Friday, April 29, 2005

Bloggers Hot on the Trail of the Polling Firm that Released the Latest Liberal Friendly Poll

A new poll by CTV and The Globe and Mail was released today. Already the poll’s results and the firm that conducted it have come under question. Brent Colbert has found three really good articles on the polling firm that conducted the poll, the GPC.

The first article is from Conservative Life. This article details the dubious history of GPC in Canada. One of the highlights was a poll commissioned by the University of Western Ontario which was so deeply flawed that its results were boycotted.

Next, the fine people at Occam’s Carbuncle ran the name of this polling firm through the Elections Canada database. It turns out they have been a big, long term supporter of the liberal party of Canada. This should come as no surprise.

Another article was written today in the Cannuckistan Chronicles. This article links to a Commons paper that details how the GPC polling firm provided stats for the botched gun registry. It also draws a link to this polling firm and many liberal friendly supporters including former premiers, cabinet ministers, senior bureaucrats and political advisors. It also includes a list of big names in liberal politics associated with the GPC and it includes work the GPC has done in other areas of international political activism. The article provides links to back up all its claims and provides evidence that it received money from the public works department, the same department that oversaw the scandal plagued sponsorship program.

Angry in the Great White North adds more research to the GPC poll with a breakdown of political contributions to the liberal party by senior members of the GPC polling firm. The breakdown of their contributions since 2003, when Paul Martin came to power, shows 100% of contributions going to the liberals. Furthermore, Otto Lang (senior counsel for GPC) was named co-chair of the Manitoba chapter of the National Liberal Campaign Committee.

This poll is not yet twelve hours old and already it is putrefying.


At 9:48 p.m., Blogger Neville72 said...

Ask for the pollster to release their internals and raw data. Those always reveal whether or not the poll is biased or not. Dubious polls here in the US are invariably based on an oversampling of one political group or another hence the pollster gets the result he or she is looking for.

If the pollster won't release the raw sampling data then you can assume with about 99% certainty that it is bogus and should be ignored.

At 11:38 p.m., Blogger Raging Ranter said...

The polls were released to give the ailing Liberals a moral boost. Volunteers and MPs are far less likely to put a lot of enthusiasm into a campaign if they think that everyone hates them, and that the numbers can only get worse. If the polls show them getting a boost, it might give them a shot in the arm. That's what these polls are; biased crap designed to boost sagging Liberal moral.

At 8:18 a.m., Blogger angry_in_t_o said...

Another tidbit -- the co-chair of the National Liberal Campaign Committee (announced in March by Paul Martin) is Otto Lang, a senior staff member at GPC, the polling firm.

Full story:

At 8:22 a.m., Anonymous Anonymous said...

The one poll, which CBC displayed on Thursday night, was done by a group headed up by Randy Gregg (I believe his first name is Randy but having a bit of a mental lapse-they spoke to him that night about the poll). If you've ever seen him on CBC, as he often provides political comment, there is no doubt which side of the fence he sits on. I am convinced one cannot rely on the first of the two polls and I'd be very skepticle with any poll that has the name Globe and Mail attached to it (not to mention CTV).


At 10:37 a.m., Blogger Neville72 said...

A recent CBS poll here in the US that was found to a be bogus use the "over/under sampling" method to produce the CBS pollster's desired results on the fillibuster debate raging in the Senate.

Recent questionaires of the voting public have shown that among US voters 37% identify themselves as Democrats, 37% identify themselves as Republicans and the rest as Independents or members of minor parties like the Greens or Libertarians.

In this case the poll results were a direct reflection of the pollster oversampling Democrats by 2% points, 39% in this poll and undersampling Republicans by 9%, 28% in this poll. By doing this any pollster can achieve most any result he/she wishes.

I would bet that's what happened in the recent Canadian poll as well.

At 11:07 a.m., Blogger an MP said...

Neville: Did they purposefully undersample the GOP, or did it just come out that way in the random selection?

If you follow proper techniques to produce a random sample, you shouldn't need to choose to overrepresent a particular group. Moreover, being selective as such or correcting for "discrepancies" in the sample is also tampering with the result.

At 12:15 p.m., Anonymous Pikey78 said...

Sounds like some Tories are worried that their 15 minutes has faded! Regardless of whether the polls are Liberal friendly or not, who cares? The only poll that matters is on Election Day. Until then everyone is releasing numbers that will benefit only their particular cause, including the Conservatives. But it doesn't matter in the end, because as much as Canadians don't trust the Liberals, they trust the right wing, religion preaching Conservatives even less!

At 12:22 p.m., Blogger Neville72 said...

Whether or not the CBS pollster purposely manipulated the sample is irrelevant to this discussion. His sample that produced a population of 39% Democrats and 28% Republicans was an invalid description of the voting population as a whole and skewed by 11%.

