Tuesday, September 16, 2008

Why Polls Matter

Three polls, two of which show a compelling battle for second, as the NDP threaten the Liberals. A third, which just happens to be the most reputable, NANOS, shows no such thing, but I doubt that will matter to our restless media, looking for an angle.

The new EKOS poll, puts the Liberals at another new low:
Con 38%(+3% yesterday)
Libs 23%(-2%)
NDP 19%(no change)
Greens 11%(no change)

Angus Reid also shows the Liberals reaching a new bottom:
39 per cent of decided voters would back the Tories in the Oct. 14 federal ballot, three points above the party's share of the vote in the 2006 electoral process (36.3%).

The Liberal Party is a distant second with 23 per cent, seven points below its 2006 total (30.2%). The New Democratic Party (NDP) is third with 18 per cent, followed by the Greens with 10 per cent, and the Bloc Québécois with nine per cent.


NANOS shows virtually no change since yesterday, the gap moves from 6% to 7%:
Cons 38%(+1%)
Libs 31%(no change)
NDP 17%(no change)
Greens 8%(-1%)

Obviously, I give NANOS more weight, in terms of what I think is really happening, but that might not be the point.

Regardless of the source, we already have a situation where the media is dying to make a story out of "second place". Any poll which supports that thesis is bad news for the Liberals, and it one reason why people who say polls are irrelevant don't understand how they influence coverage. The results might be irrelevant, but the subsequent "run with the ball" mentality of our media certainly isn't. In many respects these things tend to become self-fulfilling prophecies, as the perception of a close race, allows people to promote the situation, in turn gathering more perceived momentum.

We saw last week, how several outlets seized on one poll, amongst many, to suggest the NDP were competitive with the Liberals. Now that we have two fresh offerings, which show the Liberals at historical ABYSMAL numbers, expect a few more negative columns, not to mention Layton telling the world "he can feel something happening out there", his handlers waving the above findings.

23 comments:

Anonymous said...

Why be some glum. Here we are in the midst of a stock market meltdown and looming economic collapse and Dion decides its a good day to trot out BOB RAE of all people. Now I can breathe easy knowing that if the Liberals regain power - Bob Rae will be the leading figure keeping a steady hand at the economic tiller!!

Seriously, i don't mean to gloat - but don't you think that when the headlines are shrieking about economic collapse the Liberals could have an opportunity to remind people of their shrewd economic management when they were in power - and instead Dion decides to showcase Bob Rae of all people. Honestly, who is making the decisions at Liberal HQ???

Steve V said...

anon

Not glum at all, just trying to be fair. There is a danger of this second place narrative taking hold, so I'd rather acknowledge it, than just pretend it doesn't exist.

catherine said...

Here's a different take on this.

Good grief. Everybody relax.
The election is one-fifth over over, a Conservative majority is far from probable and only a fool would discount the formidable Liberal brand from bouncing back to keep the New Democrats at or below 30 seats.


Anyway, not sure what to make of this Harper-Layton-polling company-media generated narrative, but it still seems very contrived to me. I know one's own circle is not a good indicator of anything, but I've only seen movement in the opposite direction (toward the Liberals).

Joseph said...

Hey Steve,

Good post - and I appreciate your giving us a nice update on the tracking each day.

You know my feelings on this already, I think, but to summarize:

For the past decade, I have grown increasingly convinced pollsters are more inclined to create news or a narrative for their clients than they are to strive to accurately capture the feelings of the electorate.

They don't observe as much as they drive. I think it is clear in how they report and analyze their polls, in the questions they ask, and the way they ask them. To allude to an old phrase, they know how their bread gets buttered.

I think with the proliferation of pollsters, the practice is even more widespread.

That is not to say I don't think polls can have value, and that all pollsters partake in an overt exercise. But the more I see polling analysts "writing" the headlines and the narrative, the more I think it is undeniable some will begin to abuse that "power" just as any power will be abused.

Pollsters have always indicated that polling is both an art and a science. Well, anytime "art" becomes the measuring factor, creative license is a given.

Having said all of that, the Liberals are just going to have to deal with this narrative now. It could have been avoided to some degree (or to a great degree) if they had really launched a bold campaign 9 days ago instead of the muted approach that got us to where we are at now.

