Monday, June 1, 2009

The CBC: How To Turn Bad News For The Liberal's Into Good News..

CBC Headline : Minority government possible for Liberals, poll suggests
EKOS survey indicates Ignatieff's party would clip Tories in election tomorrow


If your a Liberal that headline would be good news, right? Wrong. Technically the poll is right in that the numbers could indeed lead to a Liberal minority government. Aside from that, this article has been spun more than cotton candy in that machine at a carnival.

Take this line for example: "News of the federal deficit topping $50 billion appears to have put the Liberals slightly ahead of the Conservatives in a poll asking Canadians how they would vote if an election were held tomorrow."

Reading that sure gives the impression that the new deficit figure has put the Liberal Party ahead in the polls, right?

Then this line: "Up until the deficit news of May 26, the Conservatives appeared to have been benefiting from the fact some more prosperous Canadians, perhaps with an eye more to the stock market than the job market, were becoming more optimistic about the economic future.

Reading tha sure makes one believe that Conservative numbers are dropping, right?

And this line: "However, their advantage appeared to vanish after Finance Minister Jim Flaherty revealed Ottawa expected a budgetary shortfall of $50 billion for the 2009-10 fiscal year that will end in March 2010"

That sure makes it sound like Conservative's had an advantage in the polls, but the recently released deficit figure has caused that advantage to disappear, right.

The fact is all these statements are wrong, in fact they are wrong based on the same CBC and Ekos polling companies information from the April poll. The numbers from April to May show a drop of -3.7% for the Liberal's, while the Tories gained +1.8%

H/T to Wilson who left this comment on Joanne's blog:

"p.s. What the CBC article doesn’t state is that Cons have GAINED support since the last poll done April 8-15,THEN Michael Ignatieff’s Liberals enjoying an upsurge, says a new poll from EKOS released exclusively to CBC News.

April…. 36.7 Libs ,
30.2 Cons
May …. 33 Libs,
32 Cons

One would have thought today's headline might read Harper enjoying an upsurge, says a new poll released exclusively to CBC news. Funny That...

22 comments:

Roy Eappen said...

Neverending msm bias in favour of the grits. people at the cbc must realize that most conservatives want them sold off ASAP.

Ted Betts said...

All partisanship aside and my reluctance to trust polls aside, this is a very fascinating poll.

First of all, it is massive. With over 10,000 polled it is the largest poll by far in Canadian history (aside from an actual election or referendum, of course!).

Second, it took place over several weeks instead of a couple of days. If you read the analysis, this why they can conclude that there was an uptick for the Liberals after the deficit was announced to be at minimum 50% greater than Harper was saying it was going to be only a month ago. The analysis indicates that they did see the Conservative slide halted after the ads - i.e. they were having some clear impact - but then at the end of the polling period when the budget numbers were revealed there was a big uptick in Liberal support. With the poll lasting over 3 weeks, it mutes out the immediate reaction to new information: meaning we don't know if the last uptick will last or not, but you have a much better sense of what support is more solid and what support is up and down with different news.

Third, the regional breakdown analysis is itself fascinating. Us consumers of information tend to focus only on the total picture which has the Liberals and Conservatives neck and neck, but the regional breakdowns of the polls have such small population samplings adn such large margins of error that they are highly unreliable. In the EKOS poll, however, the samplings in each region are themselves very big and so much more accurate/reliable.

What is interesting about that is what that means as far as seat projections are concerned. Maclean's has one here. It shows a week but clear Liberal minority given the way the support breaks in each province.

Fourth, their methodology makes this a much more reliable analysis of what Canadians are actually thinking. They did not use online polling. Remarkably, they for the first time included cell phone users. Almost one quarter of Canadian households only have cell phones; these households have different socioeconomic and voting patterns than traditional landline users; so this is a significant break from other polls.

There is still good reason to hesitate in saying this is fully reliable because, for example, it is not an election period and there is no breakdown within the provinces.

But still, wow. For a political junky, this is a wealth of much more detailed and reliable data, regardless of which way this is spun by any one side.

(Of course, it is certainly nice that this information is all good for the Liberals, but I'm trying hard not to have a partisan comment here. ;-) )

Anonymous said...

What troubles me about this poll is that 50% of respondants other than no comment or refused were from Quebec.

I may be wrong but I think Quebec only represents less than 30% of Canada's population.

This survey has not been peer reviewed and was commissioned by a partisan entity.

NOBODY who understands statistics or public analysis would ever ever give this survey any bonafide credence for reliable statistical data.

But, it does give the media some nice headlines and the general public can be easily fooled by inaccurate data.

Expert researchers can skew any question and methodology of data collection to get the results they want.

That is why a peer review is required at all stages before a seal can be applied declaring the data reliable.

This one does not pass the test. Sorry - no matter what side of the political spectrum you are - it is no more reliable than reading horoscopes.

paulsstuff said...

