Polls Apart31 October 2013
Political opinion polls: virtuous insight into the mind of the electorate or obsession that undermines the political process? In 1940, George Gallup—father of the modern political poll—declared them the tool to “take the pulse of democracy”. Though it would seem that now we have hooked our political process up to life support, scrutinising every minor fluctuation as a symptom of some type of malaise.
Newspoll’s final survey before the federal election showed the Liberals continued their firm lead over Labor, whose primary vote slipped to 33%. Notably, Tony Abbott pulled ahead of Kevin Rudd as preferred PM for the first time. The Opposition Leader was skeptical of the boost, however, telling reporters, “I do not believe these polls.”
Abbott’s response, likely a strategic move to stave off voter apathy, is common political rhetoric when it comes to opinion polls. The ABC’s James Glenday warns, just like an actor who says that they don’t read their reviews, “never believe a politician who says they don’t care about opinion polls”.
While it has become politically chic to undermine polls, Australia’s major pollsters, including Newspoll, Galaxy, Fairfax’s Nielsen and Roy Morgan, are able to mirror election day results very closely. Andrew Catsaras, pollster for ABC’s Insiders program notes that “these are all professional research agencies. Those who have polled in past elections have a very good track record”.
Catsaras’ final ‘Poll of Polls’, an aggregate of the findings of the major polls, had the Coalition’s primary vote sitting at 45.5% versus the ALP’s 34.8%. The actual result on Election Day? Liberal: 45.4% and Labor: 33%, with a surprise shift away from the Greens (8.4%) to independents and minor party members (12.4%). A similar aggregation technique was used by the New York Times’ Nate Silver to accurately predict the winner in all 50 states during the 2012 US presidential election.
Australia’s leading polls have a margin of error of just three per cent. In the case of Newspoll’s result that 43% of voters prefer Abbott as prime minister, it is likely that if you were to survey the entire population the actual result would be between 40-46%.
Those who obsessively scour through the results of every poll are probably wasting their time though. Most changes are just statistical variations that fall within the margin or error. Perhaps a few more Liberal voters were surveyed this time or vice versa.
This accuracy doesn’t mean that polls don’t warrant close scrutiny. Polls are a key way for public opinion to reach our policy makers. Polls are not just about how we intend to vote, but also the diverse range of issues we care about, the public services we need or how we want to position ourselves on the global stage.
Polling companies are increasingly facing the challenge of declining response rates through their traditional channels like telephone, mail and face-to-face. Particular groups in society, including young people and non-English speakers, are much less likely to respond to polls than older people with home phones and time to talk. In order to combat this, companies weight results to make up for a lack of response from certain groups. In practice, this means that the response of one young person has more weight, but also that there is a lack of diversity of opinion in poll responses for these groups.
Polls are now central to policy discussion and politicians are increasingly reliant on them (and similar practices like focus groups) for insight into more complex and nuanced questions. Andrew Catsaras believes this intense focus is not necessarily positive, as the “press and politicians obsess over polls, years out from an election and that distorts [the] political process”. At the same time, if politicians aren’t getting a full picture of what the electorate wants then underrepresented groups in society will be at a disadvantage. What can be done? Perhaps the best thing is to spare a couple of minutes the next time a pollster calls.