Tuesday, November 08, 2005

Watch out for the gasbags

Before we get to anything, we'll note this from Robert Tanner's "Democrats Win Gov. Races in N.J., Va." (Associated Press):

Democrats swept both governors' races Tuesday, with Sen. Jon Corzine easily winning New Jersey and Lt. Gov. Tim Kaine taking Virginia despite a last-minute campaign push for his opponent from . . . [Bully Boy].

We're noting it here and that's all we'll be noting for some time. Why? NPR will pull Cokie Roberts out of moth balls to offer "analysis," the Sunday Chat & Chews will push their "insight."
No one knows anything other than the totals. Polling data will have to be crunched and studied.
But all the gasbags will assure you, over the next few days, that the two won because of ____ or because of ____.

After the 2004 presidential election, you saw the Times rush in with their report on their polling data. The poll finished and immediately, Adam Nagourney and Janet Edler were able to write a report on it. It didn't reflect the data in a number of cases. For instance (we covered this, there are four entries on one Monday devoted to the Times "report" -- check the archives for a week in November -- I think Thanksgiving week, but I could be wrong), to prove the myth of the "values voters" giving it to Bully Boy, they combined two answers that weren't really related (they were seperate answers, two of five, to one question). If they'd combined two on the other side, it would have been disproven.

As a poli sci major, I don't tolerate instant analysis and turn it off when it comes on the radio. It's embarrassing but apparently journalists like to pass themselves off as experts, some just aren't willing to do the work required. Journalism majors, sorry to ed majors, were seen by poli sci majors as several notches below education majors or general studies ones.

They know their tiny field (or some do) and they then run with conventional wisdom and back it up with selected comments or statistics. In the poli sci field, an election will be analyzed for months. In the "journalism" field, it will be done in about ten minutes and then they'll all glom off of whatever sticks.

So the gas bags will be screaming, "The election was decided by ____" and anyone with even one research & methodology class under the belt will be cringing and wondering how they can perpetrate this crime on the country because, my opinion, it's criminal to offer "snaps" as probing insight.

For those who enjoy "snaps" watch to see who tries to push it as, "It had nothing to do with Bully Boy!" They will really be reaching. Polls for the last few months demonstrate that his numbers have fallen (a trend that holds, from various polls, tends to be reliable but you should always examine the raw data and consider the sample size -- the Times is fond of using 900 people, or less, to represent the country).

With that and with the fact that Bully Boy campaigned for the loser, a hypothesis that he aided in the defeats could be offered. It's reasonable. It's just not proven and won't be tomorrow when everyone rushes to the mike to be first out of the gate. But watch the ones who try to dismiss, citing "facts," Bully Boy's damage to the campaigns.

No one knows anything. But the nature of daily journalism is you must provide an immediate answer. So they'll all rush out to offer opinion presented as fact.

I won't take part in that here. To me, it's like watching a journalist glance at someone and then offer a diagnosis that is "correct."

Add in that (no offense to Rebecca) the p.r. firms long ago learned how to manipulate and sell data. (Faith Popcorn, anyone?) It's insane.

And brings up the issue, often debated, of whether or not political science or any of the other "soft sciences" are truly science? They aren't when "snap" is presented as informed.

The numbers need to be crunched, the data needs to be reviewed before anything's offered and then we're still dealing with hypothesis and not theory (hopefully peer reviewed hypothesis).

The press encourages that, the daily press, and they do the same thing with the box office numbers each weekend. Running with estimates that are often high balled and corrected (though never by the paper of record) later in the week when all the numbers are in.

(A number of "number one for the weekend!" films aren't, in fact, number one when the box office returns come in.)

But the dailies (paper, radio and TV) always need something to run with each day so they run with it even when there's no basis for it. The Times now prints "overnights" (but not, for some strange reason, the official totals for the week) and I doubt most people reading the "overnights" in the paper grasp that this is "select markets" and not a national indication of what was watched and what wasn't.

There's no reason to print overnights. That's like announcing the results of a presidential election with less than twenty cities in the country reporting their totals. But apparently, it's more important to toss some junk out into the daily cycle than to toss anything that's actually accurate.

The Times may let Nagourney and Elder push spin (by carefully selecting from the raw data that they don't have access to for a long enough period to even make generalizations) next Monday (or the one after) but that won't make it accurate or even useful. If you go to the archive for those four entries, please note that when I wrote it I had no idea that Elder was over polling for the Times. Had I know that, I would have been much more caustic. I assumed, wrongly, that she was a junior reporter assisting on the piece. If she has any knowledge of polling (there's no reason why anyone holding that job at the paper necessarily would), the article on the "results" of the poll are even more offensive.

The e-mail address for this site is common_ills@yahoo.com.