Cooking Books
Posted by Zach Wendling on 27 January 2008 | 4 responses
Here’s a clever site that’s been making the rounds: Books That Make You Dumb. The gist is that the creators downloaded the top ten books at every college listed on Facebook, then correlated the books with the average SAT scores at the colleges at which they are popular. The results, sorted by genre, are plotted here.
Really, this probably doesn’t tell us much about the books or their readers, but there are some interesting observations.
- The top titles (scoring roughly more than 1150):
Lolita
100 Years of Solitude
Crime and Punishment
Freakonomics
Catch 22
Atlas Shrugged
The Alchemist
Cat’s Cradle
Ender’s Game - Lolita is an outlier within the ‘Erotica’ genre (much dispute, apparently, over that classification); otherwise that genre scrapes the bottom of the barrel. Slightly above that lies the ‘African-American’ genre. All of those titles scored lower than “I Don’t Read.”
- While Atlas Shrugged scores high, the kind of committed Objectivists who would delve deeper into Ayn Rand’s works will find Anthem quite a bit lower, scoring mid-range among ‘Dystopian’ literature.
- The Da Vinci Code scores well, and those who went back to read Angels and Demons score even higher. Oddly, those who expressed a preference for “Dan Brown” scored lower than DVC.
- Interesting distinction between those who list The Bible (~1050) and those who list The Holy Bible (~925), the only title to score lower than The Purpose Driven Life (no favorite of mine) in ‘Religion.’ Those hard-working Mormons lead the pack in that genre.
- As noted elsewhere, Crime and Punishment far outranks Anna Karenina. Also noted elsewhere, some derision for tagging Tuesdays with Morrie and The Five People You Meet in Heaven as ‘Philosophy.’
4 Responses to “Cooking Books”
When I went off to college, I was still reading the likes of Robert Ludlum and Lewis Grizzard, but I managed to get a 1440 SAT score. Conclusion: no correlation!
When I went off to college, I was still reading the likes of Robert Ludlum and Lewis Grizzard, but I managed to get a 1440 SAT score. Conclusion: no correlation!
Conclusion: You are either a “comforting” outlier, or those books are culturally recomended by those particular schools.
If they are trying to make you draw the conclusion “reading book X means you are probably smart” then they screwed the math. They need to go read about Bayesian statistics.
Bayes’s law in this case:
P(smart|book) =
P(smart)*P(book|smart)
———————-
P(book)
In other words, they gave you (very approximate) P(book|smart) and didn’t tell you anything about P(smart) and P(book), which clearly would affect the outcome greatly.
If that equation is a bit too much for you, then here is my objection in words:
You would expect that the probability that someone is smart after learning that they read a certain book would be affected by three things.
The first thing is the the probability that any given person is smart.
The second thing is the probability that any given person read the book.
The third thing is the probability that any person who we know already is smart read the book.
They only gave us a dumb estimate for the third item. The first two are unknowns, and at least the second one is going to be difficult to calculate.
All of it means that you can’t know if someone is smart, given they have read a certain book unless you know those three things.