost 14 months away, so it doesn't even make sense to speculate on what would happen if the ballots were cast tomorrow. The poll doesn't predict any result at all, so "accuracy" shouldn't even be an issue.

Something to keep in mind: A shrewd political campaign manager will know how to read an interpret polls. But the campaign's PR mouthpiece probably won't; his job is to make the candidate look good, regardless of the statistics. Reporters generally get their information from the PR agents. There's an old computer-programming term to describe the typical newspaper analysis of campaign polls: GIGO.