A tie that isn't quite what it seems, and a US election sideshow: pundit gut v nerd calculator.
We’re only a few days away from knowing who will lead the US for the next four years. (Probably – I haven’t forgotten 2000).
The contest is already a media success. It has had some interesting twists and turns, and as it approaches the finish it's close and exciting.
For pundits, it’s prediction time, and we’re treated to their expert opinions (also known as gut feelings) on who will be occupying The White House.
But this year we’ve seen more of a less cocksure bunch, wielding large sets of interesting numbers but still refusing to go all in. While traditional pundits still pretend that any cherrypicked poll swing is “news”, proper statisticians are adding them all together, carefully weighed by many parameters, thus creating a much more reliable meta-analysis.
These number crunchers, led by Nate Silver, have consistently given Obama the edge. It started out slim, grew comfortable as Romney struggled, then almost evaporated after the first debate. Much of it has been recovered since, mostly due to a smallish but tenacious lead in enough of the swing states. At time of writing, Silver’s Fivethirtyeight estimates Obama’s chance of victory at 79%.
The fact that Obama never dipped below a 50% chance, even when Romney led national polls (which right now look almost exactly even), has led Silver to be dubbed “Xanax for liberals”, and has recently brought down a slew of attacks on the head of the current king of stat gurus.
It’s not surprising: the GOP is desperate to avoid a return to the narrative of a losing Romney. Silver has been called an ideologue and a joke, his sexuality alluded to. His numbers from before the 2010 election are conveniently forgotten – those were more like cocaine for conservatives than sedatives for liberals.
Also, many of the attacks on Silver seem to be from people who don’t understand probability – interpreting a seventy-something per cent chance of winning as a shoo-in. It’s no such thing.
If Obama has a 79% chance of winning, more than one in five hypothetical elections would go to Romney. That’s far from a tie, but it's not a bad shot. If you put 79 blue and 21 red balls in a bucket and picked one out without looking, your mind wouldn’t be blown if it was a red one. (If it is, you’ve lived a rather sheltered life.) There’s plenty of reason for both campaigns to keep fighting tooth and claw.
Joe Scarborough of MSNBC doesn't get this, asserting that 50.1% was the best estimate – as both campaigns had told him so. They would of course; the Romney campaign wants to be seen as winners, the Obama campaign know they’ve got a slight edge, but are terrified of anyone in their ranks letting their guard down.
On Election Day 2008 the Obama campaign sent out a message through their organisation on the ground: “McCain is doing surprisingly well – get out there!” Adrenalin flowed through the veins of thousands of volunteers, ensuring the ground campaign worked at full capacity those last few hours. I’m sure they’ll repeat the trick, it would be stupid not to.
So at this point, both campaigns and the media have interests in the same narrative – it’s a tie.
Karl Rove is too smart to really believe it’s going very well, so when he claims Romney is poised to win in the Wall Street Journal, brandishing plenty of numbers, it’s to even out the predictions of Silver and co. (He knows that he can win of course, otherwise he’d have kept quiet, but calling him favourite is a favour, not a judgement.) The Obama campaign wouldn’t want their surrogates to declare likely victory like Rove, they already look like very slim favourites and are happy with that perception.
To be sure, pundits will still declare winners, because they see speculating as part of their job, and they like to bask in the glory of being right if they happen to be. Not that there should be much of that: if you declare you’re sure of the winner now, you might be right about the outcome, but that doesn’t make you right about the confidence. It would just be an unremarkably correct guess.
Silver’s numbers aren’t necessarily right, of course. It’s a complex landscape of polls and realities, and uncharted undercurrents or unexpected events can have unpredictable results. But I’ll trust a nerd over a pundit any day, especially when the bookies agree.
Based on the imperfect, but vast and informative data we have available, I think we can say this with certainty: an Obama victory is the most likely outcome, a Romney victory is entirely plausible.
At least that’s what my gut tells me.