I am encouraged at the changes they've made. I am even pleased with some of the direction and the forcefulness with which they are now acting. I think that if they keep that up, the campaign is still young enough that they can recover.

But they did bring it on themselves with the soft opening. It is what it is.

All they can do now is work against that as well as their real competitors in this very important election.

The one heads-up I'll give, which if it happens, supports my initial premise, is when some pollsters start doing special polling just on the "race for second place" question. I wouldn't be surprised to see it because, face it, it would be a great way to keep driving that narrative, now wouldn't it?

Demosthenes said...

Joseph: the problem is not so much the polls themselves. There are variations, issues of quantification, survey design, and the adorably naive belief that online surveys are worth anything- but that's a known factor.

No, the problem is when the pollsters are called on to interpret the results. By and large, you're asking them to play amateur sociologist and/or political scientist, when their training has little to do with such things and everything to do with collecting those numbers in the first place.

They can bust out the spin, of course, but that doesn't mean it's worth anything.

As for the "second place" stuff... I suspect that a lot of Liberals will be getting off the fence on this one. Their annoyance at the central campaign be damned; this is an existential crisis for them.

(Warren looms large here—pique over a stupid 'dinosaur' line seems petty in light of the threat—but he's one of multitudes.)

Mike said...

Those Angus numbers are a joke.
In BC: 26% for the Greens(!) and 13% for the Liberals? Sorry in fact every other poll shows the Liberals no lower than 25% in BC.

Not to mention Angus seems to make clear that the Cons are at 39% of DECIDED voters and Libs at 23% of DECIDED (they don't see they factored in learners or anything). So they sampled 1,155 Canadians, for all we know 200 of those were undecided. So that leaves 955 decided voters. So 39% of decided voters would then be 373 people, or 32% of those sampled and the Libs are at 19%, with about 18% undecided if they were reporting properly. It would also mean a larger margin of error, since it doesn't sound like the numbers they are reporting include the full 1155.

Unless I am misunderstanding what they mean by 39% of decided voters, then that's some really shifty reporting in their poll to not mention how many are undecided.

Don't doubt the media will swallow the narrative anyway though, but it's hard to know what to do about polling companines that have really bad reporting methods.

Steve V said...

I'm not saying AR is accurate, just how it can be used to promote an angle, sort of along the lines of what Joseph points out. Decima doesn't have any undecideds either, and given the tertiary interest of many people, I find it strange. NANOS has a full 1/5 undecided, which is normal at this stage of a campaign.

Anonymous said...

Decima may not report on how many people they interview are undecided - but I'm certain that they do have undecided people in their sample.

I tend not to pay much attention to the undecided %. Typically about 35% of Canadians don't vote in federal elections, so in the end the vast majority of people who say they are undecided are actually non-voters.

Anonymous said...

Thru my long lifetime I have worked for many business people trying to feed five kids and I have worked for doctors lawyers ..restaurant...variety..etc...business people are all crooked with numbers.....they manipulate them to change the story....decima does not like the ndp usually having them very low around 13...since this harper push for layton to replace the libs....decima is saying they may be the opposition....AND THE BOUGHT MEDIA IS LAPPING IT UP.

ottlib said...

One problem I have with your thesis Steve is the shear proliferation of polls have probably caused many to tune them out, along with the media narrative that goes along with them.

There have been multiple polls published on almost a weekly basis since the last election. Most of them have been contradictory.

Now that situation has increased to multiple polls on a daily basis since the election call.

I suspect people have had enough and really do not care anymore what the polls say.

Trevor said...

I'm certainly no expert on polling and I tend to vote Conservative so I'm sure you'll take this with a grain of salt. but ..

Nanos is the only polling company that I've seen ask for the top TWO parties that you would support. This seems to me to give weight toward strategic voting by giving a party support levels from voters who have indicated another party as their first choice. This worked well in the last 2 elections because everytime someone said 'Harper majority' all the soft voters freaked out and voted Liberal.

Every other polling firm asks a very straight forward question 'Who would you vote for?'. When people are asked that question the smaller parties show higher support levels and the CPC gets huge numbers because their support isn't as soft as the Liberals.

