Tuesday, July 11, 2006

More poll vaulting, with thoughts from Dr. Reid ...

Several people in and about the leg -- and about town -- have sent us interesting thoughts about last week's Free Press/Probe poll. Some are too rude to repeat. However, others have some points we hadn't thought of and we will try to weave them coherently here.

It was pointed out that inadvertently, Free Press Editor Bob Cox himself may have made a pretty strong case against his own poll. In his most recent blog he writes about the debate the Free Press editorial staff had about publishing the results of their online party preference poll, which was posted Friday and produced significantly different results than the Probe outing.

The results, not published until today, said that out of more than 1,600 respondents, 67% would vote NDP, 20% would go Conservative and 10% would hang in as Liberal.

"Web polls are not scientific," Cox opines. "They simply reflect the opinions of a self-selected group – people who went onto our website and voted."

We will not get into his interesting use of the word "scientific" (any actual scientists want to get into this?), but the notion that web polls are not really credible because the participants are self-selected is a valid one. They only reflect the views of those who wish to participate.

What is a growing problem in the market research industry, as we have stated here before, is that increasingly, traditional "scientific" polling is subject to the perils of self-selection.

This is described by former Manitoba pollster Angus Reid in an interview about the last B.C. election.

Higher and higher refusal rates due to things like voicemail and call display screening are what Reid calls "the big dirty secret of the industry."

Further, he warns, is that media organizations that pay for polling are doing it today increasingly on the cheap. "No one's paying for really good high-quality polling," Reid says in the same article.

This is the man under whom Probe principles Scott Mackay and Chris Adams learned their craft and they may be well advised to listen to him.

Apart from high refusal rates leading to increased unreliability of public opinion polling (see our recent post about widely divergent public numbers about the McGuinty Ontario Liberals), the phenomenon Reid describes leads to other major problems:

One, that the polling firms may be cutting some corners to deliver very expense-adverse, but high-profile clients, like newspapers, to help grow their business. And two, the media organization has no idea what may be behind the alleged trends they are reporting.

Witness some excerpts of the Freep's coverage of the recent Probe poll:

- "The Manitoba Tories are no doubt benefiting from publicity surrounding the election of McFadyen as leader as well as the inevitable fatigue voters get when a party has been in power as long as the NDP, [Scott Mackay] added."

- "The Conservative brand is hot right now and I'm sure that is helping (the Manitoba Tories)," Mackay said.

- "The Doer government has been hounded by accountability issues and claims of mismanagement at arms-length organizations like the Crocus Investment Fund and the Workers Compensation Board. They have also faced recent problems within the child welfare system," states unattributed analysis in a news story about the poll.

- "It is an accumulation of a number of little things like the Crocus Fund, issues to do with the state of the highways and infrastructure and the sense that nothing is being done," [U of W political science prof and former Liberal candidate Allan Mills] said.

What is the bottom line to all these insightful analyses? The writers and speakers of these statements are in effect admiting they don't have a clue why the Probe numbers might be what they are. All they offer are guesses picked from the air.

Mackay as much as admits this in the original Free Press story: "The reasons for the Tory surge are hard to identify, Mackay said."

That's right. This illustrates the further problem Angus described about media getting polling on the cheap. They don't pay to ask questions that might illuminate why they see the trends they're hot to trot to report on.

Just a simple issues question about why people are switching "in droves" to the Tories would given the Free Press at least a little leg to stand on for all the prognostication. But the Freep is too cheap but to pay for a few questions on the Probe quarterly omnibus (that means it's a poll where many companies and organization have bought a question or two about just about anything -- from beer and condom preferences to favourite political party).

Witness Frances Russell's hand-wringer on Saturday about what is going on -- attributing the "slide" to the province's current crop of rough roads. She quotes the ever-quotable Mackay:

"Infrastructure is huge. It's never been this high. It is the single biggest top-of-mind issue," he continues. "In the past, classic issues like health care and crime always predominated. Infrastructure has only arisen in the last couple of years. It used to be a seasonal thing. It would happen in the spring. But now it seems to be just sticking there."

OK, fine, if that's indeed the case. But what is the correlation between infrastructure being a top-of-mind issue and voting intention? Nothing shown. No evidence whatsoever.

Health care has traditionally been at the top of that list. Does that mean it's a driver of voting intention? Sometimes it has been. Sometimes not. BUT YOU NEED TO ASK THE QUESTION IN ORDER TO DRAW THE LINK. The Free Press, Probe, no one involved in parsing these polling numbers has any evidence, just guesses, as to why they have these numbers.

Even Dan Lett thinks linking it to infrastructure doesn't hold water. But again, his reasons for why it might actually be are just his guess.

Further from Russell's piece:

"The last year has been particularly rough, including issues such as the collapse of the Crocus Investment Fund, problems in child welfare and workers' compensation, the OlyWest controversy, even the "spirited energy" Manitoba rebranding exercise."

Whaaaaa? Even "Spirited Energy"? Frances, you've got to be kidding.

This grasping at straws for reasons to explain the poll should be telling our intrepid fourth estate that maybe they should be just a little bit skeptical. But no, apparently not.

Another piece of info forwarded to us was an interesting analysis about the Probe numbers and the right track-wrong track question that has been asked on government polls more or less annually since the end of the Filmon era.

Now, right track-wrong track is not the same as party voting intention, but generally, it says whether the public thinks the government of the day is overall, well, on the right track. It's often considered a key indicator of the public mood.

In December 1998, a government health poll showed just 41% of Manitobans thought the Filmon government was "on the right track", versus 48% who thought it was on the wrong track.

In March of that year, Probe showed that the NDP and the Tories were tied 35-35, with the Libs at 26.

We all know what happened in the 1999 election.

Fast forward to February 2004 when the government poll showed 53% of Manitobans thought the Doer government was on the right track, versus 40 who thought it was on the wrong track.

In March of 2004, Probe showed 47% for the NDP, 29% for the Tories and 19% for the Libs.

In January of this year, the government poll shows 62% of Manitobans think the Doer government is on the right track, versus just 31% who think it's on the wrong track.

In March, Probe showed the NDP and Tories tied at 41% and, of course, now shows the Tories ahead at 43-38.

What this shows is at a time when in fact the government "right-track" numbers are actually increasing, the Probe numbers show things narrowing and the Tories (whose leader could probably not be identified by the vast majority, regardless of the coverage of the convention) now, we're told, in the lead.

So, in case you missed it, something is not quite making sense out there. Couple that with the Environics numbers continually showing Doer at or near the top of the premiers' popularity pile, there's even more -- regardless of what the chattering classes feel they want the picture to be -- that at least puts an increasingly large question mark on the Probe numbers.

You don't even have to believe the recent poll done for a Manitoba labour organization we relayed, that unfortunately can't be attributed, showing Doer's NDP still pretty comfortable. (If you believe competition is good, it would be good to have a competing public domain poll out there to compare. But, as mentioned, the media outlets today are just too cheap.)

However, the news media wants the story of the Tories creeping up on the NDP, and it appears the Free Press is hell-bent on painting that story and the other media are largely perfectly happy to repeat it, without much thought.

In any event, the other emerging consensus from NDP ranks appears to be that the Free Press/Probe agenda is stirring the NDP troops out of some complacency. This is a good thing going into what is very likely an election period over the next 12 months.