Friday, August 10, 2012

Polling in Quebec - comparing different pollsters recent polls

Depending on which pollster you follow the election campaign in Quebec has either been very interesting or has seen no change.

         Lib PQ CAQ QS ON Vert Oth
August 8  31 32  27  6  2   2   0
July 1    31 33  21  7  2   4   2
Change     0 -1  +6 -1  0  -2  -2

         Lib PQ CAQ QS ON Vert Oth
August 7  32 34  24  6  0   3   0
August 1  38 39  14  4  0   3   0
Change    -6 -5 +10 +2  0   0   0

         Lib PQ CAQ QS ON Vert Oth
August 8  29 32  21  8  2   3   0
May 25    31 30  22  9  2   5   2
Change    -2 +2  -1 -1  0  -2  -2

Leger and CROP both have the Liberals and PQ stagnant with minimal change, meanwhile Forum has a dramatic fall in the support for those two parties.

Forum and Leger both have significant rises in CAQ support meanwhile CROP has CAQ not changing much.  Yes, I see that CROP last polled in May, but their CAQ numbers are out of sync with the other two pollsters.

It all speaks the fundamental flaws in the polling done in Canada.    It is very hard to know which pollsters and which polls are anything close to accurate reflection of public opinion.   

When we look at the three latest polls the results look closer to each other but are still not that close.    The results for CAQ are too far apart to fit within what should happen statistically if the methods used were equally valid.   

            Lib PQ CAQ QS ON Vert Oth
Leger Aug 8  31 32  27  6  2   2   0
CROP  Aug 8  29 32  21  8  2   3   0
Forum Aug 7  32 34  24  6  0   3   0

What can we take away from the polling data?   That is hard to say with any precision.  What I think I am seeing is that there is no strong love for either the Liberals or PQ but a decent level of interest in CAQ.   Leger's poll did ask about second choices and  CAQ was by far the strongest with almost 1/2 of Liberal voters saying CAQ as their second choice.   CAQ was also the second choice leader with PQ voters.

If the election becomes about replacing Charest as premier, with momentum Francois Legault may become the primary choice rather than the PQ.    If the election becomes about stopping the PQ, CAQ may become the primary vehicle to do this among the francophones.   

Realistically we do not have nearly enough decent data to be able to project the Quebec election with any accuracy.   The one thing we can not do is average the various polls and assume that the resulting number reflects anything approaching reality.   Averaging the results only compounds the errors of the polls while at the same time giving the resulting number the patina of increased accuracy.

The reported Leger numbers for August 8th have been rounded, we can get the data to the tenth of the percentage point without "other" though the data they reported

            Lib  PQ    CAQ   QS   ON  Vert  
Leger Aug 8 31.0 32.1  26.9  5.5  2.1  2.4   

with "other" which seems to have been 27 responses out of 1417

            Lib  PQ    CAQ   QS   ON  Vert Oth 
Leger Aug 8 30.4 31.5  26.4  5.4  2.0 2.3  1.9

No comments: