(Minor revisions made on January 23, in two tranches)
I keep hearing and reading in coverage of the Canadian campaign that the NDP will bleed voter support on election day, just as it did in 2004. But did that happen in 2004? See the 2004 poll tracker at CTV: There is a very small downward movement in the NDP vote between the June 25 poll and the June 28 election, but not what I would expect from the “bleeding” analogy, and well within any poll’s margin of error. Now compare Political Arithmetik‘s 2006 poll tracker. It shows the NDP gaining in recent polls. Now that does not mean that NDP-leaning voters won’t go vote and change their minds and vote Liberal. But, taken together, these trackers suggest that there is so far no evidence that what did not happen in 2004 is happening “again” in 2006. Quite the contrary.
The more striking thing about the 2004 tracker is the sharp gains for the Liberals at the end (which appear to have been at the expense of Greens and, to a lesser degree, Conservatives, but not the NDP). Will we see the same again in 2006? My gut says yes, though not as dramatically. The late 2006 polling (see my seat estimation post below) suggests there might have already begun a late upward trend for the Liberals (although that is not evident in the Political Arithmetick tracker
because it stops at January 17). The 2006 tracker shows a downward trend for the Greens, but as already noted, not for the NDP. Quite the contrary.
(Actually, looking at the 1/22 update of the Political Arithmetik 2006 tracker, it is less clear that there is a downward trend in the Green vote, and it may even be a marginal upward trend. Could it be that the left of the political spectrum is less prone to strategic desertion–at least as of a few days before the election–because it expects a conservative plurality no matter what?)