They call it "polling" these days. I mostly ignore them. At best, they're inaccurate; at worst, they're a distraction from genuine political discussion, and I loathe their contribution to "horse race" news coverage.
Poll-reports were a joke during the last Canadian election, and I don't expect any improvement this time around. The best public analysis I heard came the day before the election, from a friend of mine who is in politics. A friend of his had been canvassing hard for his Conservative candidate, knocking on a lot of doors and shaking a lot of hands. He told my friend there was no way the Conservatives were going to win this one. My friend asked why, and the explanation went something like this:
"I've been in politics since as long as I've been conscious. I canvassed with my father on his first campaign, and I've canvassed ever since. In the last 20 years, I've developed a pretty good sense of what people's unspoken inclinations are. Usually when they tell you they haven't yet decided who they'll vote for, they're just bluffing to get you off their porch. This time around, the people telling me they were undecided were genuinely undecided. They had no idea how they were going to cast their vote. And when you are truly unsure of who you're voting for, the one thing you will almost never do is take a chance on that vote. We're in for another Liberal government."
I'm not so sure that's the way the vote will "swing" this time around. If I were to prognosticate, I'd say: another minority government, quite possibly Conservative (no majority in their cards, though: Quebec will never vote Conservative, and Ontario is still Conservative-shy after being gutted by Mike Harris). Also this: the lowest voter turn-out, ever.