Saturday, June 15, 2013

Really stupid stuff you shouldn't do with polls

Q: What's worse than writing a bogus poll story because it fits a sports narrative?
A: Writing a bogus poll story because it fits the party line!

I think we've discussed this point before, but one way you can distinguish the professional press from the party press is that the professionals screw up survey reporting evenhandedly. If the New York Times decides to round all its confidence intervals to the nearest whole percentage point, that's a dumb decision, but it applies with equal dumbness to all results -- not just ones the Times allegedly likes or dislikes. Similarly, horse-race writers will write about who purportedly clawed whose way back into contention after the candidates traded jabs when the gloves came off, regardless of how the paper's endorsement is going to go. The horse race is a silly narrative, but it usually operates independently of party alignment.

The party press is a different creature. (I'm fond of this example from 2008, in which a 3-point lead* is "Within Margin Of Error" when the Kenyan Muslim usurper is ahead but "McCain Out Front" when the home team is winning.) Its goal is to tell the audience what ought to be happening. And the risk with letting your wishes influence your results -- trying to be objective here, kids -- is not just deluding the general-purpose viewer but deluding the paymaster. If anybody deserves a clear-eyed account of the Massachusetts Senate race, it's the American side -- right, Fair 'n' Balanced Network?

The Senate race is Massachusetts is deadlocked heading into the final weeks** – again challenging the assumption that a little-known Republican cannot win national office in such a heavily Democrat state.
There are no values of "deadlocked" for which this is true -- at least, not on any evidence the story presents, or any evidence from the lone specific source the story cites. Above is Real Clear Politics' summary of the four most recent polls in the Gomez-Markey race. In the two latest, the Democrat has a 7-point lead; with a sample of 500 in each case, that means a "margin of error" of 4.4 points at your standard 95 percent confidence. If Fox News tells you that's an even bet, Fox News had better hope it gets to the county line before you and your friends with the tar and feathers do.***

But there must be other reasons Fox is fronting the story, right? 

Yet most polls show first-time Republican candidate and former Navy SEAL Gabriel Gomez within several percentage points of Democratic Rep. Edward Markey -- in the special election that sends voters to the polls June 25.

What makes "yet" the appropriate conjunction here is a question for the ages, since we seem to be trying to show that what was claimed in the lede is true (which it is, as long as "several" means "somewhere between 7 and 12"). What the dash is doing there -- perhaps to remind us that Fox can't edit for syntax either?

Political strategists gave Republicans little chance of winning when former GOP Sen. Scott Brown decided not to run -- considering he had just lost his own reelection bid for the state’s other Senate seat. And the state GOP party was left with a field of lesser-known candidates.

However, Gomez, the former Navy SEAL, emerged successfully from the primary and has continued to build name recognition beyond Boston where he works as a private equity investor.
(Hey, did we mention he's a former Navy SEAL?)

Gomez press secretary Will Ritter said Wednesday his candidate gained popularity and strengthened his poll numbers by relentless campaigning -- including 50 events since the April 30 primary.

He's certainly welcome to his opinion, but if you're a desperate Fox viewer counting on him to forestall the Kenyan Muslim apocalypse -- sorry. Gomez's poll numbers appear to be right around the peak he reached at the beginning of May. See above under tar and feathers.

“I know it sound cliché, but the more Garbriel Gomez talks, the more people like him,” Ritter said. “He wakes up at 5a.m. and wants to campaign.”

No big deal if you don't follow AP style on "5 a.m.," but would you mind spelling the candidate's given name correctly? Just a thought. (And did the flack really say "I know it sound cliche"?)

State Democrats privately acknowledged that this spring’s Boston Marathon, in which two terrorist bombs killed three and injured hundreds of others, was a tragic yet expected twist of political fate for Gomez.

"Tragic yet expected"? Do the authorities know about this?

Having crossed the finish line within minutes of the explosions, he emerged as a frequent TV eyewitness while the other candidates in both party primaries shut down their campaigns.

“What kind of evil people would do such a cowardly thing,” Gomez said on Fox TV. “We need to hold these people accountable.”

Well, that must have shaken things up.

... Markey, who entered the race as the presumptive frontrunner, has relied on campaign experience gleaned from 20 terms in Congress to try to win the race, while his campaign doesn’t over-react about polls numbers.

We have stylebooks so you don't have to guess about "over" prefixes. What the "while" is doing there is anyone's guess.

“Ed Markey has distinguished himself. And we’re in the driver’s seat,” said Markey spokesman Andrew Zucker. “Gabrielle Gomez is not for the middle class.”

If you're going to spell the candidate's name yet a third way, do you think maybe you'd want an option that's not as ... girly as "Gabrielle"?

Copy editors -- good ones, at least -- work on multiple levels at once. They try to see that copy is mechanically sound, and they they try to make sure the things it's sound about are true. We don't single one or the other out as the important part of the job, because they're both the important part of the job. If push came to shove and I was stocking the survival bunker with ammunition, aluminum-foil tricorn hats, and freeze-dried food, I suppose I'd want Fox to get the facts right at the expense of the spelling. But then there'd be no excuse at all for making up stuff about poll results, would there?

* Maximum margin of sampling error, +/- 3.3 percentage points at 95 percent confidence. Give Fox credit for scrupulous consistency in reporting its methods/
** It's already in "the final weeks"; the election is June 25, so it's heading into its final week.
*** NB, Your Editor is not predicting the outcome of this campaign, and neither are the polls. We're talking about measures of public opinion as they appear to stand at this writing. If you want predictions, take your tar and feathers elsewhere.

Labels: , , ,


Post a Comment

<< Home