Polls polls polls polls SHUTUPSHUTUPSHUTUP

Oh my God, shut up. Please, I beg of you all, just shut up. There are still 200 days until the election. If you keep up like this, one or both of us may die. 

It being a weekday, there has of course been another poll released. This one has, like a few before it, shown the Prime Minister and her party in pretty dismal electoral shape. So, that's today's column inches dealt with then. Because we love polls. We really do. In some ways, they're all we really have patience for in the current political age, the equivalent of catching up on a 5-day Test match by glancing at the scorecard for a couple of seconds. To get a sense of how insane the current situation is, today's Newspoll was pre-empted by the almost unprecedented move of news.com.au running a story about the implications of a poll that hadn't even been released yet. I mean, if that isn't a symptom of a system fast in danger of eating itself, vomiting and then eating the vomit, I don't know what is.

[Cricket may be a misleading analogy though. The last thing we need is to further reduce our political process down into a series of discrete contests between two otherwise anonymous teams. But it's hard to do much else these days unless you really have the time, patience and care to burrow beyond the poll-laden front pages of our political dialogue. I barely do, and understanding what's going on is essentially part of my job.]

In Australia we have five primary polling companies: News Ltd's Newspoll, Fairfax's Neilsen, Crikey's Essential and the unaffiliated Roy Morgan and Galaxy. Of those, Essential is the only one that significantly varies its methodologies, using rolling averages to try and give a result less affected by temporary spikes in public sentiment. It's also the only one that is generally never mentioned in any other paper. Still, even it is afflicted by the problems inherent in all polling: sample size, sample breadth and sample representativeness. On a poll with a margin of error of 3%, the results could change by 6% and the underlying trend potentially would not have changed at all. Such a large swing would suggest some movement, but there's no proof that it isn't simply the random distortion of a necessarily flawed survey. Long story short, individual polls are essentially a crapshoot where the likelihood of the final number actually being correct is little more than you might get from an actual round of craps.

Polling companies know this, of course, as do all the people that hyperventilate over the results as released. But polling companies have an interest in keeping their product in demand, and these writers have column inches to fill. So, the headlines come. Labor Headed for Poll Wipeout. Julia Gillard Deader Than Gaddafi. Rudd is Actually Jesus says New Poll. ALP Have Less Chance of Winning Than Moon Does of Exploding. Tony Abbott is The Best Because He is Handsome. Rudd Needs to Release More Harlem Shake Videos Say Poll.

The articles themselves continue in a similar vein, weaving a confection of assumption and insinuation which gives the impression that the poll contains biblical truth. Rather than the reality of it being as reliable as, say, the Bible. This is usually followed by an explanation of what it might mean come election day were these magical numbers to somehow be exactly repeated. Then there's always a quote or two from the losing party, cliché-ridden interviews that have about as much intellectual value as talking to a footy player thirty seconds after his team has won a game.

JOURNALIST: So, the polls say you're going to lose. Are you going to lose?

POLITICIAN: No, we're not going to lose. 

Whoa! Shit! Did they just say they hadn't all decided to simply commit hara kiri because the numbers were bad? Insight! [In case you want to accuse me of hyperbole, today Craig Emerson was heard to remark "We'll win government at the next election because we will be very competitive"] But what does the winning party have to say?

JOURNALIST: So, the polls say you're going to win. Are you going to win?

POLITICIAN: Yes, we're going to win.

No way!

Then it finishes with a cheery portent of imminent destruction that would not be out of place in your average Doomsday cult. Neat, tidy, done.

Have we learnt nothing from Nate Silver? Is there anyone out there actually looking at this in a sensible, non-hyperbolic fashion? Sifting the signal from the noise? Trying to slow down and make sense of the cycle of data spikes so that the whole thing doesn't so closely resemble some banned Eastern European rollercoaster called The Spinebreaker? [Crikey's Pollytics is a rare and welcome exception to this absence. His piece on polling trends and horserace analysis is kind of like this but with actual evidence and graphs and without the GIFs.]

The problem with polls is that they're as ambiguous and complex as any other part of the political process, but because of the rounded numbers they end up with they give the impression of certainty (and are reported as such), thus creating self-fulfilling prophecies from the thinnest of airs. This isn't to say that political polling is not without merit (nor that the most recent batches of them doesn't suggest a major, major problem for Gillard, et al), but somewhere along the line this flawed, incomplete beast came to stand in for the thing it purports to survey. The minion has enslaved the master, and Igor doesn't know shit about running a country. 

As it stands, polls have become political event without the need for an actual event, perhaps our most blatant example of the media and politicians creating the stories they want/need from a dry, unrelatable political process. But how deadening all of this is. How futile and empty. How confused it makes democracy even in the process of pretending to most closely resemble it. How base it makes the machinations and calculations. How harmful it is to political vision. How damaging it is to the good reputation of statistics. How difficult it is to comprehend that we are still 200 days of cookie cutter poll analysis away from an actual electoral result. 200 full calendar days. Serenity now.

Although, with all that said, I feel like it should be noted that as of today Tony Abbott has now been declared Australia's preferred Prime Minister in multiple polls in an election year. It could well be a statistically significant trend. Tony Abbott. I daresay the Mayans were out by a few months.

profile of lukeryan