The thread that all of this hangs on is whether or not people will strategic vote this time. People have seen Harper for the last 2.5 years and he's not nearly as scary as Paul Martin would have you believe. What your real worry should be is that the NDP will catch up enough and Harper will build up enough of a lead where voters will realize that strategic voting for the Liberals won't help and they will vote for their first choice. If that happens the Libs will be fighting for second place.

Steve V said...

anon

A good pollster tries to get likely voters, so I'm not sure you should incorporate those that don't vote into the results.

ottlib

I think you miss the point entirely, the media is fixated on the polls. Go look at the publication websites, or catch a newscast. People might not pay attention, but when a reporter is looking for a story, they are influenced by the polls. Where do you think these "NDP challenging the Libs" stories get their inspiration? Or the "Libs are in trouble" maelstorm of pieces and articles, it's ALL poll driven. If you actually watched Dion last week, he looked better than ever, Mon and Tues very feisty, very passionate, quite clear and articulate. That was lost, because the Cons were surging, the media quickly adopted the "majority" theme, Dion's words LOST. And, just watch, if the Libs do start bouncing back in the polls, the media will start saying Dion has really hit his stride, they will make a rebound the story, all based on what the polls tell them. As far as I'm concerned, we are too the point where polls drive the narrative.

ottlib said...

Most people vote for the party they believe will do the best job at governing.

The NDP is not that party and it never has been. That is just the objective reality and Jack Layton has not changed that.

The same applies to the Greens.

I read an interesting story just before the election which stated that only 37% of voters make their decision about who to vote for before the election and in the first couple of weeks. The remainder make that decision in the last week or two, with around 20 percent making it as they get to the polling station or when they get behind that little cardboard box at the polling station.

So folks, the only people who have made up their mind are people like us. The rest have parked their votes and are still ripe for the picking by all parties.

I personally do not believe the polls putting the Liberals in the low-20s. At this stage in the election such a situation would manifest itself in the Liberal campaign. I do not believe any party has ever come back from being 15 points down in the second week of an election campaign to win.

The Liberals are still running a campaign like they believe they are within striking distance of the Conservatives. Nanos says they are and I am inclined to believe it for the moment.

Anonymous said...

"A good pollster tries to get likely voters, so I'm not sure you should incorporate those that don't vote into the results."

First of all, as far as I know, pollsters in canada do not screen for "likely voters" - they just screen for eligible voters (ie: 97% of the population over 18). About 85% of people will claim that they will vote, but when the dust settles the turnout is rarely over 65%.

The 20% who say they are undecided are mostly apolitical non-voters.

But there are a lot of people who say they will vote one way who may still change their minds. They are the true target.

Steve V said...

"The 20% who say they are undecided are mostly apolitical non-voters."

And, what evidence to do you have to support that?

It's just a ridiculous argument to suggest that somebody who hasn't made up their mind isn't voting. I know tons of people who are engaged that haven't decided.

Hogwash, again...

bigcitylib said...

AR is from several days ago. Why use that one?

Steve V said...

They just released it last night. You mean it's worse now ;)

nuna d. above said...

Putting the AR numbers in the Hill & Knowlton election predictor gives us: CON 160 LIB 56 NDP 40 BQ 51 GRN 0 IND 1.

Steve V said...

This story illustrates my point.

Mike said...

Steve this should cheer you up no?

http://www.theglobeandmail.com/servlet/story/RTGAM.20080916.welectionpoll17/BNStory/politics/home

19 point lead to FIVE points in TWO DAYS. How's that for a narrative :).

Mike said...

Sorry to be clear, this is the infamous swing ridings Globe poll.
The Cons led 45-26 on the Ontario ridinngs two days ago, now it is 38-33. It was 39-37 for the Libs in 2006 so the change is actually within MOE (4.8%).

Even though the poll is somewhat dubious this is what CTV and the Globe run with so it makes a good narrative tonight.

Steve V said...

mike

Watch Duffy not have time for Donolo tomorrow :) One thing he did say today too, something to watch for, the number of people who want something different, a change, is up 10% since the beginning of the campaign.

Dame said...

hey Look ..

http://www.theglobeandmail.com/servlet/story/RTGAM.20080916.welectionpoll17/BNStory/politics/home

marta