Saturday, July 28, 2007

Fun With Numbers: Memphis Mayor's Race Department

FOX13 just released their own poll on the Memphis mayor's race but it is, as the two previous Commercial Appeal polls were, generally useless.

If you look at the only piece of hard data on the poll, it's a +/- 7% margin of error (MoE). Such a large error margin is usually not a good sign, so I called Yacoubian Market Research, the folks hired to conduct the poll, to get more information. Turns out, the sample size (the number of people asked the poll questions) was just 300. That's a very small sample size, one prone to a lot of error in its results. Hence, the 7 percent MoE.

Berje Yacoubian was satisfied by it. He pointed out that the only real result you could draw from the poll, the FOX13 story notwithstanding, was that Herenton was in trouble. His support has slipped dramatically. "Over a year, his popularity is low and sinking."

He said that this poll was meant primarily as a baseline against which polls between now and election day can be compared. "I'm not trying to figure out who is going to win or lose."

The numbers in the poll that matter (outside of the "Who would do a better job?" stuff)

He did offer some interesting observations based both on his thirty years of polling and what he's learned in this election cycle:

1. White voters break out 52% for Chumney and 20% for Morris. As the number of undecideds dwindles as we approach the election, that's bad news for Morris.

2. Chumney seems to be the choice of the not-Herenton voter. Morris isn't getting what Yacoubian considers his "expected level of support."

3. Conversely, he notes that as Chumney weakens, a possible trend in the flawed Commercial Appeal polls, Morris does benefit, although that's an obvious "last man standing" artefact.

4. Yacoubian expects Herenton to wait until September to launch a "four week campaign." It will dramatically change the race, and not to Herenton's benefit as voters will be forced to consider what Herenton's done up to now.

The poll did find that Morris is the second choice of most voters, some encouraging news to Morris supporters. Anything that cripples Herenton or Chumney stands to benefit him. But with such a large MoE, Herenton and Chumney are in a dead heat (something the Commercial Appeal poll found) and Morris is just below the MoE compared to both. Without a larger, better poll, it's pretty much anyone's game right now.

Yacoubian also didn't have kind words for the Commercial Appeal polls. He felt they traded their "neutrality for cheerleading. They made news so they could report it."

I agree with him. Wharton was not a candidate. He did not pull an application at any time. He had expressly stated he wasn't a candidate and wouldn't stab his good friend Herenton in the back. He was very happy as County Mayor, a position he felt (with some justification) was superior to the City Mayor. He and Herenton had a long personal history that Wharton would have had to callously jettison and then repudiate in order to run a campaign. For the Commercial Appeal to have pushed him as a possible candidate was partisanship of the strongest kind.

For the Commercial Appeal to have included him in the polls was just the worst kind of pot-stirring, a kind of underhanded journalistic story-making that Chris Peck should deplore. At a past Dutch Treat Luncheon, editor Blake Fontenay defended it as just reflecting the public's stated wishes. I didn't buy it then, and would be harder pressed today to buy it. They were ginning up a slim, dim fantasy.

The Commercial Appeal always claims to "tell the stories of Greater Memphis." But it has absolutely turned a blind eye to the most obvious story of all. For all that the campaign coverage has focused on the foibles and follies of Mayor Herenton, and for all the coverage given to his erstwhile opponents -- both real and imagined -- Herenton is still the man to beat. The story is his strength as the incumbent. Barring surprise or shock, he's going to win. But you don't see too much of that in the Commercial Appeal, do you?

You should wonder why.

It has bothered me to no end to see the reporters in this city make the most absurd statements about some of these polls. Jackson Baker and Richard Thompson have been among the worst, but let's not leave out the Commercial Appeal editorial staff.

Even when a poll's margin of error is 4%, Baker and Thompson have both referred to changes of less than the MoE as if they were actual evidence of real, concretely measured, results. Jackson Baker I know should know better because we've previously discussed this very thing. But no, he'll happily cheerlead for Carol Chumney anyway:
I had just seen a report on Fox 13 News about a fresh poll taken by my friend Berje Yacoubian showing Chumney to be leading a second-place Herenton and a third-place Herman Morris.
Nope. As we've just discussed, her "lead" is only half of the MoE, meaning it's anyone's guess who is ahead in that poll.

I mean really, is it so hard, so expensive for these folks to head on down to any bookstore and just pick up a standard text on statistics? They don't even have to read the whole thing! Just the first few chapters should give them enough understanding to make their analysis useful instead of ridiculous. Heck, eve a good used bookstore ought to have something, if money's an issue. Or do I have to embarrass them by making a gift of a book and then posting a picture of me making the presentation here on Half-Bakered?

Because you know I will.

No comments: