Thursday, August 27, 2009

Bizarre Polling

What, you mean we have another poll that doesn't confirm that Ipsos sample that has everyone "buzzing"?. Shocking:
Conservatives – 32.6 (-0.2)
Liberals – 30.9 (+0.7)
NDP – 15.7 (-1.4)
Green – 11.3 (+0.3)
BQ – 9.5 (+0.8) (37.1) (+2.1)
Undecided – 15.1 (-0.7)

Funny, that EKOS actually shows a slight narrowing, a statistical tie.


Anonymous said...

Darrel Bricker to the Senate!!

CanadianSense said...

Are you comparing different designs from different Research Polling Companies to produce a similar result taken on different dates?

Are these companies are getting the desired answer they are seeking.

Who is the sponsor for these designed Polls?

Who benefits from this Pollster design?

Steve V said...


You don't make sense.

Gayle said...

Sense makes perfect sense.

Any poll that is good for the conservatives is a neutral and objective poll. Any poll that is bad for the conservatives is horribly biased.

What is so hard about that?

CanadianSense said...

I will try to make it easier, if I was unclear. Try to compare the Ispos to the last Ipsos (same design).

Before being so quick to criticize and pass judgement please examine below.

Let's Compare 2 EKOS Polls

same design, rolling dates?

Aug 12-18,2009

Aug 19-25,2009

Take a look at the cities

12-18 19-25

Vancouver 28.6% 41.6% 33.1% 30.4%
Calgary 65.8% 18.2% 69.5% 15.0% Toronto 34.2% 39.9% 32.8% 39.7%
Ottawa 44.0% 39.1% 42.3% 43.8%
Montreal 13.1% 32.0% 14.4% 28.0%

So taking a look at the significant drop in Toronto, Montreal, Vancouver captured by EKOS "designed" Poll-give you pause?

Does Ispos 'outlier' have any more validity in your eyes or is it a BS Poll still?

How much did the CPC gain in POP and the Liberals lose in those traditional Liberal strongholds?

Before you panic, it will only have made a difference in close races etc.

CanadianSense said...

Gayle I have no interest in engaging in partisan debate with you. You are free to believe and condemn any negative news that does not conform to your twisted reality.

I am also free to engage and debate with Steve V. on polling bias and design.

I have tried to explain comparing Polls from different companies with different dates and designs is difficult and unreliable.

The cleanest way is to only compare the same 'design' by each company and look for trends

Compare EKOS to EKOS, NIK to NIK etc.

If they are the same design, executed in the same manner that is your TREND from that DESIGN.

Any polling company can design a Poll for a preferred option.

Who was the most accurate in the last election?

Does that mean they will be again?


Gayle said...

Sense - if you were not interested in a partisan debate, perhaps you should not have suggested that the sponsor of the polls have anything to do with the results.

And I do suggest you explain yourself better. I see no significant drop in Toronto from the last poll to this one.

CanadianSense said...

Steve V.

Here is my breakdown from 2 EKOS Polls in Metropolitan Areas comparing the Aug 12-18 to Aug 19-25,2009.

Van 4.5% -13.3%
Cal 3.7% -3.2%
To -1.4% -0.2%
Ott -1.7% 4.7%
Mon 1.3% -4.0%

I was trying to explain how the criticism for Ispos Reid may be unfair. Regardless of the picture.

Each polling company can DESIGN a Poll to give the preferred outcome.

I have shown using the last two EKOS Polls a large drop took place with the same DESIGN. (That is more important)

I have not attacked the validity of any polling company and do not use them for SEAT projection for a campaign and election that has not official begun.

I am also not interested in talking to you based on reading your "perception of reality" posting pattern.

Barney Frank was correct about people like you. Each party has furniture.

I applaud the Blog owner who is spending time sharing his ideas. I may not agree with all his views but I can appreciate them.

CanadianSense said...

Steve V,

My apologies regarding my last post and not being more specific in my comments.

The Barney Frank comment is meant for Gayle whom I believe reflects the fringe element.

Each party has their furniture.

I admire your effort on behalf of your party.


Steve V said...

Yes, let's compare poll to poll, without this mind boggling, very confusing, sort of batshit crazy presentation you've given. Ekos is pretty consistent poll to poll. Ekos is also consistent with every other pollster, using their own methods, with one exception- IPSOS. Ipsos has a very poor record, and has shown time and again to overstate Con votes, understate the Libs and NDP. That is based on election results, not some partisan spin.

I'm quite fair with the polls. As a matter of fact, I've lauded the online Angus Reid poll, whenever everyone attacked it. I did this because their track record was quite impressive. I did this, even though their numbers weren't exactly Liberal friendly.

I am not being unfair to Ipsos. What is unfair is silly columnists and rabid partisans seizing on a clear OUTLIER (which has since been proven with two polls after, not to mention the one the day before) as evidence of a trend, which doesn't exist. An objective attempt waits to see if a trend develops, wants more data. An objective attempt puts a strong critical eye on a poll that is so out of whack with everything else we've seen.

BTW, I'm sure Gayle will be thrilled with the "fringe" tag.

And, you'll note I deleted the comment calling you a "retard".

Steve V said...

Oh, and btw, apart from Vancouver (where the margin of error is large), all the other examples show little statistical change poll to poll. You've actually completely and utterly contradicted your point with the data you've presented. Do you even know that? The Toronto Liberal vote changed a whopping .2% in one week, Cons 1.4%. That's actually very credible replication poll to poll, way, way below the margin of error. Are you for real? Don't answer that :)

Anonymous said...


You should at least admit that your first comment today was so vague that any interpretation could be made from reading it.