Nice analysis Ted. I agree with much of it. I just don't give polls between elections a whole lot of merit. We Canadian's by our nature are very wishy-washy.

It's like the polls in the days of the coalition showing Conservative support in the high 40's. Those numbers would never had held in an election.

Looking at past polling between elections, Paul Martin should still be PM, Stephane Dion should still be leader of the Liberal Party, and Stephen Harper should now be running a majority government.

There are too many possibilites of game-chngers in a 36 day campaign.

Ted Betts said...

Oh, I agree Paul. I am always pretty wary about polls. Especially individual polls; trendlines among many polls can tell you something.

I think though that polling can tell you something, lots. The problem is that there are so many poor polls out there and so much weak guessing that counts as analysis. PLUS, a most important plus, too much assumption on the importance of a snapshot in time.

But like I said, polling can tell you things, provide you with valuable information. When it gets this large a sampling and this detailed, it can tell you quite a bit.

The primary take-away from this that is good for Liberals is not that they are still in a minority government position despite hundreds and hundreds of thousands of dollars spent on the negative attack ad. What is good for the Liberals is that their support numbers appear to be less volatile than in the past. You are right to say we are wishy washy voters, but there is a clear sense in this poll that Canadians are looking seriously at the Liberals as a serious option for governing and are not looking away as soon as the Conservatives try to re-direct their attention.

There is also a lot you could say good or bad about the Conservatives in this poll that would be pretty meaningless. But what I think you can say more certainly that is good for the Cons out of this poll is that Canadians outside of Quebec are not done/fed up/angry with the Conservatives - despite the economy, three years of governing, three years of flip flops and broken promises and abandoned principles, mismanagement of our finances, the backlash against the ad campaign, etc.

When Martin ran the 2004 election, Canadians were fed up with the Liberals but not convinced about Harper. By 2006, they were even more fed up with the Liberals and so it would not have mattered how bad Harper was (he wasn't - he ran a good campaign) or how good the Martin was (he wasn't - he ran an awful campaign). The media gave up on him. Canadians gave up and weren't even listening to him.

This poll shows the Conservatives, although clearly on a decline, are far from that point.

Anonymous said...

I checked the seat projections and there is no way the Liberals would get 11 seats in BC. They will be lucky to hold on to the 2 they won by less than 50 votes so if anything they will lose seats in BC.

Remember that BC just re-elected their conservative government.

Ted Betts said...

Anonymous 11:29,

In your zeal to attack and dismiss a poll that is overall more positive for the Liberals than the Conservatives, I think you must not have even bothered to read it. I think you are referring to the CTV poll that surveyed 1000 Canadians and 500 Quebecers, not the EKOS poll that surveyed over 10,000 Canadians.

You also have it wrong. That they interviewed more Quebecers just means the Quebec numbers are more accurate, not that the overall numbers are wrong. This is because they weight the regions by their populations. So if you interviewed 1000 voters in Alberta and 1,000,000 in Ontario and every single Quebecer, the 1M Ontarians still account for only 38% of the overall poll numbers, the Quebecers surveyed would still only account for 23% of the overall poll numbers, etc.

The CTV poll is worth dismissing not for this reason, but because they only interviewed 1000 which is very small and it included online polling.

Ted Betts said...

Anonymous 11:46,

Seat projections is a very unreliable game to play, regardless of how fun it can be.

As a precise tool, it is pretty useless outside of an election and even then only if your size samples are large enough in the region.

Still they can tell you things based on history and that is how they work. Historically, if a party has at least X%, they will get at least Y number of seats. It was explained to me once by a Harris Tory riding campaign chair during the 2003 election and why they knew internally they were toast even when the polls were showing them fairly close to the Liberals.

But it is more of a rule of thumb than a science.

paulsstuff said...

I looked at the regional breakdowns Ted, and I think the Quebec numbers are a little on the generous side for the Conservative's as are the B.C. numbers for the Liberal's.

I think the most accurate seat prediction I saw last election was still off by 27 seats. Given the tightness in the polls that makes them pretty much useless.

As a Conservative supporter I am quite pleased with these numbers. sitting governments around the globe would look at those polling numbers with envy, considering the beating some are taking.

I think by mid-summer the ecomonomic picture will be quite a bit more rosy than today, and that will correspond to Conservative support rising.

Anonymous said...

wah wah wah the MSM hate Conservtives.

Please. Quit your bellyaching. It's pathetic.

revanche said...

Ah yes, another poll to commissioned to fill another 24 hour news cycle.

The thing to remember about polls is 50% of respondents will stay home on election day.

The next election will be determined on the ground. The party with the best ground game will win. A simple polls can never account for that.

The lack of a solid organization in Quebec is the only reason Harper doesn't have a majority right now.

paulsstuff said...

