Bush has 7-point lead (oh, never mind)

That's what the headline really should have read on this morning's USA Today story about its latest presidential poll because down four grafs was this:

Bush's lead remains within the survey's error margin at Labor Day, the traditional start of the campaign's homestretch. ... By historical standards, the race is too close to call.

In other words, the results don't mean squat. The margin of error is 4 percent (percent, not percentage points, because it refers back to the base). So Kerry could just as easily be ahead 50 percent to 44 percent. When will we as journalists learn that statistical oscillations within the margin of error are not screaming headline news? (Technically, Bush could be ahead 90-10 or Kerry 80-20, too. Remember, one time in 20, the entire thing could be wrong, but I'm not advocating we scrap the whole process -- just report it correctly).

It is interesting that the margin seems to have widened, but we can't be sure. So report it, but don't scream it.

The more significant news was buried in USAT's story -- improvement on views of Bush's qualities to be president (+6) vs. Kerry (-14). Bush's improvement over Kerry in ability to handle terrorism (27 points, up from 10 points last month) and the surge in terrorism as an important issue on par with the economy: All these are much more significant as potential trend markers.

In the first of three panel studies done by the Center for the Study of Elections and Democracy at Brigham Young University and the University of Wisconsin Advertising Project -- in July -- 81 percent of those questioned (N=2,782) said they already had made up their minds on whom to vote for. And neither candidate could find a majority of people willing to say "_____ cares about people like me." (Bush 47.1 percent, Kerry 46.2 percent) And on who was seen as a strong leader, 54 percent said that described Bush extremely or quite well, vs. 37 percent for Kerry. So this is not something new after the conventions; the groundwork already was there.

(There is much more in this study, which oversampled battleground states and tends to have a slightly older demographic. Link to full data.)

To me, that -- and the stuff buried by USAT -- are far meatier than trumpeting a statistic that might not even be true. When will we learn to handle numbers so that the public doesn't hold us in even less regard?

Columbia Journalism Review's Campaign Desk had a great article on polls and such. It should be required reading for every editor and reporter before touching a poll story.

And if you want to read a rather spirited outpouring on this, check Romenesko's letters column, where one writer started things off by suggesting:

A three point lead for Bush, even in a poll with a four point margin of error, means that there is a small probability that Kerry might actually be tied, or even ahead, but that it is far more likely that Bush is ahead (and most likely that he is ahead by three points.) Polls are an estimate of what millions think based on what a few hundred think, nothing more than a probability. If they have any use, then reporters ought to report a small advantage as an advantage.

Early returns find him in the minority.