Election night was tough for our friends in the Liberal Party. We know what that's like. And so we can well understand why they and some of their apologists have been tempted to try to de-legitimize the key election outcomes -- the rise of the NDP to Official Opposition and a Conservative majority, realigning Canadian politics around these two national political parties -- by saying the latter only occurred as a result of vote splitting by the former.

The argument goes like this: Stephen Harper gained more than 20 seats in Ontario by increasing his vote by only five percentage points. The victory in these key ridings (so it is claimed) happened because the NDP supposedly drained votes from the Liberals, giving the Conservatives more seats.

Poring over the entrails leads me to a couple of observations.

First, as is usually the case, the change in the distribution of seats which commands headlines is an imperfect reflection of the change in the distribution of votes.

The NDP breakthrough in Quebec was remarkable and historically important and unprecedented, but so was the NDP surge in English Canada. The NDP's 30.6% share of the national vote was based on 43% of the vote in Quebec, and a very high 28% share of the vote in the rest of Canada.

Strikingly, limited seat gains outside Quebec poorly reflects the actual change in the NDP vote. Between 2008 and 2011, the NDP share of the national vote rose from 18.2% to 30.6%, a gain of more than 12 percentage points which is a huge shift.

Two women are sitting on a bench. After a while the first woman says, "Oy!"

The second woman replies, "Oy!"

The first woman says, "All right, enough about the children."

Kind of like how Canadians think of their political choices these days, seems like.

For many months now, we have been treated to a steady stream of articles and television reports about political polls, breathlessly over-thinking variations that fall within the margin of error. Analysts embarrass themselves trying to think of things to say about gross swings in regional numbers, which are then contradicted by other polls and then by the same poll, taken a week or two later.