These 8 Charts Highlight the Lack of Diversity in Movies
From the #OscarsSoWhite movement to the halls of academia, discussions of diversity in film have become increasingly salient and reliant on data. As data scientists at Pilot, we hope to dive deeper beyond basic summary statistics and introduce more rigorous and data-driven analysis to the conversation. Over the past year, our team analyzed thousands of films to identify the following industry trends across the intersections of gender, age, and race in the acting and directing professions.
1. Women Receive Fewer Acting Roles After Age 30 and Have 60% Less Opportunities Than Men
The below graph depicts the total number of major roles in our dataset at every possible age. The graph shows that men (green) and women (blue) have roughly the same number of starring roles until their late 20s. At around age 30, women hit an opportunity ceiling. Meanwhile, men of the same age starred in 60% more roles. Men ultimately reached a higher ceiling at the age of 40.
Although the number of roles for both genders declined after their respective ceilings, 50-year-old actors still had the same number of total roles (about 150) as 30-year-old women at the height of their careers. From ages 50 to 80, men had more than twice as many major roles as women of the same age.
2. Women Held 34% of Acting Roles and 6.4% of Directing Jobs
Women accounted for 34% of major acting roles in our dataset, about half that of men. On the other side of the camera, women directed a mere 6.4% of films during this 17-year period.
The two graphs below depict mean and median box office by gender. Mean box office values are higher than medians because blockbusters cause high standard deviation — they account for the majority of film revenue despite being a minority of films.
In these graphs, circles are sized proportionally to the percentage of total roles associated with each gender. Mouse over each graph to see exact values. Superimposed circles show the largest group (Male) for comparison. You can click on the mean or median labels to toggle displaying each dataset. Graphs are interactive for desktops, tablets, and mobile devices and can be embedded using the link below.
Although the mean and median box office of films directed by women are lower than those of films directed by men, these differences do not exist between actors and actresses, suggesting that female directors in particular are afforded less opportunities to work on high-grossing blockbusters.
3. Over 80% of Actors & Directors Are White
83% of major roles in our dataset were played by white actors and actresses. Black talent accounted for a mere 9%, while every other category made up less than 10% combined. Meanwhile, almost 90% of films featured white directors, with the other 10% split between the remaining groups.
In contrast, the 2010 United States census reported that 72.4% of Americans self-identified as White (including Middle Eastern), 12.6% as Black, 4.8% as Asian, 6.2% as Other, and 2.9% as Mixed. Additionally, 16.3% of respondents also identified as Hispanic and Latinx of any race.
The following two charts depict mean and median box office by race. Similar to the above graphs, circles are proportionally sized to the percentage of total roles associated with each race. Mouse over the graph to see exact values. Superimposed circles show the largest group (White) for comparison. You can click on the mean or median labels to toggle displaying each dataset. Graphs are interactive for desktops, tablets, and mobile devices and can be embedded using the link below.
4. Minorities Are Most Often Cast Alongside Whites and Their Own Race
Are minorities likely to be typecast in racially homogeneous films featuring minority-specific narratives? Put another way, does the landscape of films with diverse casts more resemble Straight Outta Compton or Furious 7?
We calculated a "typecast" metric by identifying the total number of joint appearances between actors of each racial category. The graph below depicts these collaboration frequencies. Mouse over each bar to view the exact value. You can click on any square in the legend or any of the bars to toggle highlighting the data for a particular race. Graphs are interactive for desktops, tablets, and mobile devices and can be embedded using the link below.
For every minority racial category, actors were most likely to be cast alongside white actors followed by their own race. (The only exception is Mixed, where the second-most frequent category is Black.)
Accounting for the fact that 83% of roles in our dataset were played by white actors, this suggests a very homogeneous brand of diversity — one in which minorities usually appear in films alongside individuals of their own race but not other minority groups.
5. Who Has the Most Influence in Hollywood?
Using graph theory, a clique in our dataset can be defined as a group of 3 or more distinct people who have all worked together on a film. When we visualize the massive network of our dataset — almost 6,000 actors and directors who are connected by 2,500 films — we can quantify the relationships between individuals in a similar way to the popular game Six Degrees of Kevin Bacon.
The more cliques an individual belongs to, the more connected they are to different groups of people, and the more likely it is that they've had a longer film career.
The two graphs below depict the mean number of cliques by race and gender. Mouse over each row to see exact values. Graphs are interactive for desktops, tablets, and mobile devices and can be embedded using the link below.
Marginal differences in clique size between genders or across race may not be significant. However, actors and actresses in the Asian, Nonwhite Latinx, Nonwhite Middle Eastern, and Other categories have clique counts that are about twice as small as those of individuals in the Black, Mixed, and White categories. Moreover, male directors appear to be more than twice as well-connected than female directors across almost every racial category.
With an emphasis on graph theory and network analysis, we hope that our contribution to this discussion provides greater clarity for everyone and helps to build a shared foundation of data surrounding film diversity.
Especially as the end of awards season approaches, it is incumbent upon all of us in the film industry to continue this conversation and convert data-driven analysis into actionable insights.
— The Pilot Team
Notes on Methodology
- We analyzed about 2,500 films from 2000-2016 that received a wide or limited theatrical release in the U.S. and earned at least $1M domestically. All revenues were calculated using domestic box office only.
- We only considered top-billed actors and actresses with major speaking roles — in total, approximately 4,300 unique actors and 1,400 directors.
- We recognize that classifications of race are often arbitrary and contentiously debated. This analysis uses 7 major categories: White, Black, Nonwhite Latinx, Asian, Nonwhite Middle Eastern, Mixed, and Other. Similarly, this analysis uses a binary classification of gender: Male or Female.
- In the future, we hope to conduct additional analyses of other dimensions, such as whether an individual identifies as LGBTQ.