Now that they’ve succeeded in terrifying parents and children across the country, the Homeland Security folks have admitted (in a way timed to run in the seldom-read Saturday newspapers) that the headlines they arrranged for yesterday, just ahead of the debate, such as “Terrorists Targeting Our Schools” were false.
Unless you happened to read the short squib on p. 20 of today’s LA Times, or checked Talking Points Memo, you might well have missed it the correction. And of course the atmosphere of fear remains even after the specific warning turns out to have been either a blunder or a hoax. That reflects reasonable news judgment: “danger” is news, “no danger” isn’t news.
But at some point someone might notice that DHS (like the Justice Department) is headed, not by a security professional, but by a career Republican politician. If Tom Ridge’s troops are doing fake terror alerts to move votes, someone ought to call them on it. If they’re doing it out of sheer incompetence, that’s par for the course. (Note that the maker of the anti-Kerry documentary Sinclair plans to air in prime time later this month is a former Ridge aide in Pennsylvania, who later worked in the Homeland Security publicity shop.)
Crying wolf is a dangerous game to play. The folks in Georgia left looking silly after they mobilized to defend schools that weren’t under attack will be less swift to mobilize next time.
And next time might be for real.
Since it’s come up in connection with this story, I’d like to offer a statistical caveat about the claim that terror alerts boost Bush’s popularity. There’s another possibility equally consistent with the gross facts, and as far as I know no one has done the statistical work to disambiguate the interpretation.
The alternative hypothesis is that Bush’s popularity fluctuates around a trend, and that terror alerts are more likely to be issued when it is below trend, because Tom Ridge and John Ashcroft are trying to stop the slide. If that’s right, then the apparent increase in Bush’s popularity as a result of a warning might be, at least partly, regression toward the mean.
I offer this as a problem set for a graduate statistics class. The more of the apparent power of a warning turn out to be a mean-regression effect, the less actual capacity the Administration has to gain votes by creating an October surprise.