Wednesday, March 2, 2011

When polls go horribly, horribly wrong

As Warren Kinsella noted, a new poll out done by Ipsos Reid has pretty much blown any hope that Liberals may have had left.

The poll, done over the same period as that hopeful Ekos poll, shows the Conservatives on a very strong 43% of the vote, the Liberals on a slightly-improved but horrible-in-context 27%, the NDP collapsed at 14%, the Bloc strong at 41% in Quebec, and the Greens at a sad 5%. This poll, with all its regionals calculated in and whatnot, would give the Conservatives with an excess of 170 seats, according to basic swing projections. The Liberals would follow with 65-70 seats, then the Bloc around 50 seats, and the NDP around 20. A strong, scary majority, though thus far the only poll to really give them one.

It's not necessarily all bad news, however. Contrary to what other polls are saying, the Liberals are up in Quebec with 27% of the vote there, much better than the 16-20% most are showing. The Liberals are also up in BC (21%) and Alberta (17%), while struggling with 33% in Ontario, and the same in Atlantic Canada.

For the Conservatives to get this kind of result, they hit the country in all the right places - 48% in BC, 52% in Atlantic Canada, 45% in Ontario. They earned 19% in Quebec, and though that's actually a drop, it's still enough to give them about 10 seats. But its definitely somewhere they can afford the drop.

With these huge numbers in the sweet spots of Canada, the Conservatives are virtually assured a majority. This is the kind of electoral coalition that, for them, leads to success. The question, however, is whether or not they could actually pull it off. Ipsos is showing they can, but I have my doubts. Remember how often we saw these polls back during the 2008 election?

The NDP have some real issues, though. In every region with the exception of Alberta (to an extent, 10% is good for them there), they are severely struggling with support. 22% in BC, 14% in Ontario, 12% in Atlantic Canada, 6% in Quebec - these are not good numbers. Makes you wonder where a lot of the Conservative vote is coming from - Liberals, or Dippers? Or are they taking from the Liberals, and the Dippers are shifting to the Grits to make up the shortfall?

What would have been interesting to see in this poll was where the Conservative surge was coming from in Ontario. Are they really on the warpath in Toronto? Has the Ford effect kicked in? Or are the Liberals hanging on in the GTA, while the Conservatives romp around the rest of the province with ease? Really don't know, and you could make sound arguments for both.

Optimistically, at the very least this poll shows the Liberals in a relatively OK position, especially when compared to something like Abacus. Cheer up, everyone, we're not dead yet.


  1. Enough with the frickin' polls already!

  2. Polls are the spice of political life, my friend.

  3. Meh, this poll is an outlier. Every other polling outfit has the conservative lead at it's more traditional upper level including one today...

    ... not to say that's a good result (it's not) but until the industry consensus says otherwise I'm more inclined to write the Ipsos poll off as a statistical aberration.

  4. Ipsos Reid huh? The same people who put Smitherman leading Ford right?

    Ya, they're never wrong. The Liberals are doomed, just like Rob Ford.

  5. Only 1001 people polled? Margins of error of 10% in BC and Alberta, 12% in Saskatchewan, Manitoba and the Atlantic provinces? No wonder more recent polls are already showing it to be an outlier.

  6. Hey, all I do is report what the polls say, and occasionally incite panic. Also remember that this was done last week, and shows results similar to Angus Reid and Abacus, in opposition to the narrowing that Ekos showed. Given that while the 43% may be something of a stretch, its not necessarily an outlier - it falls into a trend we've seen develop, and so far that only Ekos has bucked. It would seem to me that Ekos would be more of an outlier at this point.

  7. I'm not panicking. In fact, I'm entirely unfazed by these polls.

    But you are wrong about this one being consistent with a trend and EKOS not. The Harris Decima poll today is consistent with EKOS and with double the sample size.

    What we are not seeing is the rising undecideds. That doesn't matter on election day, so they say, so they don't include it. But it has been rising which squishes the gap and indicates that voters are very volatile and voter support is soft.

    It will continue to go up and down. I wouldn't get any more excited about a poll or several polls that showed the Liberals pull even.

    This is not how voters are thinking.

    Almost every Conservative I speak with is far less happy with Harper today than in 2006 and 2008, and those who don't support him are far far angrier with him. Thus the undecideds go up and where they land is the important question.

  8. Sample sizes aren't of a huge concern - whether or not they're following trends is, at least per what this discussion is. And to note, the Angus Reid poll which Ipsos agrees with had a tremendously larger sample size than what Harris has.

    But the rest of what you're saying is known facts. It doesn't change the reality that the Conservatives have seen a definite swing in the vote in their favour recently, and that polls are continuing to reinforce that. I'd also note that the two-percent narrowing in HD doesn't really reinforce the argument that its in line with Ekos, especially considering the margin of error, and the huge drop Ekos saw for the Conservatives, not present in HD. Statistical noise, more or less.

  9. One thing I've noticed is that the Conservatives are winning polls of support, but when it comes to polls on policy, the Conservative view is no where near the popular choice. Should we decrease corporate taxes? Polls say no. Are we happy with the level of transparency in the government? Polls say no. Do we trust the current government? Polls say no.

    The race is MUCH closer than most people think. The Conservatives have just been able to influence people with attack ads. When it comes down to Conservative ideals and policy...The majority of Canadians do not follow the Conservatives.

  10. But the same issue happens often with the Republicans south of the border. They're consistently on the wrong side of the majority of issues, but they're given support because of weak leadership in the Democrats, or voter backlash against one very major aspect of their history, policy, or leaders.

    Now, the sad fact is that the Liberals in 2008 faced all of those triggers for loss of voter support, and we're at least facing a lack of direction and a leader whose numbers are dragging us down. So while the race may be close, I honestly wouldn't be surprised at these numbers on election day.