Thursday, April 21, 2011

Is the Game Changing?

We have a whole series of federal polls come out that show there has been a significant shift in the federal election.    The last three elections have seen few changes in seats and this election looked like as few as than 20 seats would be changing hands, but with the recent change in trends in the polls.

I should add a personal caveat here.   From my background in doing a lot of statistical analysis, I have serious concerns about the methodology and results of most polls in this country.  I have highlighted the issue that many more people give an answer to "How would you vote?" than will actually vote.   There is also the issue with how far apart the polls are from each other and how far they are from the final result.   It should only be in one election in three that a 2000 person poll should not be within +-1% of the final result of all the parties.   There are a host of sources or error that are not accounted for in polling and mean we have results in the polls that are not nearly as accurate as people think.

That said, I look for broad trends in the polling and what is happening over time with any one company.   The trends are telling me something significant is happening.

Until the debates the NDP was down from their 2008 result and the Conservatives and Liberals were up.  Since the debate the support for the NDP has been growing at close to one percentage point per day nationally.    The Conservatives and Liberals look to be hovering close to their 2008 numbers, a loss of about two percentage points each in the last week.

The real shock comes in Quebec.    Polls came out yesterday and today that indicate that the NDP is in the lead in Quebec.   CROP, Forum and Ekos all have the NDP in the lead.  Of most interest to me is the CROP poll because it was a 1000 person sample.  I do not know how big the Forum regional number was but I am estimating it was about 400 to 450.   The Ekos Quebec regional number was 587, which is a decent size.

Ekos has a small Montreal sample, but it fits with the CROP Montreal numbers and shows that the NDP is in the lead in the city and the Conservatives are non-existent.

The trend of the polls in Quebec has been that the NDP has been ahead of their 2008 results in every poll.   Where are the NDP at in Quebec?   At this point if I were to try and peg their realistic support, I think it is at 26-31%, I may change my opinion if more polls confirm the latest results over a number of days.

Meanwhile the Conservatives are down in 17 of 19 polls with samples of at least 300 in Quebec.   the Liberals are down in 18 of 19.   I see the Conservatives at around 16-18% and the Liberals at 19-20%

Before the debate the Bloc was running close to where they were in the 2008 election, since the debate their support has dropped significantly, being regional samples and small, the various numbers jump around in the province.   I see the Bloc in around 32-34% of the vote in Quebec at the moment.

So what does this all mean?

If at the end of the election the NDP does win more votes than the Bloc in Quebec, it will translate into seats, somewhere between 10 and 30 seats.   The seats come equally from the Bloc and the Liberals with only one from the Conservatives.

A higher NDP in Quebec may influence the vote in Ontario and boost the NDP in that province.   The NDP trend in Ontario in this election has been a slow but now noticeable increase.  They seem to now be close to their 2008 result.   A positive boost from Quebec could move the NDP to higher numbers in Ontario than the best election results under Broadbent - 21.8% in 1980.

I was planning on revising my prediction based on the campaign and polls and such today, but things are in too much flux for me to make a stab at it today.
Post a Comment