Jada Pinkett Smith and Spike Lee have been lambasting the Oscars on their lack of diversity. It's hard to blame their reaction.
Their focus was on four categories: best lead and supporting actors for both men and women. There are five nominees within each of these four categories. This year, all 20 of those nominees were white. And the same was true last year.
Is this streak of all-white nominees that bizarre? To answer that question we have to determine what the probability is of all the nominees being white--just by chance.
We weren't able to get a breakdown of race within gender, but we were able to get figures on race in film. The percentage of non-white lead actors in 2011 was 10.5%, according to UCLA's Bunche Center 2014 Hollywood Diversity Report . According to another report put out by the Media, Diversity, & Social Change Initiative , the percentage of all actors that were nonwhite in 2014 was 26.9%. We'll use that last figure for the supporting actor category. For comparison, nonwhites made up 22.6% of the US population in the most recent Census estimate.
We can use this film information to calculate the probability that not a single nonwhite actor would be nominated in any of the four categories. The odds of that happening this year were a mere 1.43%. That means that if you randomly nominated actors within these respective categories you'd expect this all-white outcome to occur only about once every 70 years. The larger red flag was that the exact same thing happened last year, too.
How can we know, however, whether this was some super anomaly or part of a larger trend?
The 2014 Hollywood Diversity Report pointed out that between 2002 and 2012 that about 20% of nominees were nonwhite. A recent article in The Economist explored this further finding that while the numbers for nonwhites were especially gloomy before 2000, since then, nonwhites have gotten a more proportional share of nominations. Using data from Time Magazine indicates that since 2000, nonwhites have made up 15% of nominees. The last two two outlier years pulled down the average a bit.
So what about this 15% nonwhite nominees figure? Technically, it's still a little low, even when we factor in randomness. By calculating a 95% confidence interval, we'd expect the upper bound of nonwhite nominees to be 18.9%. This statistic means that if we kept sampling this range of nominees over time, we'd expect our estimate to capture the true percentage of nonwhite nominees 95% of the time. Because our upper bound estimate is lower than the proportion of nonwhites from either the film industry as a whole or the general population, this is why there may still be concern.
To be fair, part of this representation issue is outside the Academy's reach. Nonwhites are underrepresented in lead movie roles. Actor George Clooney has questionedwhether nonwhites are getting the same opportunities for higher quality films. And many large films still cast white actors for roles where a character's nonwhite race is inherent to the role, or their real-life counterpart was not white. These switched-race lead roles have been in big-budget movies like 21,The Lone Ranger,The Social Network, Prince of Persia, and Pay It Forward.
People's focus on the academy also isn't surprising given that only 7% of the Academy's voters are nonwhite. The Academy has admitted their own concerns about this in a recent statement. They promised to double their numbers of nonwhite and female voters by 2020. (Also, only one in four of their voters are female, and the average age is over 60.)
Fixing the Academy's voting demographics would be quite an ordeal. The Academy currently has over 6,000 voting members. Such a large existing membership makes it difficult to change its overall demographics. If the Academy wanted to give nonwhites the same makeup as the US population--which is actually an easier bar than the percentage of nonwhite actors--the Academy would have to induct 1,215 nonwhite voters consecutively before it accomplished its goal demographics. And this would assume all existing voters stayed how they are.
Another question is whether the Academy wants its nominees to be diverse--not just in race but in the type of nominees overall. A proportional voting method would produce this diverse result. It's unclear from the Academy's voting rules, however, whether it uses a proportional voting method for its actor nominations. (The Academy didn't respond to our inquiry.)
The Academy clearly does, however, use a proportional voting method for one of its other categories--Best Visual Effects. There, it uses a method called reweighted range voting (RRV), which was developed by one of our organization's co-founders at The Center for Election Science. RRV works by letting voters score all the candidates on a scale. The method then reweights voters' ballots after they successfully nominate a candidate.This reweighting gives other voters a chance to nominate someone they want. That way a majority bloc of voters doesn't control every last nominee.
Another quality of this proportional method--and proportional voting methods in general--is that the outcome becomes more proportional when you increase the number of winners, or in this case, nominees. The Oscars could accomplish more proportionality by increasing its nominees from five to eight, for instance.
And while they're at it, the Academy may as well use RRV's single-winner counterpart, score voting, to select the actual winner. There, voters score all the options on a scale, and the candidate with the highest score wins. This simple method works particularly well when there are many similar options--like Oscar nominees. Even simpler is approval voting where voters choose as many options as they want. This is in sharp contrast to having voters choose only one option, which is susceptible to a host of drawbacks, including similar options dividing the vote. For more evidence that a choose-one method is bad, look to the US government. That's how it does elections, too.
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).