Saturday, September 18, 2010

Clues 'n' polls

And we have a winner for Clueless Hed Orthography Trick of the Year!

Like it or not (I emphatically don't), you can see a sort of rudimentary logic in the tediously familiar pickin'-and-grinnin' hed. The desk is tryin' to show it's not a complete tool of the big-city liberal pantywaist elite, so it finds a way of sayin' it once watched three minutes of "Hee Haw" too. Here, though, we cain't even tell what they're reachin' for -- maybe a clever allusion to the stylebook's "rock 'n' roll" entry?*

As a dialect trick, it's a failure. The alliteration isn't worth the puzzlement, and neither is the reporting, which gets us** to another point:

Deal leads Barnes 45 percent to 41 percent, according to the poll conducted for the Georgia Newspaper Partnership. But, given the margin of error of plus or minus 4 percentage points, the race could be even closer.

Well -- sorta. But given the margin of error, the race could also be even less close! Indeed, given any margin of error at all, the gap could be either wider or narrower! Sure, the sample value of 45-41 could reflect a population value of 43-43. That's just as likely as a population value of 47-39 (or 47-43, or 43-39). Rather than talking about how close this result is to being a tie, we'd do better to talk about how likely it is -- from "not very" to "almost certain" -- to represent a real difference in the population.

A visitor last week took issue with the idea that reporters (and editors) ought to be picky when they describe statistical results as "significant." That's fine. We can negotiate on that point. (Come to that, if you guys will stop saying either "statistical dead heat" or "within the margin of error," the hostages could go free today.***) But I don't want us to give up on the idea of "significance" altogether, partly because of the role it played in one of the most successful Big Lies of the right-wing media this year: the alleged "Climategate" scandal.

Remember the gleeful "Climategate U-turn" heds, in which the "embattled" scientist allegedly dropped a "bombshell" by admitting there had been no global warming since 1995? He didn't, of course. Asked by the Beeb whether he would agree that there had been no statistically significant warming since 1995, he said "Yes, but only just." The positive trend, he added, was "quite close" to the significance level.

If we give up on "significant" altogether, we're handing the weasels a free pass. We shouldn't do that. They already have their own media system.

(While we're at it? Don't write "voters weigh in" kickers on stories like this. The respondents are considered to be likely voters; they aren't "voters" until they pull the lever on Election Day. And don't write stuff like "Some 625 likely voters were interviewed by telephone. Voters were randomly selected and distributed across Georgia," period. Unless they were distributed by the same person who sampled them.****)

* It pains me to admit it, but it isn't much of a stretch to imagine that happening.
** Yes, the link goes to the Rome paper, not to the AJC. It's really hard to find the day's print stories on the AJC Web site, which is hugely annoying.
*** Didja hear about the copy editors who hijacked a plane full of Gannett publishers? They're going to release one an hour until their demands are met.

**** And since you can't calculate a confidence interval from an approximation, never write "Some 625" when you mean "625." (That's a way of getting around the stylebook's ban on starting sentences with numbers, but it's not a valid excuse.)

Labels: ,


Blogger Brian B said...

Specifically, a 4-point lead with a 4-point MOE is a clear lead with 84% confidence.

(That assumes that this poll, as is the norm, reports its MOE at a 95% confidence level.)

Not exactly neck and neck.

7:26 PM, September 18, 2010  
Anonymous Anonymous said...

I would be overjoyed if journos would report "the survey found that there's an X% chance that Y would have beat Z if the election had been held last Monday, Tuesday, and Wednesday", but I'm not holding out hope as that would destory their precious horse-race narrative.

9:23 PM, September 19, 2010  

Post a Comment

Links to this post:

Create a Link

<< Home