Quote:
Originally Posted by Cowperson
Your own link is only confirming what I was saying . . . .
http://www.sfu.ca/%7Eaheard/elections/polls.html
. . . . . in the link I later added . . . . . .
http://network.nationalpost.com/np/b...-majority.aspx
. . . . . that the Conservatives were trending towards a majority in early to mid-September.
Then the full weight of the economic collapse began to be reflected in heavy equity losses and gathering panic on a global basis as we marched into late September/early October where, again, a critical two week period at the end of the former/start of the latter had markets losing about 33% . . . . . and polling, which had already flattened, took a dramatic shift at pretty much the identical moment.
Again, others might argue there were other, political factors involved in the Conservative polling reversal. Fine. I don't believe it but everyone has an opinion.
However, the point that the Conservatives were trending towards a majority in early September still stands.
Cowperson |
I don't really see what you call a "reversal." When looking at the numbers it's important to look at the average and not get too attached to a single, outlier poll. What I see is a slight downward correction as the campaign begins, from a peak of 38% to around 33-35%--both of which are VERY close to their final numbers of 37.5%.
When a single poll contradicts that trend, you have to ask yourself which explanation satisfies the test of Occam's Razor: that ALL the other polls are wrong, or that the single poll is probably an outlier.
The poll you cited has, you won't be surprised to learn, additional problems, because its prediction of a majority depends on smaller, regional samples. Their overall sampling error is based on a sample size of 1200--small but serviceable. But their "seat predictions" break that sample down into regions, and though they don't give us sample sizes for the smaller sub-groups, they have to be huge. How many Atlantic Canadians were in the poll? How many Ontarians? 200? 300?
If so--we're talking about a margin of error of somewhere between "this poll is utterly useless." and "this poll was made by Kevin Lowe."
In any case, we're splitting hairs: you claim they were polling at 43%, and they clearly never were apart from an outlier in September. (though indeed, they polled that high in May, and again in November). Even assuming that's true, those are hardly encouraging numbers for a PM who is literally opposed by nobody. As it happens, he was closer to 35-37% for most of the election cycle--and well before the stock market crash. If Harper wants to use that as an alibi, I suppose he can, but the numbers just don't bear it out.
Here's another listing of polls from last year and this--a bit more complete. You'll see what I'm talking about.
http://www.electionalmanac.com/canada/polls.php