Tuesday, November 13, 2007

Selzer Gloats Over Star Poll Results

Ann Selzer, president of Selzer & Co, an Iowa-based public opinion research firm, witnessed her professional reputation coming under intense attack when her firm conducted a poll on behalf of the Star and WTHR, which showed Mayor Peterson only slightly ahead of Republican Greg Ballard and well below 50%. Selzer pens a My View column to share her thoughts about the criticism in the wake of Ballard's upset win over Peterson. She writes:

Our last poll before the election, published Oct. 25, in the Star, showed a very tight race. That by itself set this poll apart from everything else that had been released up to that point. While incumbent Bart Peterson held a lead, it was slim and vulnerable. His approval rating was just barely 50%, which is low for an incumbent. It was lower among independent voters.

Independent voters may have given Ballard the edge. They are typically less likely to show up on Election Day than the party loyalists, and Ballard led this group by six percentage points.

Another theory to explain Ballard's victory is that fewer African-American turned out on Election Day than early polls may have estimated. They favored Peterson by more than 7-to-1 margin, according to our poll, so if that group turned out to be less than is proportionate to their presence in Marion County, that would affect the outcome as well . . .

Our poll did not show Ballard winning, but we did see his rise. His victory was certainly foreseeable based on our data. And it was the only public poll showing this to be a close race.

You may recall the poll's controversy ignited after the Star initially released results showing a dead heat race with Peterson leading Ballard only 41%-39%. Soon after those numbers first appeared at the Star's online website, they were altered to show a Peterson lead of 43%-39%. The Star added a disclaimer explaining that the results had been revised to more closely reflect voter turnout based on the 2004 election. Selzer explains the pressure her firm came under at the time:

There were pressures in the marketplace that, had we given in, would have shown Peterson much farther ahead. I'm glad we held our ground in the face of some nasty commentary. The readers of the Star and the viewers of WTHR were well served by having early notice of an unexpectedly tight race. What we delivered was our best work to reveal the truth of political intentions in Marion County.

With all due respect to Ms. Selzer, I must disagree with her contention she didn't give in to marketplace pressures. The altering of the poll results after its initial release reveals to this writer her firm and the Star did indeed give into pressure from the Peterson campaign. The fact that the poll results were altered and done so in such a public manner cast doubt about the validity of her poll, a fact seized upon by the Peterson campaign and others within the media. As a consequence, the Selzer poll got dismissed by many as completely unreliable. It's little comfort that her poll results were closer than other opinion polls. As I recall, there was only one other poll later in the campaign, a WISH-TV poll, which showed Peterson hanging near the 50% mark with many expressing disapproval of his policies.

This whole ordeal leads one to wonder how often pollsters fudge their numbers to satisfy a marketplace pressure. The fact Selzer acknowledges such forces exist is enough to give one pause. And contrary to her claims today, we know her firm gave into such pressure last month after it initially released the poll she now boasts got it more right than anyone else. One of my biggest fears about polls is how they often become a self-fulfilling prophecy. Too often they are used to convince us a certain political race isn't close at all so we shouldn't even bother to come out and vote. I know for a fact that some local Republican leaders cited the WISH-TV poll as evidence Ballard couldn't win and the party should instead focus on the council races. Personally, I read that same poll a completely opposite way because it showed such a high degree of dissatisfaction with the city's direction and weak support for the mayor. I now must question whether the WISH-TV pollster inflated Peterson's margin to match some anticipated voter turnout scenario. It makes you wonder.


Anonymous said...

I've been involved in multiple polls for many candidates over the years. I always subscribed to the theory that you hired pollsters based on theri tactics and professional accountability. That's never cheap.

Trust me--Selzer's dramatic and unprecedented withdrawl of results, recalculation and re-release, is being much-discussed among pols nationwide.

They used a statistically and demographically unprofessional and unrealistic sample group. They rolled the dice and predicted it correctly.

But their credibility is seriously questioned.

And FWIW, my polling experience tells me the more-espensive polls are generally more accurate and reliable, even when they give you results you don't like.

Selzer's poll was likely cheap. Her sample group, and the questioner (I got their call) were entirely unprofessional.

The Mayo'rs folks didn't ask that the poll be withdrawn. That was none other than Amos Brown. In yet another example of his unbiased and straight-forward journalistic style.

The Star and 13 milked the story for all it was worth. Don't forget, to explain the unusual polling methods and results, 13 rolled out Mrs. Ryearson, without disclosing her personal relationship...

All in all, not journalism's finest hour. Yeha, they got it right, and in this media market, that's all that will likely be remembered. But among pols, even Republicans, their tactics are in serious doubt.

Hell a broken clock is right twice a day. It wasn't difficult to predict this mayor was in trouble: he had a trifecta of two tax revolts and a close relationship withi a council leadership team whose tactics were given the boot last Tuesday. (Justifiably)

When Selzer puts together an accurate sample group, and gets another one right, her professional image will be back to 50-50.

No,. this is not sore loser talk. It's an objective review of political data and tactics. Ballard didn't ask for this poll...in fact, I think it gave him a startle, and almost made him peak too early.

Gary R. Welsh said...

anon 9:55, I worked in a state rep campaign over in Danville, Illinois back in 1986. MRG out of Michigan was doing polling for HRCC. We didn't think their polling in our district was accurate at all. They were about to write our race off in the final days. A veterinarian, who studied polling as a hobby, devised a poll for us. We assembled a group of 12 callers and sampled a little more than 300 voters in the two county district. The poll predicted the outcome of the race almost to the tee. My candidate won 52-48%. The poll results even had the county results down to a tee. It showed us in a dead tie in one of the counties--we won that county by less than 100 votes. The poll cost us nothing and was much more reliable than the poll numbers MRG was feeding the campaign committee. So the amount of money you spend on a poll is not always indicative of its accuracy.

Gary R. Welsh said...

And I should add on that comment, the chief of staff for the House Republicans tried to get me fired because I did what we had to do in that race to win and didn't follow her horrible advice. My boss asked her, "How can you fire the guy for winning a race?" A few months later she ran off and married the chief of staff for House Speaker Mike Madigan. I learned long ago that often your real enemies are within your own party.

Anonymous said...

ah yes, the infamous marriage between Chris Freveletti and Gary LaPaille. Are they still married?

Gary R. Welsh said...

I believe they are, anon. 1:25. I've tried to block out my memory of them.

Anonymous said...

Gary, your personal poll story was fascinating.

But over time, the good pollsters are usually right. Their techniques are time-honored and expensive. Woe is the politician who fails to listen to what pollsters say.

Not that politicians should blow in the wind...principles are to be held, for sure.

But Selzer used faulty methodology. And frankly, they predicted Peterson on top by a few...didn't he lose last week?

I wouldn't trust their results.

But I'll tell you this--I know for a fact, Bart was in the field at the exact same time as Selzer. He used trusted pollsters whow ere respected and expensive. Those poll results were never shared, not even with most council candidates.

I'm betting they showed Bart was in trouble, and not likely to win. If there were anything hopeful at all in those rresults, his campaign would've leaked it. They didn't. Mum was the word.