Libs & Cons In Virtual Tie!
Given MOE, one could say Libs & Cons in virtual tie, maybe as close as 30.2-31.8. Or one could give this spin: "Given the margin of error, the results are essentially no different than the 36-27 split garnered by the two parties earlier this fall". Both my interpretation & CG's are spin, given MOE. So why not just say, race seems to be tightening slightly, 34-28, though given MOE of 2.2, Cons could be in same dominant position as fall or Libs could be in virtual tie. My point here is not this or any particular poll but the cliche spin on the MOE, sometimes used to infer dramatic advance by 2nd party, sometimes used to extrapolate huge lead for 1st party. C'mon, that's just silly, either way.
posted by Eugene Forsey Liberal @ 7:42 PM
2 Comments:
At 8:18 AM,
Anonymous said…
Your statement is right, if you use only the poll you are commenting on as information. However, trend lines are also information one should use. If a poll spikes a trend line for no apparent reason one might consider the error is pushing the MOE limits, or it may also be the 20th poll. But if a poll is a consistent with the trend line it is fair to give the poll more weight than the MOE would seem to allow. So using the trend line it is highly improbable that the Conservatives and Liberals are in a tie, even a virtual tie.
At 8:15 PM,
Eugene Forsey Liberal said…
I concur. But that's no reason to oversell poll changes. My title was deliberately misleading so as to illustrate how similarly misleading the CG's spin was. The last couple of weeks, given the different polls, indicate a slight tightening. So one could oversell it either way, using the MOE. And one shouldn't. Of course, I understand that it's a bit boring to be more punctilious and people are selling a product, but it is dangerous, as the spin becomes the perceived reality when delivered by an apparently neutral source like CG.
Post a Comment
<< Home