Nate Silver once was a respected mathematical analyst. His baseball-related work, such as that at Baseball Prospectus and on PECOTA, showed that he has the ability to make solid, reasoned arguments using mathematical tools.
But now, he's flushed his own reputation into the toilet with his campaign against Strategic Vision. The pretend math, and lack of serious analysis and justification, in his series of posts against the company is so bad, I expect him any day now to start ranting about how he hasn't seen a given poll, but he still thinks that Obama has the consistency to pull it out just like the Reds used to. Nate Silver has become the Joe Morgan of politics.
The plain truth is, much like a Joe Morgan broadcast, the Nate Silver articles leave one knowing nothing he didn't know to begin with. Take the original piece. Here, Silver's analysis boils down to this:
- Manufacture two sets of data using a methodology with no justification given. Why strip out everything but Democrats and Republicans? Either the polls are doctored or they aren't.
- Make pretty pictures.
- Eyeball the pictures.
- Scream that they aren't consistent enough, not like his old Reds teams, so they must be FRAUDS!
He attempts to provide a thin veneer of justification for his work by citing Benford's Law. However that's completely ridiculous, as Benford's Law applies to a) early digits of numbers in data sets spanning b) many orders of magnitude c) smoothly. Silver's work covers a) last digits of numbers in data sets spanning b) a range of about 30-60 c) bunched together around 50 because polls are more likely to be taken in close races. Even mentioning Benford's Law in this context by most people would show a fundamental lack of understanding, much like Joe Morgan and other analysts when they use Wins to praise pitchers and RBIs to praise batters.
However Nate Silver knows better. He's not the Joe Morgan of politics. He's more like Joe's old teammate Pete Rose here. Rose was great as a player, and a fraud as a manager, while Silver was great as a baseball analyst and has now become a fraud as a political analyst.
Compound that Benford's Law deception with the use of a picture of a correlated data set. He asserts out of thin air that the distribution of last digits should be uniform. How is this the case? We all know that close races are polled more often than blowouts, and Silver in particular should, since he spent the whole last Presidential election watching some states come in more frequently than others. All it would take for Strategic Vision to get a distribution like he shows, is to have a bunch of polls that show something like, oh, R 48 D 49 Other 1 Undecided 2. But we don't see that because, guess what, Silver stripped out the Others and Undecideds!
He hasn't backed down since that original article, either. The willful mathematical incompetence continues in a followup article, in which he exhibits the same mathematical ham-handedness:
- Asserts a distribution of last digits without justification
- Invokes Benford's Law in a way only a mathematical illiterate could
- Heavily relies on charts and not established statistical tests to draw conclusions about data sets.
In another followup, Silver attempts to refute a specific poll by... making up his own simulated poll results. And apparently distance from Atlanta, GA has a proven correlation with fraud, or something. Perhaps Coca Cola makes you better at math?
Nate Silver once had a reputation. Even if his political commentary was left-leaning, his math could be trusted. Not anymore. He has shredded that solid reputation to become a political mercenary, attacking a firm's integrity for partisan political reasons. I'm sure he won't even notice that Republicans and independents no longer have any reason to trust him, with all the rabid cheering he'll get from the radical left. But deep down, I wonder if he felt it when he shed that last bit of integrity to get page views.
Is Strategic Vision making up poll results? I have no idea, but that's just it: Nate Silver's rabid crusade won't tell me that. Actual, mathematically-sound analysis would have to be done to draw any conclusions about that. Silver has done none, because Silver is only interested in scoring political points for the Democrats, rather than using math to ferret out truth.