David Moscrop laments the role of opinion polls in shaping political events - and there's certainly reason for caution in presuming that immediate polls will have a lasting effect. But I'll argue that at least as politics are now covered, polls in fact serve as an important check on the tendency of campaign coverage to become completely detached from the views of the public.
After all, the same citizens whose votes determine the outcome of a campaign are generally expected to follow that campaign with varying levels of care through media intermediaries. And I discussed the problem with the direct impact of media here, as the most subtle of campaign narratives - whether or not they're generated deliberately or based on facts - can swing enough votes to change the outcome of an election.
It's certainly fair to point out that polls can help to shape those narratives. And the effect can run in both directions: just as they can offer a signal as to which parties are viably positioned to offer an alternative government (as happened for the Alberta NDP in this month's election), they can also offer a warning that the public may wish to reconsider a trend (as arguably happened for the Wildrose Party in 2012).
But that serves only as a side effect of polling rather than a primary purpose. At their core, polls are the basic evidence-based means of measuring public opinion - which seems like rather an important factor in talking about how the public will choose to be governed.
With that in mind, let's ask this question: what's the alternative to paying attention to polls as a means of assessing where parties stand, for the purpose of both strategic voting (as identified by Moscrop) and merely talking about the progress of a campaign?
While it's easy to find elections where polls have come under fire for failing to reflect outcomes (see the UK's recent vote or Alberta's 2012 election), omitting them from election coverage won't stop pundits from offering their own prognostications - which at best reflect an unstated set of personal biases and assumptions, and at worst are downright intended to shape the campaign narrative to favour one party. And this month's Alberta vote reflects an obvious example where the polls told a far more accurate story than the insiders.
Which leads to this question: would Alberta have been better served by not knowing that enough voters were receptive to an NDP government to create the potential for change?
Before answering "yes", it's well worth questioning the alternative of having campaign narratives shaped entirely by the people who are able to spin stories in the absence of evidence - and not at all by the public whose interest is intended to be served by the election.
Of course, one might validly point out that we'd be better off with a radically different form of campaign coverage which focuses far more party platforms and values, and far less on spin from all directions. But until we've taken some giant leaps in that direction, we're best off treating polling as a check to test whether narratives match public opinion - not as a problem to be eradicated in favour of even more air time for evidence-free speculation.