The Canadian Press - Federal Tories, Liberals locked in statistical tie, poll suggests, Joan Bryden, The Canadian Press, 15/12/2010
http://www.winnipegfreepress.com/canada/breakingnews/federal-tories-liberals-locked-in-statistical-tie-poll-suggests-111935134.html
The two polls were taken fairly closely together with no, as far as I am aware, 'poll moving moment' for any of the parties in between.
Also, the Liberals are polling the same in both polls Nationally. Regionally the polls overall are the same, well within the margins of error, in Ontario and Atlantic regions and almost identical for S-M (Saskatchewan/Manitoba).
The number seem to diverge greatly in Que, BC & Alberta were the differences for the Con are well outside the margin of error, but the differences in numbers for other parties are within the margin of error.
The difference in Que might be explained by the fact that for the HD polls their numbers add up to 95% as opposed to the IR adds up to 99%. In other words the Con number for HR may be a typo and perhaps more like 11+4 = 15 which would bring it within the margin of error with the IP number. This would give the Con's a few more points Federally as well, to say 33 - 34 (est. only).
It seems to me that normally the explanation is the variation in questions/questioning and perhaps the media used. There is also that 1 in 20 where the poll simply doesn't accurately reflect opinion, or the margin of error is huge.
However, you might expect that if it is a problem with questioning, or the 1 in 20, that it would bias the results for all the parties, which does not explain the similarities (almost equal) numbers for the Liberals across the board or the result for all but BC & Alb.
A comparison with other recent polls suggests that the Ipsos-Reid polling number for the Con's is way off and it is the results for BC & Alberta that may very well be causing it. Why would their poll be so wrong for the Con's and not the other party and in BC & Alb and not the other regions.
It would be interesting to have spokes persons for both Pollsters discuss these differences. It would sure help to understand the significance of their polls, as released to the public anyway.
excerpt: Lloyd MacILquham cicblog.com/comments.html
| Harris-Decima The Canadian Press 16 Dec.'10 - Nationally: Con: 31 Lib: 29 NDP: 15 Green: 11 Block: ? Regionally: - Ont Con: 36 Lib: 36 NDP: 14 Green: 12 Block: N/A - Que Con: 11 Lib: 23 NDP: 10 Green: 7 Block: 44 - BC Con: 32 Lib: 21 NDP: 24 Green: 21 Block: N/A - Alb Con: 47 Lib: 24 NDP: 14 Green: 12 Block: N/A - S-M Con: 48 Lib: 25 NDP: 22 Green: 5 Block: N/A - Atl Con: 35 Lib: 42 NDP: 17 Green: 6 Block: N/A | Ipsos Reid Poll Postmedia News and Global TV 10 Dec.'10 - Nationally: Con: 39 Lib: 29 NDP: 12 Green: 9 Block: 10 Regionally: - Ont Con: 39 Lib: 34 NDP: 14 Green: 11 Block: N/A - Que Con: 21 Lib: 22 NDP: 7 Green: 8 Block: 41 - BC Con: 44 Lib: 26 NDP: 18 Green: 12 Block: N/A - Alb Con: 65 Lib: 16 NDP: 6 Green: 12 Block: N/A - S-M Con: 49 Lib: 28 NDP: 19 Green: 2 Block: N/A - Atl Con: 39 Lib: 44 NDP: 14 Green: 1 Block: N/A |
| 2.2 percentage points, 19 times in 20. The margin of error is larger for regional results. | margin of error of 3.1 percentage points. The margin of error for the regional results are: B.C. (9 per cent); Alberta (9.8 per cent); Manitoba/Saskatchewan (12.1 per cent); Ontario (4.9 per cent); Quebec (6.2 per cent); Atlantic (12.1 per cent). |
