While we are in the heart of awards season, I wanted to do something about the Oscars for Cinemath, our semi-regular feature that combines the wonder of movies with the tedium of mathematical analysis. This edition is inspired by something Eric D. Snider wrote in his writeup on Marty, the shortest Best Picture in Oscar history:
“Don’t expect that record to be broken anytime soon, either. Oscar winners are gettin’ longer, not shorter.”
That got me curious. That sounds reasonable, but could it also be statistically true? I went over to IMDB to check out the runtimes of Oscar Best Picture nominees from 1928-2010—while I was there, I grabbed data on the genres and public ratings to see what it takes to win Hollywood’s top honor. Hit the jump for the analysis.
All data on runtimes, genres, and ratings comes from IMDB. Major props to the user who assembled all 485 Best Picture nominees on the same list. It was here, but has unfortunately been taken down since I exported it.
We’ll start off the claim that Best Pictures are getting longer. (I understand Snider was not writing literally, so I am not trying to prove him right or wrong—but it works as a research question.) I did not do any rigorous statistical testing, but it is apparent in the graph below that there is no evidence of this effect. There’s Marty (90 minutes) on the low end, Gone with the Wind (226 minutes) on the high end, and everything fluctuates wildly but consistently in between.
I found that the majority of the movies I watch are between 90-120 minutes, but Best Pictures tend to be much longer:
- When separated by decade, the 1990s had the longest Best Pictures with an average 154.5 minute runtime (median: 152 minutes).
- If you look at the 10-year moving average, the period from 1956-1965 had the longest Best Pictures with an average 162.8 minute runtime (median: 165.5 minutes). That speaks to just how lengthy Best Pictures tend to be.
- Nearly 70% of all Best Pictures winners are longer than 2 hours.
- Nearly 80% of all Best Picture winners after 1960 are longer than 2 hours.
The histogram below breaks down the runtimes of the Best Picture winners.
Notably, Oscar voters like to give the award to the longest movie nominated. For each set of nominees, I ranked the movies by runtime. The longest nominee has won Best Picture more than 40% of the time, as seen in the histogram below, which breaks down the rank of Best Picture runtimes among the nominees.
I don’t believe longer movies are inherently better. But they may have a competitive advantage at the Oscars, in the sense that longer movies feel more “important.” One can see the same effect when classifying the Best Picture winners by genre.
IMDB has a genre section on each movie page, which allows for a movie to be identified with one or more genre keywords. It is subject to certain biases, so I am hesitant to conclude too much here. But I like the table below that shows how often a genre keyword was present on the IMDB page of a Best Picture nominee or winner. The percentage columns show the percentage of Best Picture nominees/winners with a certain genre. So unsurprisingly, drama is a listed genre for 85% of nominees and 89% of winners. In comparison, the sci-fi genre has produced 6 nominees and 0 winners in eight decades.
The interesting part to me is the difference column, which shows the percentage difference between the genre’s presence among nominees vs. winners. A positive difference suggests the genre has a stronger association with winning once nominated; a negative difference suggests a losing genre. Look at the difference between war movies and comedies. I imagine that comedy is the second most common genre keyword on IMDB after drama. However, comedies have less than 1/4 of the presence dramas do among the nominees, and the comedies that do get nominated don’t win very often. There are relatively few war movies (it is a specific genre), but Oscar voters love them: 1/5 of our best pictures are war movies!
Comedies are light. War movies are heavy. War movies get the Oscar. Perhaps more evidence of the Importance Effect.
As expected, the public IMDB rating is generally high for Best Picture winners. A few notes on the spectrum:
- The lowest-rated film is 1931′s Cimarron at 6.1.
- There are only 7 films with lower than a 7.0 rating—4 of these won the award before 1937. Maybe Oscar voters were still working out the kinks, or perhaps there is too much distance between the tastes of the Depression era and the internet age.
- The lowest-rated postwar film is 1952′s The Greatest Show on Earth at 6.7.
- Of course, the highest rated Best Picture is The Godfather 9.2.
The graph and histogram below show the breakdown of IMDB ratings on Best Pictures.
As with the runtimes, I ranked each set of nominees by IMDB rating. The highest-rated nominee won Best Picture nearly 40% of the time. The Best Picture was one of the top 2 highest-rated nominees about 60% of the time. The distribution of the rank of the Best Picture winners by IMDB ratings is in the histogram below.
Of course, the IMDB rating is not independent from the Oscar voting. A Best Picture win solidifies a certain status in the film community, which could very well lead to a bump in the IMDB rating. The point is, no matter the reason, IMDB users and Oscar voters are generally in agreement.
If you would like to win the Best Picture, be important. Be the longest and highest-rated nominee. Also, tell a story about war, and don’t be funny. So best of luck, War Horse.