St. Paul's vote a national battlefront?

That was the Star headline this morning. Since they included a question mark at the end, I'll answer them: no.

Putting aside that most of the "senior anonymous insiders" quoted in the story are simply trying to manage expectations, let's assume for a second the Tories do win St. Paul's. They won't, but we did just see a 20 point margin of victory safe seat switch hands in a by election a few months ago, so it's not ridiculous to speculate about this.

What is ridiculous is to assume this is some sort of Harper versus Ignatieff litmus test, that will impact when the next federal election happens. By elections, by their very nature, are pretty awful indicators of public opinion - trying to extrapolate out the results of a provincial by election to the federal political scene is about as useful as this.

Here's the federal Liberal vote share over the last 40 years, plotted against the provincial Liberal vote share in the election that preceded it:



For the more numerically inclined among you, there's a -0.583 correlation between the two variables - in normal speak, that means when the provincial Liberals do well, the federal Liberals don't. And, even though I used 40 years of data, it's been that way since the dawn of time (or, at least, the dawn of confederation).

If you prefer, we can use the change in provincial vote to predict the change in federal vote...at least that should control some of the secondary factors, right?



Well, once again, we get a negative correlation (-0.259).

So even if by some miracle the PCs do win St. Paul's, it won't tell you a thing about the upcoming federal election. If it did, you'd have to conclude that after the John Tory by election fiasco, Michael Ignatieff is poised paint rural Ontario red this fall.