"The thing to remember about polls is 50% of respondents will stay home on election day."

Excellant point. Last election Conservative's did a great job getting their voters into the booths on election day. and they learned quite a bit from that to do even better next election.

Ted Betts said...

Paul:

I hope the economy returns, even though I agree with you that that will mean a recovery of support of sorts for the Conservatives.

Not sure what line of work you are in, but I deal every day with business leaders who are trying to figure out the economy (not just the markets, but the economy) as it relates to a myriad of different industries. And they have good money - and their own! - riding on it which, as a capitalist, I believe is a better predictor than economic pundits and certainly better than self-serving politicians.

The conclusion is pretty consistent: we are just starting to see the depths of the economy and it will take another 9 months before we see real recovery.

Also, there is a fairly consistent conclusion that markets may go up, not because the economy is getting better but because the next 6 months of bad economy have already been factored into their analysis.

The markets will also go up because troubled companies have gotten so bad that they have been forced to restructure and the market is responding to announced restructuring plans. The economy won't start to truly recover though until several months after those restructuring plans are fully implemented.

In between, and as part of those restructuring plans (like GM) are hundreds of thousands of jobs lost, which means less cash for consumer purchases, which supports the view of a recovery 9-10 months from now at the earliest.

paulsstuff said...

I've been in the auto sector for over 30 years Ted, and this is the first time I've ever seen them miss so far off with maket projections. Normally they can predict sales 1-2 years ahead of time. Now they can't predict next week. Really quite amazing.

In reality the sales of vehicles was unsusatinable, hitting roughly 17 million yearly in North America. 12 million is a more realistic number.

As for the Conservative's, I hope they can improve on those Quebec numbers, but if not, I have no problem cheering for the Liberal's to kick those Bloc asses out of Parliament.

By the way, I can't remember where, but someone poted that a number of Bloc MP's need another 13 months of service to qualify for an MP pension. My guess is they are more worried about that pension than their constituents, and won't be to quick to force an election.

Joanne (True Blue) said...

Who says Liberals and Conservatives can't get along? ;)

paulsstuff said...

I think Ted is a decent enough guy Joanne. We just differ on politics. If I'm not mistaken Ted was a Mike Harris supporter (I'm sure Ted will quickly correct me if I'm wrong). I love debating with people of differing views, whether it be politics, economy, sports, whatever.

Ted and others come hear voicing their opinions in a civil way is what blogs are all about. Joanne is a perfect example of that. Joanne's blog (my fave), welcomes different opinions. Unfortuneatly from time to time th trolls roll in, not to debate, but to prevent debate. And Joanne knows better than anybody when that line has been crossed.

revanche said...

Ted, I could have a beer with you.

I can't say the same about some of the knee-jerk idealogues in my own party right now though...

Anonymous said...

Here are a few little tricks used to skew poll results.
- If you want a more socialist entitlement social services dependent population you get a larger percentage of your sample during the day on weekdays.
You will get proportionately more retirees, welfare recipients, unemployed, stay at home Moms, etc.
These demographics are more inclinded to the nanny state government should give you everything politics offered by the NDP and Liberals.
It is a favourite ploy of pollsters.
They also have an easier time recruiting polling phone workers during weekdays so it helps with their staffing.

Anonymous said...

They had to poll 50% in Quebec in order to get the results they wanted - just edging the liberals a bit ahead of the Tories in numbers. That is also why they polled so many - they realized by polling the normal amount of people - the Tories were ahead so they polled more in Quebec where they knew they would get a greater number for their liberal party.... Canadians are not that stupid - but note - they still pick PM Harper over the visiting Iggy - even though the media is pushing him in our face all the time...

wilson said...

'Harper preferred as PM choice'

and that will make the difference in an election.

Funny how CBC polled and polled but Cons were ahead,
then they polled and polled some more,
and on a bad news day...Cons and Libs were tied.
Voila, we get the poll.

Torstar was smarter, they just polled for the 2 days following the bad news $50B deficit....Cons and Libs tied.

tao_taier said...

Polls are a joke no matter what the result...

You would have to break them up by riding in-this-country since that is how votes are tolled come an election ballot, for the "Local" Party Candidate Representative.

And repeat for a year if polled weekly for a general trend or monthly if by the day.

Poll 100 (or 6-11% of last voter turn out) different people each day for a month then tally up the numbers and start all over again for the next... same riding over again for a year.
Then match up the trend line and start all over again.

That should start a polling company off to a more accurate read of each riding.

then maybe polls would start to reflect something closer to reality... or at least further from the current delusions.

tao_taier said...

Forgot to add my main point, is that when Ontario first started shedding jobs the income revenue loss also hits the federal budget negatively...

In other words, when Ontario leads in job losses by vast numbers even before the global recession hit, while other provinces were booming as they were, you know then that this is definitely NOT a Federal or "Conservative Made recession".