- Quebec Margin of Error, +/- 64% ??? That's One Big Error

Posted: 12:38 PM on April 25, 2011
Nanos Poll
Strength in Ontario puts ‘squeaker of a majority’ within Harper’s reach, Jane Taber, Globe and Mail Update, April 25, 2011
http://www.theglobeandmail.com/news/politics/ottawa-notebook/strength-in-ontario-puts-squeaker-of-a-majority-within-harpers-reach/article1997430/

The Polls methodologies may be suspect, both in its questioning and how they take their samples.

"Allan Gregg, chairman of Harris-Decima, which provides political polling for the Canadian Press. He is also a regular member of The National's At Issue panel on CBC TV.

Gregg has been doing political polling since the 1970s . . .
'there's broad consensus among pollsters that proliferating political polls suffer from a combination of methodological problems, commercial pressures and an unhealthy relationship with the media.

'The dirty little secret of the polling business,' he went on, 'is that our ability to yield results accurately from samples that reflect the total population has probably never been worse in the 30 to 35 years that the discipline has been active in Canada.'"
(CBC, 25 Apr.'11)

Gregg explains that 30 years ago 70-80 % of people called by phone answered the polls now it is 15%. Also, with cell phones the sampling can be worse.

One would expect that with only 15% answering the Polls a proportionately larger % that answer are highly motivated politically. This may explain the high numbers for the Con's as well as the large fluctuations in their numbers.

NDP are also typified by highly motivated supporters that one would expect would also jump at the opportunity to express their views.

Internet Polls or polls that are passively open to people to answer may, obviously, be biased. For example the Crop Poll that started this whole thing about an NDP 'surge' in Quebec.

Then, of course, there is taking a National Poll, one designed for National use and restricting the results to a Province.

If Polls are sensitive to questioning and sampling methodologies and even adjusted to take these things into account.

Then how in the world can someone justify taking National results and restricting it to Provincial results.

It is not simply a question of increasing the Margin Of Error because the sample size is decreased. The results of a Poll may vary greatly because of the design of the questioning from Province to Province as well as the sampling methodologies.

If anything, the Polls should be done the other way around - take separate Provincial Polls and combine them to make a National Poll - but what Polling Company would spend the money on doing that.

Oh and by the way, did I mention this article states:

"In Quebec . . . (There is a margin of error of plus or minus 64 percentage points, 19 times out of 20, for the provincial sample.)"

Taber - That's one big error.

excerpt: Lloyd MacILquham cicblog.com/comments.html