Wednesday, September 29, 2010

Angus Reid and Reality: Are they actually connected?

I may or may not have said it before here, but I've got a big issue with Angus Reid a good portion of the time. The numbers always seem a little strange, always seem to buck a trend, and always give a better result to the token third party at the expense of the second party, particularly if they're progressive or left-of-centre parties.

You can put this down to their method, to incompetence, or to plain bias. Either way, if you look at Angus Reid in the long-term, their numbers simply don't seem to fit the mold. The most recent Canadian and Ontario polls demonstrate this perfectly.

Here are the numbers for the federal voter intention poll:

Con - 34%
Lib - 26%
NDP - 18%
Green - 11%
Bloc - 10%

If you look at the regional data, you can see that except for Alberta (17%) and the Atlantic provinces (48%), the Liberals are more or less in line with their 2008 results. So, 26% makes sense in these results.

That's all well and fine, but this is certainly a poll that bucks the trend, as all of the most recent polls tend to disagree with the static position of the Liberals, including Harris-Decima, Ipsos-Reid, Ekos, and Environics, all pollsters who had polls out in September, the first two polling in the latter weeks of this month. Indeed, the last two polls have shown a big drop for the NDP into the low teens, and while Angus did reflect a drop for the NDP nationally (1 point to 18%), they also reflected a low result for them in the Atlantic provinces (12%), lower than what they had back in August (15%) - while other polls have shown the NDP regaining ground in the Atlantic. The make-up for the NDP comes in Quebec, where they've managed 17% in this AR poll - something that is actually reflected in some other polls, but definitely not all.

But, I digress; Angus Reid has done this many-a-time before, and not just in Canada, where the third party seems to be extraordinarily powerful as compared to the second party's position, while the leading party sits perfectly well. For example, even before the Prime Ministerial debates in the UK shot Nick Clegg into the sky, Angus Reid consistently had Gordon Brown's Labour at their lowest scores, and Nick Clegg at the high end of the spectrum - while the rest of pollsters would have different numbers, with Labour and the Conservatives closer.

Case and point: on the 6th-7th of April polling done by AR, the Conservatives were on 37%, consistent with other pollsters; Labour on 27%, and the Lib Dems on 22%.

Three other pollsters from the same time peroid:
YouGov (C - 37%, L - 32%, LD - 19%);
ICM (C - 38%, L - 30%, LD - 21%);
Populus (C - 39%, L - 32%, LD - 21%).

Let's go back as far as 2009 in British polling and see the same effect, all polls in the same time period:

Angus Reid: Cons - 40, Labour - 24, Lib Dems 20

ComRes: C - 38%, L - 29%, LD - 19%; YouGov: C - 40%, L - 28%, LD - 18%.

Maybe I'm simply building a pattern up in my mind, but there has to be a habit there. I can even demonstrate it with Canadian polling, during the height of the Coalition crisis, same time period:

Angus Reid: Cons - 42%, Libs - 22%, NDP - 18%

Ipsos Reid: C - 45%, L - 26%, NDP - 12%;
Ekos: C- 42.2%, L - 23.6%, NDP - 15.0%;
COMPAS: C - 51%, L - 20%, NDP - 10% - and the list goes on.

I'm not crazy, right? There's definitely something odd going on there.

That brings us to the Ontario polling that Angus Reid did, with Tim Hudak's Progressive Conservatives on 41%, McGuinty Liberals on 29%, and Horwath's NDP on 22%, and the Greens on 8%.

The last poll that Angus Reid did for Ontario on Dec. 4 had basically the same results, but 41-27-20. That rubbed against Ipsos Reid's findings, indeed a few days later on Dec. 13, had the Liberals on 38%, PCs on 34%, and the NDP on 15%. What do you think?

You can draw your own conclusions from what I've pointed out here, but this is mine: Angus Reid, most likely due to their methodology, whether they weigh the third party results more than the second party results, or whatever, seem to give the third parties more of an influence and position as compared to the second party, which is always consistently lower than what other pollsters reflect. That leads to some suspicious and less-than-reliable results, in my opinion. Like EKOS, the polling always seems a little more whackadoo than it should be.

No comments:

Post a Comment