Friday, February 15, 2008

Don Martin To Be Sued?

The other day, Scott sent me an email, with a link to John Wright of Ipsos Reid threatening legal action, if anyone were to question their polling:
CanWest has been running polls by Ipsos-Reid, which have consistently been giving the Conservatives a showing five or six points higher than other pollsters. Ipsos pollster John Wright, who says he's been getting heat on this from bloggers, says there is no bias and anyone who makes such a suggestion will be sued. Other pollsters I talked to aren't suggesting any bias. They just find the discrepancy, as pollster Nik Nanos put it, "strange."


Don Martin today on Politics offered up the latest poll from Ipsos for his employer. Should Mr. Martin retain legal counsel?:

"Our latest poll results. Ipsos Reid Conservative36, Liberal 29, which is sort of the way Ipsos-Reid works, they tend to give the Conservatives more support than others."

Don Martin

Strange isn't it? When the guy who works for the company, that commissions the poll, speaks to Ipsos being an OUTLIER, you know you have little credibility. Nobody is suggesting bias, well maybe Martin, but not myself ;) How about consistently BAD polling, is that okay?

UPDATE

The poll is now online:
The poll, conducted by Ipsos-Reid exclusively for Canwest News Service and Global National, said the Conservatives dropped one point to 36 per cent since the last survey two weeks ago, while Grit support remained at 29 per cent. The NDP slipped one point to 13 per cent and the Green party held steady at 10 per cent

I would describe the Ontario numbers as quite STRANGE, the only pollster to show the Tories ahead, while every other outfit has them down considerably:
In vote-rich Ontario, the two leading parties remain in a statistical tie with the Conservatives at 37 per cent while the Liberals are at 34.

Meanwhile, the NDP slipped another point to 14 per cent in the central province while the Green party rose to 13 per cent of decided voters.

Strange.

14 comments:

Oxford County Liberals said...

What's stranger is Don Martin being almost apologetic when talking about these poll numbers.

northwestern_lad said...

i'd comment, but I don't think that Ezra Levant's bloggers union will cover my legal bills ;)

Ti-Guy said...

John Wright, who says he's been getting heat on this from bloggers, says there is no bias and anyone who makes such a suggestion will be sued.

You and your outfit are biased.

...come and get me.

Karen said...

Geez, I referred to that poll tonight and completely dismissed it as bias. I might be next, lol.

Anyone who knows John Wright, knows that he is right leaning. Does that make his polls bias? I guess it does only if he is writing the questions but hey, as we've seen with Harper, there can be an attitude of "give me what I want".

I don't know Wright personally, but I do know people who do, so my comment is not built on sand.

Anonymous said...

Martin was almost embarrassed, he immediately used the "same trend" argument to try and find some relevance to other polls. Those Ontario numbers are a joke frankly.

Militant Dipper said...

Fuck Ipsos I haven't paid attention to their bullshit polls for 2 years.

Anonymous said...

Even Harper got sued for calling a polling company liberal.

They don't mess around.

Greg said...

Ipsos doesn't bother me much. Nanos is the only poll with a track record of any credibility.

Steve V said...

greg

I completely agree. The only part I don't like, an outfit like NANOS only conducts a poll every quarter, whereas Ipsos clearly conducts the most polls. The disporportionate releases tend to give Ipsos more ink, more framing, because we are bombarded with their results, almost weekly. I don't know how many times I've seen outlets just run with an Ipsos offering and take it as indicative. What bothers me is the way people just accept the numbers, without questioning, take the narrative from there, while any counter poll might be weeks away.

liberazzi said...

Even ol Duf didnt go ga ga for once over the Ipsos poll and was giving more credence to the Nanos poll, which is saying that the Libs are basically sitting pretty at the moment. You have a handful of polls stating its within the MOE and then you have Ipsos stating something completely different. Are we not allowed to question that? I believe Nanos usually comes out pretty bang on when you compare polling numbers to election numbers. I am not sure similarily how Ipsos does?

Robert McClelland said...

This poll is nothing new. Every time election speculation heats up the majority of pollsters all start showing that the cons are way ahead of the libs. The only difference this time is that Ipsos jumped the gun. They should have waited a week so that this surge in con support could be pegged to the approval by Canadians for their super terrific happy fun hour budget.

RuralSandi said...

Nanos may be more accurate because he doesn't just poll to mood of the day.

I see that the polling companies are now worried about political interference.....doesn't Harper leave anyone alone?

Anonymous said...

There was one funny nugget buried in this "hallowed and immaculate" poll ; ) . . .

Harper has the edge when it comes to questions on leadership, with 41- per- cent support, versus 25 per cent for Stephane Dion, 18 per cent for Jack Layton and six per cent for Gilles Duceppe.

So, given their (shall we carefully say) "different" and "advanced" polling algorithm, wouldn't that mean the leadership gap has essentially disappeared?

The give-away that there was a brilliant silver lining was the tag line of the poll wasn't "Dion polls lower than many small tortoises on leadership."

Monkey Loves to Fight said...

Ipsos has actually been pretty good during elections, although in between, I agree they are overly generous to the Tories. The earlier poll was not totally out of whack with the others as the Tories always do better when parliament is out of session than when it is in.

I think the discrepancy is more because they are a lot of soft and leaning voters who depending on how you phrase the question will answer differently. Once the election is underway, people's opinions become firmer and so this discrepancy disappears.

I've taken a number of stats courses in university so I sort of understand how this occurs. Another big problem is how to deal with the undecideds which are typically around 30%. Some polls include this, others remove them completely and re-distribute the 70% who are decided. Others then ask a follow up as to who you would vote for right now or who you lean towards.

My guess is that if you remove the undecideds completely, the Tories do better than if you include leaners as the Tories have the strongest base in terms of committed supporters, but less room for growth than the Liberals.