We saw another perfect example of this in the early exit polls taken as voters left the polling places during the 2004 US presidential election when early results showed John Kerry running away with the vote and Democrats cluelessly elated. Later when the raw data of the exit polls were analyzed it was found that the people asking the questions outside of the polls had queried a disproportionate number of women and Democrats thus giving the news media and gleeful Democrats false hope.

Were they doing this purposely to effect the vote in the west where the polls hadn't yet closed. Who knows? All I know is that pollsters are fully capable of controlling the results by simply manipulating the sample population.

Here in the US in last couple of
years it's become standard practice for pollsters to be expected to release their raw data when the polls are released. If they don't they are immediately under suspicion. That's how the bogus CBS poll was quickly eviscerated by US bloggers and dismissed as badly flawed and not worthy of attention.

If Canadians would insist on seeing the same information at the time any and every poll is released(and maybe you already do)then at least you'll know which polls are serious and which polls are not.

At 12:40 p.m., Blogger an MP said...

There is a difference between a quantitative, random sample as done by a polling firm and Exit polls, which rely on using available subjects. You can't rely on the latter, as you are inevitably getting a bad sample.Exit polls do little more than give newsfolk something to talk about while the results roll in.

However, its really dangerous to "correct" a random sample or push to make sure that a certain party is always adequately represented, as you always have to base those numbers on other surveys. It is perfectly acceptable within quantitative research to just use a random sample, and not correct for it.

At 12:59 p.m., Blogger Neville72 said...

Your points are well taken however a truly reliable and valid random sample should not overrepresent one group or another by 11% in this case. If it does it fails to accurately describe the population it seeks to poll. Moreover, if you can't see the raw data how do you know whether or not the sampling was truly random. You don't.

Look, I don't put a huge importance on poll results. As was said before, the only poll that really matters is the one taken on election day.

That said, however, as we've seen in recent headlines in Canadian newspapers, polls can make headlines as they did for 1 day in the US in the case of the aforementioned bogus CBS poll before it was discredited.

If it's insisted that poll data be released along with the final tally you'll have much less of these shenanigans going on. What amazes me the most about the CBS poll is that they still haven't learned their lesson. They still, even after the Dan Rather fiasco, think they can get away with these kinds of hijinks. Sadly for them, the bloggers are on to 'em.

At 1:31 p.m., Anonymous Anonymous said...

Wow... the MSM is running hard with the poll numbers...

The CPCs are getting killed by the questions framed by Liberal polling firms (e.g. Does Stephen Harper have a hidden agenda?)

They need to go on the offensive with their own polls. Apparently, idiot Canadians believe polls... even rigged ones.

How about asking Quebeckers: "Would you be more likely to vote to separate if Ontario elects a majority of Liberal MPs again in the next election?"

It's time to fight fire with fire.

At 2:29 p.m., Anonymous Anonymous said...

Polls like these pander to our 'diversity'.

Too much diversity is killing the 'right' in Canada. Far too many Canadians are willing to forgive Adscam, for the sake of advancing their political agenda.

For's all about ME and my feelings, and 'checkmating' the paternalistic 'right'.

Some kids never grow up.

At 5:10 p.m., Blogger an MP said...

I agree that more information provided is better than less, and think that it is probably better that as much data as possible be made available.

As for the framing of the questions, it's important to look at the order that questions are asked. If the hidden agenda question appears before those measuring whu peopel would vote for, that is a problem. If it appears after then you aren't tainting your sample.

Regardless, that question doesn't qualify as push-polling. If it asked "Do you think that Stephen Harper has a hidden agenda to ban abortion and ban homosexuality?", you'd probably have a case. That's as close to a value-neutral term as you can get for an issue that is coming up daily in the polls, and it's the one that gave the Liberals the win last time.

At 6:01 p.m., Anonymous 905 Tory said...

Pikey78 said:
"religion preaching conservatives."

Take a hike buddy. The Liberal party treats its own agenda as a "religion" to be believed and supported by all. Political power is the Liberal god.

Besides, this kind of rhetoric is useless. Deal with the issues -- adscam, healthcare, tax relief. The fear and smear crap has to go.

At 9:10 p.m., Blogger Bill said...

Here's a variation on the theory for you. These skewed polls are actually a conservative plot to lull Liberal voters into a false sense of security. These people are the same ones who don't want an election, so they will just stay at home, secure in their belief that the Liberals will win again anyway.

On the contrary, conservative minded people really want the election, will turn out in record numbers and Stephen Harper will be the next Prime Minister...

At 10:45 p.m., Blogger Hamm172 said..., has an informal poll on their site that has had over 22,000 respondants. Over 58% are in favour of an election to bring down the government.

At 7:49 p.m., Anonymous Anonymous said...

Whoever checked the one polling company's affiliation in the Elections Canada database should do it for all firms and realease the results to the Sun chain of papers, the National Post, Western Standard and all known small-c talk show hosts. This has the makings of another scandal, call it "Pollinggate."


Post a Comment

<< Home