Pre-election peaks and doldrums, or, a lesson for Alf (UPDATED)

I’m mostly going to leave readers to draw their own conclusions about the latest epistle from Liberal Party president Alf Apps. His analyses of the press gallery and of his party’s leader constitute fair comment. But I admit I’m flummoxed that a Liberal official is still poring over polling data from between elections to seek comfort.

Here’s a secret about elections: they don’t happen between elections. Elections are almost always simultaneous with elections. So one thing people who are interested in politics — journalists, presidents of major political parties — should do is pay some attention to the way voter behaviour on election days compares to voters’ predictions of  their own behaviour when elections are distant and hypothetical.

To help, I’ve swiped a chart from the estimable Nanos Research, which shows party-support trends since 2002. (Pause.) Ok, for whatever reason, WordPress doesn’t want to insert the chart in this post, so just click on this link to load your own .pdf: Nanos trend

Now, just about the most reliable trend in the chart is that every time there’s been an election, Conservative support has jumped smartly upward while Liberal support declined as sharply. You see it happen in 2004, when what looked like a Liberal rout of the Conservatives turned into the loss of the Liberal majority. (Memory plays tricks. It was not at all clear, on the day Paul Martin dropped the writ, that his majority was even in danger, and when Jean Lapierre said several days into the campaign that he expected a Liberal minority, it was covered as a big gaffe.) You see it again in 2006, and you see it most spectacularly in 2008. But you also see it in the autumn of 2009, after Michael Ignatieff announced that Stephen Harper’s “time was up” and we seemed to be headed for an election.

As a rule of thumb, the Harper-era Conservative writ-period bounce seems to be about five percentage points or a little more. The Liberal writ-period decline is comparable. Which means if the two parties are tied in voter support on the day a campaign begins, the Liberals should, as a rule of thumb, expect to be 10 points behind when people actually vote. Right now the two parties are not tied.

Of course history isn’t fate. There will be elections where the Conservatives don’t benefit from a 10-point swing during the writ period. But if you’re writing an 18-page memo about polls sometime soon you might want to mention this very robust trend.

UPDATE, Sunday: Many commentators say three data points (2004, 2006, 2008) is a flimsy data set. Quite true. Here are two more. The 13th link from the bottom (“Update on the Federal Political Landscape”) from a list of old Ekos polls shows you an Ekos/Torstar poll from 2002; like Nanos, Ekos gives a longish time series of its party-preference polling. What kind of jumps out is that the two lowest troughs in Liberal support since the late 1990s are the two moments when Canadians actually voted: the 1997 election and the 2000 election. If anything, the combined swing illustrated by Liberal declines and PC/Reform/Alliance gains is more than 10 points.

So that’s five federal elections in a row, 1997 to 2008, where the trend is for the Liberals to bottom out and the (assorted conservative, then Conservative Party) opponent to have better results than recent polls had indicated. Note that this isn’t about “governments” dipping while “oppositions” gain: conservative parties posted writ-period gains against the Liberals without regard to which of them was in power.

I’m told this pattern goes back decades. I’m looking for confirmation.