Tuesday, July 25, 2006

The real push-poll problem

While LKO is concerned about the Progressive Bloggers poll question on Lebanon, how has the bias in the Ipsos-Reid poll and resulting commentary claiming that Canadians largely support Harper's position gone without comment?
Almost two in three Canadians believe Israel's military action in Lebanon was either somewhat or completely justified, a new poll has revealed.

The survey, conducted online by Ipsos-Reid for CanWest News Service and Global National, found that 64 per cent of Canadians believed Israel's action was either somewhat or completely justified.
In other words, the premise of the article lumps together what was presumably the middle option with one of the extremes without differentiating between the two. A quick trip to the Ipsos-Reid site breaks the proportions down as 20% "completely justified" and 44% "somewhat justified", which paints a rather different picture than the suggestion that the two can be considered equally. (Indeed, one could equally describe the poll instead as reflecting a view that 80% of respondents see Israel's response as less than completely justified.)

But having started from that biased perspective, the headline and commentary go several steps further, implying that the combined group is absolutely united in support of the extreme position:
"When you look across the country, Mr. Harper has good support in every region on almost every measure of both foreign policy and his approach to this matter, except in the province of Quebec, where it tilts clearly to an Israeli compromise," said Wright.
One would think the credibility of the article and poll couldn't get much worse. But then there's this gem of a question from the original poll:
When asked which side of the conflict should make a major compromise in order to have a ceasefire, 63 per cent of Canadians said it was "those who kidnapped the Israeli soldiers," while 53 per cent of Quebecers said it was the Israeli government.
Let's leave aside the sheer absurdity of the implication that "those who kidnapped the Israeli soldiers" would actually be at the bargaining table such as to be able to make compromises. Even ignoring that obvious problem, does Ipsos-Reid really believe it's possible to get a fair result by comparing a neutral descriptor of one side to a loaded description of the actions of the other? Or by analogy, how much credence would be given to a poll asking "Who should compromise, Hezbollah or the army currently bombing Lebanese civilians"?

In sum, it seems painfully clear that the poll is bound to receive tons of attention as a sign of support for Harper's Middle East policy. But the poll was obviously flawed to begin with, and even from that starting point the results have to be twisted beyond recognition to be described as backing Harper's view. Which should only highlight the lengths the Cons and their friends in the media need to go to in pretending that Canadians share the values of their current PM.

No comments:

Post a Comment