Does McCain’s choice of Palin show him to be a reckless fool, another cowboy who can’t be trusted with the keys to the national car? Those who don’t want to see McCain win argue that it does, because Palin can hardly be expected to govern well. Those who do want to see him win *also* argue that it does, because it makes him less likely to win the election.

This second group is wrong. Down 4 points in remarkably steady polling over months, only two months to go before the election, McCain was like a boxer losing on the scorecard in the last round: all that’s left to do is throw haymakers. Mathematically, this is expressed in a shift in probability distribution:

What I have plotted here is two probability distributions for McCain’s vote share, one before (blue) and one after (red) the VP pick. Let’s simplify the race and discount third party candidates so that for this example, McCain needs 50% of the vote to win. The area under each curve to the right of 50% is the probability that McCain will exceed 50% of the vote and so win the election.

The blue curve has an expected value of 48%, which means the most likely outcome is that McCain loses by 4 points, which is about what we’d expect. The variance is 1 point, which means that there is very little chance of McCain doing worse than 46% (an eight point loss, still a near blowout) but also very little chance of doing better than 50%. I think this is a fair assessment of his campaign’s position before Palin. More importantly, I think this is a fair assessment of where the McCain camp thought they were.

Palin moved them to a probability distribution more like the red curve. The most likely outcome is now slightly worse, but because the variance was so tight before, the most likely outcome is now worse than was really possible before the pick. But the variance is also much higher, which is just to say that Palin might be the haymaker that lands. The shaded area under the curve, to the right of 50%, is much higher than it was before the pick.

This simulation, however rough, correlates nicely with three realistic points:

1. McCain’s expected vote share is lower.

2. McCain’s chance of winning is higher.

3. McCain’s chance of losing disastrously is now much higher; and his chance of losing disastrously is much higher than his chance of winning.

There is no way to know now whether this gamble will succeed; that’s why they call it a gamble. What is certain is that if it does McCain will be thought a political genius – if not, an imbecile. They will be wrong in either case: it’s simply the obvious play.

September 7, 2008 at 7:25 pm

[…] distant cousins in the ‘the Great’ family, but I must say I’m not convinced by Frederick’s analysis of the Palin pick. The basic insight is a good one: since there is relatively little effective difference between a […]