Tuesday, April 1, 2014

An April Fools prediction on Quebec

308 did an excellent job today explaining why the CAQ internal poll should not be counted as a regular poll.

I'll expand on that.

1 - We don't know all the details.
This video clip from Yes Minister explains it very well.
Skip to 30 seconds in if you want to skip the joke before the explanatory and relevant joke.

Without knowing all the questions asked - word for word - and the order they were asked in, we can't know the true veracity of the poll.

2 - Good News.
Poll averaging has become the standard in the poll prediction and projection blogosphere, and with good reason. Averaging out scientific polls produces a more accurate result. The problem comes when you add non-scientific polls to the mix.

With parties deciding what polls get published, all the "bad news" polls are hidden away, and without those, you do not get a true average.

3 - Money.
A party could pay for 14 polls in the span of a week. The Media, and the polling companies themselves, can only pay for a limited amount. This can skew the poll results if a party decides that during it's climax it needs more polls.

Again, this screws with poll averaging.

Thus, it's not a good idea to include party polls with regular polls. Despite that, being April Fools, I have decided to release my prediction based on the new CAQ poll.

Note as well that I actually agree with some of the base assumptions behind this. As eluded to in my post about who won the debate.

I think while Couillard and Marois may have both botched a debate each, Legault did what he needed to. The CAQ may thus be able to save a larger portion of it's caucus than expected.

1 comment: