In defence of polls
This election has proved something of a trial for opinion pollsters. Not only have there been an unprecedented number of polls — sometimes three in one day — but with the closeness of the race for much of the campaign there has also been a marked disparity of findings. On the same day, one poll will have one party in the lead and another in third, while another poll will have the reverse.
Once upon a time Canada used to ban polls in the final days of a campaign, before this was thrown out by the Supreme Court. Later, the blackout was shortened to election day, but this too was repealed. And with good reason: whatever anyone thinks of them (there’s probably a poll on that, too), opinion polls provide citizens with useful information — namely, what their fellow citizens are thinking.
To be sure, we could probably get by with fewer polls. We folks in the media tend to pay far too much attention to them — to the proverbial horse race, rather than the conversation the parties are supposed to be having with the voters and each other. And yet voters have every right to consult the polls as part of their deliberations.
Critics complain polls lead to strategic voting or “bandwagon effects,” and of course it’s true: sometimes they do. If that is how some voters choose to vote, rather than on a straight-up comparison of the parties’ policies and candidates, that is their privilege. In a democracy, it is not for others to say what are valid or invalid reasons for voting, least of all the state. That is the prerogative of the citizen, in the privacy of the voting booth and his conscience.
Strategic voting is an entirely legitimate exercise, if an unpleasant one: the voter is often presented with the dilemma of voting for a party he dislikes, rather than the party he likes, in order to prevent the party he detests from getting in. But that is a function of the electoral system, not the polls: first past the post is a “winner take all” system, in which only the leading candidate is elected in each riding. In such circumstances, polls are not just the messenger. They are a vital part of the message.
It would, then, be anti-democratic to withhold such information — not least since a ban on polls would not actually ban the taking of polls: only the publishing of them. Experience tells us the parties would continue to conduct surveys for their own purposes, as would others with the means to do so and an interest in the result; these would inevitably leak to the media and other well-connected insiders; rumours, some accurate, would spread online. The only ones left in the dark would be the ordinary voters.
Pollsters are doubtless finding the going tough these days. Many wouldbe respondents don’t answer their phones. People change their minds more often than they once did. Others simply lie about their intentions. Yet while individual polls can and do err, the pollsters’ track record, collectively, is much better than they are given credit for: we remember the occasional fiascos, not the many more where the pollsters got it right.
Polling is a serious social science, whose methods matter greatly to its professionals. There is no reason to think that it cannot be of service to the public. Certainly there is no alternative. People who sneer at the polls as inaccurate should say what numbers they think would be more accurate, or what methodology would produce them. “Get out and talk to the people in my neighbourhood” is not a valid sampling technique.
Still, polling data should be handled with care. National numbers typically matter less than regional. Sample size matters, as — to tread onto controversial turf — does methodology: online polls versus telephone, for instance. Voters are advised to consult a range of polls, with an eye to averages and trends instead of single data points. In doing so, they will gain a better sense of their fellow voters’ state of mind, and make a more informed choice at the ballot box.