Sunday, November 4, 2012

Book Review, Sunday, November 4, 2012: The Signal and the Noise

Madame L has been enjoying this book more than she thought possible but, because she has to return it to her library, she cannot give a complete review here. 

Also, annoyingly, Madame L is not able to turn down page corners and mark passages with underlinings and highlightings, because she wants to stay in the good graces of her local librarians. 

Therefore Madame L will be buying her own copy of this book, which she only wishes any of her college statistics teachers had had available instead of the stultifying and incomplete textbooks they foisted on her and her classmates.

Nate Silver is brilliant, not only at statistics, but at history and economics and baseball and politics and, well, everything else covered in this book. He introduces us to Bayesian theory, of which, in three semesters of statistics classes in college Madame L only heard, "That's for a more advanced course."

That's the brilliance of this book: It explains these concepts that college math and stats teachers act like are too difficult for us and which, Madame L suspects, were actually too difficult for them. 

So, what's the signal? It's the correct conclusion to be learned from a statistical model. 

And the noise? It's the information that tends to drown out the correct conclusion. 

And why do "so many predictions fail---but some don't"?  Duh. Because they're doing it wrong.

And here's the thing: Bayesian theory, in Nate Silver's hands, allows us to sort the signal from the noise. And that's why Nate Silver's blog is the main source Madame L goes to for the real information on the current state of the vote and the possible outcomes on the election.

For more about the book, please stay tuned for Madame L's more complete review, which she promises for shortly after the election, when she will have her own copy.

No comments: