Polls Are For Dogs

Bias in the mainstream media? That's just a conservative fairy tale isn't it?
"Ottawa — A new poll suggests that at year's end, the Conservatives and Liberals were in a virtual dead heat in political support."

"The Decima Research survey, made available to The Canadian Press, suggests the Conservatives had 34 per cent support, compared with 31 per cent for the Liberals, a spread within the poll's 3.1-percentage-point margin of error."

Some good statistical analysis a goings on there. Margins of error are there for a reason. We know the poll reveals a Tory lead, but that statistically it is insignificant.

So then you'd figure that that rigorous statistical analysis would be applied to regional results which show a Liberal lead in Ontario and Quebec? Think again.
" In Ontario and Quebec, however, the survey numbers indicate the Liberals had an edge on the Tories....The poll gave the Liberals 40 per cent support in Ontario, compared with 35 per cent for the Conservatives... The margins of error are higher for regional samples."
That "higher" margin of error was almost surely above 5% seeing that the national confidence interval was over 3%. Given that the poll shows a lead, but it is statistically insignificant as it was in the national poll.

They were more than willing to point out the statistical insignificance of a 3 point lead in a poll with a 3.1 margin of error.

But when the lead is for the Liberals with a larger margin of error, it's no longer a "dead heat" it's and "edge"?

Can't the Liberal apologists at the Globe come up with more subtle ways of injecting their bias into everything they write?

No comments:

Post a Comment