Monday, November 28, 2011

Nanos = Voodoo Poll?

I have huge respect for Nik Nanos and his firm, but this recent poll is just a bit much, even for a die-hard Liberal:

Con - 35.6% (-1.1%)
Lib - 28.1% (+4.7%)
NDP - 27.3% (-2.7%)

So, somehow the Liberals have managed to claw their way back up to second place in the popular vote (though really, its within the margin of error between us and the Dippers), coming in first in Ontario (38.8% to the Con's 37.2%), and coming into Dion-level territory in Quebec (23.6% vs. 37.7% for the NDP, and 20.1% for the Cons). This is not a fantastic poll in historical circumstances, but it's better than usual for us.

Putting it through the projection system, we'd get 129 Conservatives, 92 New Democrats, 86 Liberals, and Lizzy May. Easily enough for a coalition government that might be slightly confusing (both parties have similar percentages of seats).

But wait, why are the NDP still in second place in terms of seats? Well, even with their low numbers in Ontario (19%) and comparatively to 2011 in Quebec, they still maintain 52 of their 59 seats in Quebec, and 15 of their 22 seats in Ontario. Considering their general stability in the smaller parts of the country, this is enough to keep them afloat. The Liberals, meanwhile, need to have greater distance between themselves and the NDP in order to come out second (or first) again.

But we've seen this before. In early August Nanos had a similar poll that was promptly turned around next month. Their August 2nd poll was never supported by any other polling company, though their subsequent polls (until now) fell back in line. I would say this poll is likely the same - unless other polls show the same thing.

However, what event in Ontario could have possibly boosted the Liberals that well? One might say the October election, which saw the McGuinty Liberals re-elected to a near-miss minority, could have prompted the huge rise in Liberal fortunes (they would win 46 seats in Ontario with the Nanos poll)? However, Nanos' previous October poll showed no "McGuinty bump."

Considering that Ontario accounts for the Liberal's random rise, and considering the lack of obvious stimuli which prompted this rise... you know where I'm going with this.

The poll does confirm a few things us, however. The NDP are facing a bit of a pinch in Quebec that's dropping them from +40% to 35-40%. None of the polls can decide who is actually benefiting from this cause, with Nanos giving the Liberals and Conservatives the rise, Leger giving it to the Bloc, and CROP giving it to the Conservatives and the Bloc. Overall, however, the trend seems fairly strong - the NDP are slightly down, with the Conservatives and Bloc being the main beneficiaries, and the Liberals probably strong again among non-francophones.

What do you think? I'm giving Nanos a chance in my aggregate projection, though with less weight than usual given the funky results of the poll. It's got some legitimate trends (i.e. Quebec, Atlantic, etc.), but the Ontario numbers are just so out of step with other polls so far. It'll be interesting to see what happens, but for now, I think its a bit of a voodoo poll.


  1. Nanos reported the Liberals and NDP tied in August and what happened before their next poll in September? Jack died. The NDP were losing ground the summer then they were boosted by Jack's death but it now appears they are losing ground again.

  2. Ehh I considered that but the fact is that no other poll at any point showed what Nanos showed, and that's pre and post Layton's passing. In fact, the NDP remained remarkably stable over that period. Either Nanos is super sensitive to these things or something's wrong.