New High Score

Analytics. Visualization. Data. Design. Marketing.

Posts Tagged ‘Matthew Effect

The Matthew Effect and The Oscars

leave a comment »

Very belatedly, I stumbled upon Anand Rajaraman’s post, Oscar Halo: Academy Awards and the Matthew Effect. I find it most interesting, for a few reasons.

1. I didn’t know that this sort of a thing had a name, and a fancy Bible-inspired one at that. To quote the Wikipedia entry,

The Matthew effect in sociology is the phenomenon that “the rich get richer and the poor get poorer”. Those who possess power and economic or social capital can leverage those resources to gain more power or capital. The Matthew effect results in a power law distribution of resources. The term was first coined by sociologist Robert K. Merton and takes its name from a line in the biblical Gospel of Matthew:

“For to all those who have, more will be given, and they will have an abundance; but from those who have nothing, even what they have will be taken away.” — Matthew 25:29, New Revised Standard Version.

Nice. I always thought it was called capitalism, or something 😉

2. Rajaraman finds that the Matthew Effect does seem to manifest itself in the scores for the Oscar nominations. He infers this from the fact that the distributions of N(k) (which denotes number of movies that got k nominations) follow a power law. Indeed it does, and the R2 of a power-law fitted curve, if it can be considered for so few observations, is pretty high. The chart below is off the data he’s put up on his site.

The Number of Films that were nominated for an Oscar

The Number of Films that were nominated for an Oscar

It’s interesting to note that the number of films at 7 are more than the ones at 6! Goes to show, that once you’re past the first 5 nominations, you can safely expect a couple more, probably. He takes 4 nominations as the cut-off for the ‘rich’ and ‘poor’ films classification, but I think 6 would be a better cut-off because that’s when the slope changes quite a bit. However, that might result in a very small sample size for the ‘rich’ movie group. Another option might be to break it into three groups, 1 nomination films, 2-5 nominations and 6+ nominations.

3. Next he looks at categories, and find that the ‘rich’ categories usually also get nominated for Best Picture, Best Director and Editing. On the other hand, Music and Special Effects clearly do not suffer from this bias. I’d like to look at the ‘Market Basket Analysis’ that he does and see if there any categories where a nomination might predict a nomination in another category. I intend writing him to see if he’s willing to share the data. Otherwise, I’ll have to learn Python to get it off the Academy website.

4. He then looks at wins, and tries to predict if a high number of nominations means a higher number of wins. He uses a system of ‘expected wins’ (plain probability) and ‘observed wins’ and calculates a ratio which he calls the ‘win boost’. A ‘win boost’ over 1 signifies a bias. As expected the ‘rich’ films have a strong ‘win boost’. As another Anand points out here, that’s not really a huge surprise. To quote this other Anand,

Winning an Oscar is all about being in the right place at the right time, so yes the Matthew Effect must dominate again. You have to find the right combination of Hollywood liberal guilt, Hollywood elitist condescension, and Hollywood self-preening and then make it work in your movie’s favor.


Anyhow, it’s interesting, the analysis he does, using Matthew Effect (I still can’t get over the fact that this had a name like this) on the Oscar nominations. In the end, he tries to answer his initial question (which was whether A R Rahman piggybacked on the success of the film to be nominated, or not).

The statistics on the Music category say that the Matthew effect likely did not help Mr Rahman in securing his nominations; but now that he has been nominated, his chances of winning are greatly boosted because he is associated with Slumdog’s 10 nominations.

Fair enough. Of course, we know now that he won.


Written by newhighscore

June 29, 2009 at 7:19 pm

Posted in Analytics

Tagged with ,