This blog is still alive, just in semi-hibernation. When I want to write something longer than a tweet about something other than math or sci-fi, here is where I'll write it.
Friday, August 3, 2012
Unable to agree on reality.
As regular readers know, I'm being obsessive about polling data, just like I was in 2008.
For anyone considering an intervention, I ask you to reconsider. I'm 56, and I think we know what stuff I obsess about by now. Do not expect it to change.
Two companies, Gallup and Rasmussen, are putting out daily tracking polls for the national presidential numbers. For anyone with memory loss, let me say that the national numbers are crap because the electoral vote count is what matters. Go to Wikipedia and look up the 2000 election if you are unclear on the concept.
Although we have two companies engaging in serious numerical work that is about as useful as doing an astronomically accurate horoscope, it does have some value in that it lets us compare the data gathering methods of the two companies. If you are an old timer like me, you might still think Gallup is the gold standard of polling companies, but they lag far behind new companies like Rasmussen and PPP, and even produce less polls than some universities, notably Quinnipiac.
What we can see is that Rasmussen gives Mitt Romney a lot more credit than Gallup does. In May, Gallup thought the race about even, while Rasmussen gave the nod to Romney. In June, Gallup said the tide was moving in Obama's favor slightly, while Rasmussen thought it was going the other way significantly. In July, the difference increased dramatically, with Gallup showing the tide turning strongly for Obama while Rasmussen thought Romney might have slipped slightly but was solidly ahead.
Again, let me say that the national polls are meaningless. More than that, if we look at Rasmussen's polls in the battleground states, they still show an electoral college advantage for Obama. If we look at the odds on Intrade, the betting site has Obama's odds at about 58% to 42%. If my Confidence of Victory numbers are anywhere close to right, Obama's actually over a 90% favorite right now, so betting him at these numbers should be positive expentancy. (Before any benighted commenter wants to talk about "the wisdom of the markets", Intrade was skewed in 2008 by one gambling fool convinced that McCain would shock the world.)
The polls have been all over the place this year. My best advice is to look at the median, ignoring the polls that are rosiest for either candidate. This is the first presidential election post-Citizens United, and no one knows how the needle will move after Labor Day when hundreds of millions of dollars will be spent on ads, possibly even adding up to more than a billion when both sides are tallied. Still, the most reliable numbers say Romney is trailing and needs a big boost, possibly a miraculous boost to win. He may have a better chance to win than Rafalca has to medal (still in the hunt but 30th of 32 qualifying horses after the preliminary), but I wouldn't put money on either of them right now.