Corporate Research Associates (of the NS Conservative Party)
Perhaps someone that knows more about polling can explain to me how a poll of 500 people taken over two weeks can be considered a valid snapshot of voter intentions. I have looked at the numbers of Corporate Research Associates second strange poll of this provincial campaign and I am having a hard time reconciling how this at all makes any sense.
The numbers break down like this:
Conservatives 38%
NDP 36%
Liberal 20%
None of the above or others (Green?) 6%
Undecided 15%
There are so many problems with interpreting this poll that I don't know where to begin. First, as my five year-old daughter figured out when I showed it to her 38 - 36 < 4%, the stated accuracy of the poll, therefore asserting that
The numbers break down like this:
Conservatives 38%
NDP 36%
Liberal 20%
None of the above or others (Green?) 6%
Undecided 15%
There are so many problems with interpreting this poll that I don't know where to begin. First, as my five year-old daughter figured out when I showed it to her 38 - 36 < 4%, the stated accuracy of the poll, therefore asserting that
it is probable that the Tories would win re-election with an outside chance ofis pure BS. The error bar means not that they are close, but that statistically the Conservatives and the NDP are tied. It would be every bit as valid to say that
a majority government
it is probable that the NDP would win re-election with an outside chance
of a majority government
These two phrases are eqally (in)valid with respect to this poll. I find it telling that this is not how the poll is being spun.
And, if a poll is supposed to be a snapshot of the attitudes of the public at a given time, why are we asking questions of two weeks and saying things like "support for the NDP over the past week has been increasing"? This poll is every bit as fishy as the one that CRA published at the beginning of the campaign, which asked questions over ten days, during which time the Liberals and NDP had only released part of the campaign platforms.
If I was a Liberal, yeah, right, but if I was, I'd be very unhappy at the way this poll has been conducted and is being interpreted. And they are.
