Thursday, March 27, 2014

Georgia Senate Race

A new poll is out on the GOP primary battle among a gaggle of Republican hopefuls. This new poll, like a previous one, has the David Perdue leading the pack.

Okay, that's the news lede. Let's look at the story itself, specifically how it reports the methodological details.

Now there are certain things you should always report in a poll story: the margin of error, how many surveyed, and how they were surveyed. It's that last one that gets messy. Telephone survey? Fine. But what kind of telephone survey? A robo-poll (those annoying push-a-button-to-answer things)? Were cell phones included in the sample? If not, why the hell not? Purists would also want other details included -- and I'm among them -- but let's stick to the basics here and look at the story. Here's the lede:
Voters will soon be heading to the polls to decide some major races, including who will replace retiring U.S. Sen. Saxby Chambliss.
Really, that's your lede?

Really, that isn't your lede, but this online version is written like you'd do it on local TV. The above sentence is fine as a lead-in for TV news (okay, it still kinda sucks, but we're talking local TV here), but it is inappropriate for a print version of the story, which is what we have.

The next two grafs are quotes from voters. Huh? They come out of nowhere. I'd fail this story if a student wrote it in my class. Well, it's clean otherwise. Maybe a "C."

Okay, but enough bitching about the writing. Let's get to the poll details.
  • 600 likely Republican voters
  • margin of error, 4 percent
  • um, and that's about it
What kind of poll? Robo-calls? If so, those do not include cell phones. That skews your sample older and more conservative and, frankly, makes it less useful. But I can't tell if it's a robo-poll. Those quotes don't help. They could have been at the end of a robo-poll where folks could say what they thought. Or it could be a traditional telephone poll and they just typed what the respondent said.

We don't know. And that's the problem.

No comments: