Canadian polls: democracy or sheep-herding?

One of the more interesting debates around the kitchen table as we proceed to a fall election is whether voter preference polls help or hinder democracy.

I argue they are a help, at least when they are reliable. The questions we ask ourselves as we approach an election are typically: which party/leader/cabinet do we want to govern? Which do we wish not to govern? And which parties unlikely to form government need and deserve our support if they are to remain effective in the House? Votes, in these deliberations, are inherently strategic.

If I prefer party X to Party Y, but am strongly opposed to Party Z, and if Party X isn't likely to win in my riding, it is most effective for me to vote for Party Y (if, that is, Party Y has a chance of forming government).

If Party Y is likely to form government or if the candidate for Party Y in my riding is likely to win regardless of what I do, then I might as well vote for Party X, my first choice (adding to the pool of votes it can employ to claim financial resources in the next election). I need to know, however, what other folks are likely to do if my vote is to be most effective in advancing my preferred outcome.

The counter to this argument is that knowing what others are likely to do encourages our neighbours to vote as sheep, simply to go along with what others are about to do. And sheep, of course, make bad democratic citizens.

However, there are few elections in which we are not told how folks are going to vote: by cab drivers, haircutters, party spinners, elites, and especially by journalists. Voter preference polls cut out these middle-persons, those who don't always know which way the wind is blowing, though they claim to, and those who have their own axes to grind.

The trick, of course, is to have public opinion polls that are reliable, predictive. Misleading public polls are much worse than none at all.

How, then, to distinguish the good from the bad and the ugly?

Good, bad and ugly polls

The worst sort of poll is one that’s not really about gathering reliable data at all, but is about manipulating public opinion. The worst of these is the push poll, one that leads the respondent, with the nature and tenor of prior questioning, to an answer the "pollster" really wants to hear. Some polls, short of the push poll, are nonetheless spoiled by framing the key questions.

Secondly, polls can have short shelf lives, predicting most effectively by the time they are made public what we are going to do yesterday.

Polls conducted online are arguably less effective than those conducted by telephone. Telephone interviews are better even if we suffer too much unwelcome interference in our households by folks trying to sell us something, despite the fact not all Canadians (especially those who rely totally on their cell phones) have listed numbers and the suspicion that not all of us tell the truth.

Finally, polling firms are typically employed by newspapers and other media firms, often on a continuing contract. While professional integrity should mitigate this, I worry the temptation to render polls that support the editorial line of the media firm for which one works is too strong a temptation.

The difference between the good, bad and the ugly in polling is found in the sophistication, insight and neutrality of those composing the questions and gathering data. As some readers know, I trust Nik Nannos's work the most, as he seems to have discovered a formula for getting it right. (His most recent poll, for the record, has us, as of Aug 20-27, at LP 35 per cent; CP 33 per cent; NDP 17 per cent; BQ eight per cent; and GP seven per cent.)

A mere vote is a modest political resource, but not one to be squandered. I like to believe I've gathered the best information possible, trying to make my vote count as much as it can. Even if I sometimes get it wrong.