Monday, September 16, 2002

More Polls, More Blather

Yesterday, I dissected some poll numbers from the Commercial Appeal story claiming an 8 point lead for Bredesen in the Governor's race. Today, the CA has a second story, again from the Knoxville News-Sentinel's Tom Humphrey. And it's more of the same.

The same organization conducted this poll, and claims an 18 point lead for Lamar! Alexander over Bob Clement:
"Alexander has all but got this thing
wrapped up - unless something really
strange happens," said Dr. Michael
Gant, a University of Tennessee,
Knoxville, political science professor
and director of the UT Social Science Research Institute.
Well, hoo doggies! Let's just call the whole thing off and save some taxpayer's buck why don't we? The sages have spoken. It's clear to see why they went back to this guy for their numbers.

But let's look at those numbers. Remember, the margin of error is again four points.

Alexander: 45%
Clement: 27%
Undecided: 20%
Other: 8%

First, the Undecideds are a large lot. In a re-election, they'd be expected to largely go to the incumbent. But in an open-seat race like this, it's a harder call. Many may stay with the seat's party -- Republican. After all, at the State level, Tennessee is still a Republican-leaning place, newpaper agendas aside. But read further into the article and you come to these poll numbers:
Clement spokesman Carol Andrews questioned poll results,
saying a Democratic National Senatorial Committee survey
released last week found the race considerably closer - Alexander
at 49 percent and Clement at 42 percent.

An Alexander poll in late August put the margin at 14 points,
53-39 percent.
Now I'd be far more willing to take these numbers, even being at the opposite edges of the margin-of-error spectrum as they are, because they closely track each other. Each finds the candidates within four points in both polls.

Let's take that 50/40 split to the Gant poll's Undecideds. Doing so, you get -- Alexander: 55% and Clement: 35%. And that tracks with the two partisan polls, but with bad news for Clement!

But Gant's propheseying aside, it's still not good news for Lamar!. Remember, he's been running saturation ads for weeks now and Clement has been pretty much not. Expect those numbers to tighten up when Clement launches his own assault. Give me another poll, after a couple of weeks of Clement ads and we'll see then.

Let's let Gant pontificat some more, with some numbers:
Gant said that Clement can be expected to draw closer when his
advertising gets under way, but said it would take some "ugly
skeletons" from an Alexander closet to put the Democrat ahead.

"Once they start spending, they're going to close the gap. But it's
a huge gap now," said Gant. "Even with a good scenario for
Clement, I can't see Alexander dropping much below 55 percent
by Election Day."

The poll indicated considerable support for Alexander across the
political spectrum. He was ahead in all three of the state's
geographic grand divisions and by more than 2-to-1 in his native
East Tennessee, which is predominantly Republican.

Twenty-two percent of poll participants who consider themselves
Democrats said they would vote for Alexander; 8 percent of those
considering themselves Republicans said they would prefer

A third of those who said they would vote for Democrat Phil
Bredesen in the governor's race said they preferred Alexander for
the Senate. As reported on Sunday, the poll found Bredesen
leading Republican Van Hilleary by 8 percentage points in the
governor's race.

"Bredesen and Alexander look an awful lot alike to me on many
issues," said Gant. "You can make the case that the difference
between them is no greater, and maybe less, than the difference
between Alexander and Hilleary."
Jeez, does this guy love the sound of his own voice. He's a hired pollster, not a commentator, for heaven's sake! Let the numbers do the talking.

The article does have some interesting numbers about the softness of party identification and fungibility of voters. The willingness of folks to cross party lines is pretty impressive, at least as this poll sees it. But let's wait until mid-October until we start really calling these races, OK?

OK, then.

(Apologies to South Knox Bubba for stealing his line there.)

Until next time.

INSTANT UPDATE: Just after posting this, I found Justin's Elephant Rants, Part One and Part Two, on the subject! He catches some good things I missed and helped jog me on something I didn't cover above.

The way that polls are constructed influences the results you get. In the two campaign polls I mention above, the choice offer to pollees was simple A or B, Alexander or Clement, so the results were pretty stark as well. But the Gant polls, as Justin shows, make use of "support" and "strongly support," "oppose" and "strongly oppose" constructions that create vaguer, more malleable results. In a small sample like the Gant polls (roughly 600 people), to break them into smaller categories creates larger and larger margins of error, and consequently less reliable results.

Also, journalists are, as a group, deeply deficient in math, sciences, economics, etc. Most only study journalism, English or humanities in college. Very few have any serious background in the necessary physics or science or math needed to critically analyze what they write about. In fact, many lack true "critical thinking" skills, only learning the reporter's "neutral/oppositional" stance. It shows, time and again, in gullible and sloppy reporting.

I am involved, in my real life, in a science-based hobby. From time to time various papers around the country write about it. In the hobby's USENET newsgroup, we are usually apprehensive about these stories, because the reporters who cover us almost always are utterly lacking in sufficient science to understand what they write about. Again and again, the most egregious statements get made by reporters who are simultaneously ignorant and self-confident -- lacking even the most basic science education, but thinking themselves completely capable of writing knowledgeably about it. Of course, their errors get into the story and our corrections are buried later. And the reporters go on to other areas they are still ignorant in, misleading the public and misinforming them.

It's long past time journalism schools addressed this situation. But I don't hold out any hope of that happening. Reporters are possessed of a supernaturally arrogant self-esteem born of immunity to outside criticism.

Until next time.

No comments: