Tuesday, February 26, 2008

Tuesday Statisticz: Reverse Engineering the Oscars

As you all know by know, No Country for Old Men took home Best Picture at the Oscars on Sunday. Naturally, I was shocked. It's not even the highest rated of the nominees on imdb! That's why I was so sure that There Will Be Blood would win.

So, all of this got me thinking. If it does not vote according to the likes of the general populace, then which demographic does the Academy vote most alike? As you know, if you click on the actual number of votes next to the rating of a movie, it takes you to this page, which breaks down the ratings based on age, gender, and some other categories. I've charted the scores that each Best Picture nominee received according to their demographic. Click on the image to enlarge it.

Note that the rating scale ranges from 0 to 10, but all of these scores fall between 6 and 9.5, so I chose those numbers to make the differences easier to see. "M" refers to males, "F" refers to females, and neither refers to the combination. The IMDb staff refers to people who work at imdb (none of whom bothered to vote on Atonement), and the "top 1000" refers to the people who have voted on the most movies on imdb. The website never lets people know if they are in the top 1000, for all I know I could be, with my measly 250 or so votes.

The most interesting parts of this chart are the places where No Country for Old Men outperformed the field. These are the type of voters that the Oscar voting population acts most alike. As you can see, No Country for Old Men has a large edge (.5) in the Top 1000 voters, a medium-sized edge (.3) in males over 45, and a very small edge (.1) in non-US voters. We can essentially throw out the non-US votes because (at least for No Country for Old Men), they represent over half of the total votes, and because it is such a small edge over There Will Be Blood. One other point of note is that teenage girls do not seem to have much influence, as they loathed the Oscar winner, awarding it a 6.6 average, albeit with a low sample size of 301.

It is hard to say exactly who these top 1000 voters are, but it is a safe bet that if you have seen and voted on more than 750 movies or so, you will have become pretty pretentious about what kind of movies you like. Don't believe me? The top 1000 voters consistently kill frat-boyish comedies, like the 5.9 they gave to Old School, while they love pretentious comedies, like the 8.1 they gave to the (admittedly hilarious) Life of Brian. The rest of the voters gave the two movies comparable scores of 7.9 to 8.2.

So, which demographics vote most similar to the Academy? Old, pretentious men. I guess it's not such a bad country for them after all.