Having seen Gayle comment on any number of issues, your remarks are unwarranted. Her instinct that your first comment was partisan in nature was as good an interpretation as any. Perhaps the response was just a lack of patience at yet another person obfuscating to defend Ipsos Reid's pattern of bringing just the right poll to the forefront at just the appropriate time to fit snugly with whatever the latest Conservative talking points happen to be.

I've observed the pattern for a couple of years now. It's suspiciously uncanny how their findings and analysis fit the conservative's talking points without fail.

If your intent wasn't to dismiss that criticism by introducing some other target, you could easily have clarified your first nebulous remarks instead of casting daggers at someone who saw those remarks in an unfavorable light.

If a polling company can shape the poll to create the results they want at any given time, then why would that prevent them from "tweaking" the process and results over time.

You seem to assume they scrupulously follow the same flawed pattern, as if that is some pure inviolable truth, even as they nefariously shape the rest of their analysis to achieve skewed results.

If you're implying the intent of the polling company is to twist the outcome, then I wouldn't suggest you put so much faith in the consistency of polls from the same organization.

Gayle said...

Steve and Joseph, thanks.

I still wonder why Mr./Ms Sense was asking about who sponsored the polls if s/he is not implying that polls are biased.

In any event, as Steve points out, there is very little difference in the two polling results from week to week, aside from Vancouver. I notice that rather than answer my question about that Mr/Ms Sense opted to launch a personal attack.

PS - I am not a member of the LPC, nor do I vote for them. Perhaps people should not be so quick to jump to conclusions.

CanadianSense said...

Steve V.

I don't rely on public polling data. I suggested it would be better if you only compared each Designed Poll with itself over a period of Time for a pattern.

I asked if the drop noted in EKOS in Cities may have been picked up by ISPOS poll.

A research company is capable of designing Polls, do we agree?

The execution affects also affects the Poll. (Time of day, length etc)

Repeat the same with each company.

I have asked questions, I am not implying I know the answers.

I am merely pointing out the CBC pay for their Polls, as does CTV, and Global.

I am not going to pretend Research or Polling can not be manipulated to create "designer Polls". I am also NOT suggesting any of it has been.

I simply stated it is possible. If a TV network, Drug Company, Oil Industry etc.. (customer) orders a "designer poll" their preferred opinion can be presented in a favourable light.

I may have not been able to present the significant drop in Montreal or Vancouver with this format. I am not really interested or consumed over a few Public Polls and I have spent too much time trying to explain the 'outlier' Poll.


Steve V said...


And the Vancouver total has a massive 12.1 margin of error. Only a fool points to that as indicative of anything.

Steve V said...


Just for the record, Montreal doesn't support your "significant drop", it's marginal at best.


CanadianSense said...


The post may not have been clear. You have your opinion regarding Gayle. I don't share it.

I am NOT defending the 'outlier' by presenting or using other Polls showing a drop in Lib support in Metropolitan Centres. I did a quick scan of his "Bizarre Polling" post and did a hasty post apparently.

I suggested Steve use the same Designer Poll to look for trends more accurately.

I was not aware of the drops in the Metro until his posting. I did a quick scan and noticed in EKOS a big drop in Van/Montreal/Cal in %.

I also remember skimming CROP articles a 5% drop in Quebec for the Liberals, CPC and NDP are up?

I made a generalization that a pattern may have found that the 'outlier' Poll may have found some evidence of a drop in support.

I have NO media or preferred access to any of the Polls regarding their Design and execution.

I think the last Ispos was a few months old and Crop's as well making comparison just a very BIG stretch.

Joesph if you are sure of those "Biases" than you should discount those Polls accordingly.

Steve V said...

"I did a quick scan and noticed in EKOS a big drop in Van/Montreal/Cal in %."

For the last time, until you develop the simpliest understanding of margin of error, I would just stop talking. There was no big drop in Cal or Mon, you will see that fluctuation almost everytime. Only a fool develops a thesis using this "evidence".

There's another 30 seconds I'll never get back...

Patrick said...

Heh. If that Ipsos poll were accurate, it sure would have been a very sudden change.

CanadianSense said...

Steve V.

Your post reflects your hostility towards the media reporting on an outlier Ispos Poll.

Crop,Ispos,Ekos Poll pattern show a drop in support in Quebec?


Ispos/Ekos show a drop in BC?


The majority of the Polls show the Liberals up from their low% in October 2008.


The majority of the Polls show the CPC below the October 2008 numbers.


I am aware the Regional numbers have a very large MOE.

You seem hung up on the 11% outlier Poll and the modest movements of Polls in general.

I understand you are happy the Liberals are polling better than October 2008.

Campaigns matter and regardless of the designer Poll flogged by the media.

Each party will have to deliver the voters.

I wish you and your party well.


Anonymous said...

Off topic, but of interest from the Globe and Mail;

Having built up employment insurance reform as the single issue that avoided a summer election, Liberals now say it's not an issue worthy of triggering a vote in the fall.

Liberal Senator David Smith, a co-chairman of the national Liberal campaign, said even if his party's talks with the government fail to produce an agreement, that alone will not be enough to warrant an election.

That still leaves open the possibility of a no-confidence vote on other issues, such as the state of the economy, but the senior Liberal is clearly throwing cold water on election speculation as MPs prepare for a caucus retreat in Sudbury.

“To the extent that at some point [the government will fall], I do not believe it will be on unemployment insurance,” Mr. Smith said. “It's an important issue and it's an issue we want to address ... but I don't see that as the defining issue that would trigger an election.”

Mr. Smith said his comments do not rule out a fall election however he went on to say there have already been plenty of elections recently.