Thursday, August 2, 2012

Polling in Quebec

Three companies have done multiple polls in Quebec since April 1st of this year and the results look rather different.   We had two new polls released in the last day, one from Forum and another from Leger (3 pm: I have found out about an internal one done by CROP for the CAQ)

Pollster       Lib PQ  CAQ QS Vert Oth
Forum Aug  1   38  39  14   4   3   0 - 1617 IVR
Leger July 31  31  33  21   7   4   4 - 1648 internet panel
CROP July 25   31  30  24   -   -  15 - 1000 internet panel

The pollsters have some dramatically different results even though they were in the field at the same time.   Forum has 77% opting for one of the two major parties while Leger only has 64% and CROP 61%.   That is a huge difference and these polls indicate dramatically different election results.  They are far enough apart that someone is very much incorrect.   The gap is large enough that it is possible for all of them to be fundamentally wrong.

Forum was using a telephone based random sample methodology, Leger was using an opt in internet panel.   CROP is also using an internet opt in panel method.   I am less and less certain that internet panels are giving us useful data.

Pollster      Lib PQ  CAQ QS Vert Oth
Forum Aug  1  38  39  14   4   3   0
Forum May 17  34  33  18  10   4   1
Forum May 15  35  33  19  10   3   0
Forum Apr 23  35  35  16   8   6   0
Forum Apr 10  34  39  18   6   6   0
Leger Jul 31  31  33  21   7   4   4
Leger Jun 14  33  32  19   9   4   3
Leger May 21  32  32  21  10   4   1
Leger May  2  28  31  24   9   4   4
Leger Apr  4  27  33  22   7   5   6
CROP  Jul 25  31  30  24   -   -  15 (internal party poll)
CROP  May 25  31  30  22   9   5   4
CROP  May  3  31  25  24   8   8   4
CROP  Apr 23  30  28  25   8   7   2


There seems to be a significant difference between the results of Forum and those of CROP and Leger.  Is this difference due to methodology used?  In April and May Forum was getting 67-73% for the top two parties versus 59-64% for Leger and 56-61% for CROP.   The difference seems to come with CAQ and other being lower with Forum versus the other two.


Segma Recherche released one poll in May which was a traditional telephone poll.  Their results would seem to fit with Leger with respect to the Liberals and PQ but Forum with CAQ.  They also had the highest result for the Greens since May 2011.   There was also an internal poll for the PQ done by Repere July 13-17th.  Liberals 30%, PQ 33% and CAQ 21% - no idea what form the poll was.  Also, it was internal and my experience of internal polls is that they tend to have less rigorous overview than public polls.


I do not know how to use the polling data in a useful manner to project the Quebec election if the polling results are so far apart.

1 comment:

DL said...

Interestingly in BC there is no evidence at all that online polls inflate the support of "minor parties" (ie: Greens and BC Conservatives) compared to IVR polls. This seems to be a purely Quebec phenomenon