Thursday, February 21, 2008

Polling mysticism





















So today's poll has the Tories "flirting with a majority." Now wait... didn't yesterday's poll have them in "a dogfight" with the Liberals?


Apparently, yesterday's competitive Liberals are now suddenly today's underdogs fighting even to stay above historic party lows (of which their reported 27% would certainly be). The NDP, whose 12% in today's poll put them well below their 2006 support levels, were just two weeks ago at 19% and apparently well-positioned to make gains. Back then, the Greens found themselves at 7%, a standing barely above their 2006 election result and far short of their apparent rapid ascendance to 13% in the latest poll.

As political observers, we all ooh and ahh at each new horse race figure that's printed in the papers, trying to read into the numbers some sort of truth about who's ahead and who's behind and who's going to be sorry and who's going to be rewarded. The art of poll watching has taken on an almost mystical quality, with its high priests quick to ascribe some dramatic trend or, in some cases, validate the various notions and myths their particular political tribe feels a need to hold on to (which includes partisan false bravado).

For political junkies, prognosticating election results after every poll makes for fun exercise, but it's fairly pointless, for a number of reasons:

  • First, most of us who read about polls regularly know some of the reasons that poll numbers jump around: sampling (± a certain percentage, 19 times out of 20) and various sorts of non-sampling errors, and the way survey questions are asked (debates continue about what question wording best replicates how voters are likely to respond upon arriving at the polling booth, something that Blackberry Addicts were never exactly shy about raising).

  • Second, polls taken in between elections are often more about party brands than they are about the true competitive standing of each party relative to one another. It's what I often call a "parking lot" effect, in which poll respondents instinctively "park" their vote in between elections. Asked by a pollster who they'll vote for and they'll tend to blurt out whatever party brand they roughly associate with or that happens to be top of mind. The Liberals and Greens tend to be beneficiaries of this tendency in most parts of Canada.

  • Third, pre-election campaign polls don't take into account party organization, finances, gaffes or infighting, strategic positioning, the name recognition of incumbent "star" candidates, media exposure, and party messaging, all of which play a huge role in that complex game of chess that plays itself out in each election campaign.


For these reasons, the hype that follows most polls tends to be pretty hollow. A better analysis might take into account some of the above factors.


The Greens, currently buoyant in most polls, have considerably less ground organization, campaign experience, cash resources, star candidates, or media exposure than the other three national parties and will almost surely sink when the next election campaign goes live. Nik Nanos, of SES Research, suggests that the Greens typically drop 1/3 of their vote from the last poll to the actual election. That's after the deflation of the Green balloon that's likely to occur during the campaign itself.

The Liberals, in their current state of endless infighting, fundraising difficulties, mass retirement of MPs and loss of valuable potential candidates, to name just a few of their problems, had best shape up and fast if they expect to hold the 27% to 35% the polls currently peg them at.

Worse for the party is the unenviable position they keep backing themselves into: critiquing the government but then doing all they can to make sure it isn't defeated. For them, voting to defeat the government in a confidence motion means risking catastrophic losses in an election (for some of the reasons noted above), yet supporting the government or abstaining from confidence motions means branding themselves as a "weak" opposition that has little alternative agenda to that of the government.

A weak, indistinct, vision-less Liberal Party is one that's going to find itself wedged in an election between the Conservatives and the NDP when centre-right Liberals find they have no reason to vote for a "me too" Liberal rather than a Tory and when centre-left Liberals find that the alternative vision they're looking for lies only with the NDP. Is it any wonder, then, that the Tories and NDP are literally trying to goad the Liberals into an election, while the Liberals are finding ever new and interesting ways to contort themselves in an attempt to avoid just that?

We live in very interesting times.

Photo: a map of the 2006 federal election results.

2 comments:

Michael said...

There are lies, damn lies, and statistics!

cherenkov said...

Good analysis, PT.