Polygraph‘s Hanah Anderson and Matt Daniels undertook a massive analysis of the dialogue of approximately 2,000 films, counting those characters who spoke at least 100 words. With the data, they’ve producing a series of visuals that powerfully illustrate male dominance in the American film industry.
We’ve seen data like this before and it tells the same disturbing story: across the industry, whatever the sub-genre, men and their voices take center stage.
They have some other nice insights, too, like the silencing of women as they get older and the enhancing of men’s older voices.
But knowledge is power. My favorite thing about this project is that it enables any of us — absolutely anyone — to look up the gender imbalance in dialogue in any of those 2,000 movies. This means that you can know ahead of time how well women’s and men’s voices are represented and decide whether to watch. The dialogue in Adaptation, for example, is 70% male; Good Will Hunting, 85% male; The Revenant, 100% male.
We could even let the site choose the movies for us. Anderson and Daniels include a convenient dot graph that spans the breadth of inclusion, with each dot representing a movie. You can just click on the distribution that appeals to you and choose a movie from there. Clueless, Gosford Park, and The Wizard of Oz all come in at a perfect 50/50 split. Or, you can select a decade, genre, and gender balance and get suggestions.
Polygraph has enabled us to put our money where our principles are. If enough of us decide that we won’t buy any movie that tilts too far male, it would put pressure on filmmakers to make movies that better reflected real life. This data makes it possible to do just that.
Lisa Wade, PhD is an Associate Professor at Tulane University. She is the author of American Hookup, a book about college sexual culture; a textbook about gender; and a forthcoming introductory text: Terrible Magnificent Sociology. You can follow her on Twitter and Instagram.
Comments 5
Jamie Riehl — May 12, 2016
Very cool tool. One observation- it seems that a larger number of the very-high female dialog scripts are marked as such in the title (For Colored Girls, 3 Women, Boys on the Side etc.) Which makes sense, given that such films are note-worthy by their rarity - because it's an unusual product, the fact has to be advertised to attract the audience.
Umlud — May 17, 2016
It's a really interesting tool, but it would be interesting to see the impact that "screen time" (measured as number of words) would have on gender bias. From a quick-and-dirty analysis of the data, on average females spoke 28% of the dialogue.
I extended this bit of statistical procrastination a bit further and built a (very simple) regression model based on the total number of characters in a film, based on the total number of words, and then used the resulting linear equation to model what the number of female characters should be in a film, given the number of female-spoken words. NOTE that this assessment ONLY indicates the DISTRIBUTION of how the number of words in a script should indicate the number of actors in the film. IF the overall ratio of actors/words translates to female actors/female character words, then the implication is that overrepresentation of men is a simple case of numbers. IF, however, the ratio DOES NOT translate to female actors/female character words, then the implication is that - even with the lower number of words in a script - there is an "over-representation" of female actors (if the model under-predicts the actual number), and an "under-representation" of female actors (if the model over-predicts the actual number).
The MODELED mean number female actors, given the female word count is 5.25 actors. NOTE that this model assumes that the number of female actors speaking the number of words for female script can basically be modeled by the overall actors/words model.
The ACTUAL mean number of female actors is 3.45. NOTE that this indicates that there are - on average - fewer female actors speaking a greater number of words than the overall conditions imply.
This result seems to imply two immediate things to me. The first is that films simply cast fewer female actors than the number of words would imply OUGHT to be there. The second is that the number of female parts in a film (given the ENTIRE script) may be closer to the overall actors/words relationship, but that more female roles have less than 100 words (which is the cutoff for the study that generated the data).
Sorry for not posting graphics; the interface challenges between Disqus and SAS are something that I didn't want to begin to confront. Needless to say, if you want to do the calculations yourself, head over to the Polygraph website and download the data to have a play-around yourself.
Umlud — May 17, 2016
I don't know about the absoluteness of this statement:
"...whatever the sub-genre, men and their voices take center stage."
At the genre level, there is no doubt in the veracity of this statement. However, at the sub-genre level? I dunno. I suppose it depends on how one defines the "sub-" part.
Nieuwsronde: de goed nieuws editie | De Zesde Clan — June 17, 2016
[…] een film kijken en daarbij NIET overvallen worden door een mannelijke monocultuur? Vrees niet, de scenario database helpt je om films te vinden die hun vrouwelijke personages serieus nemen. Tik genre en gewenste genderverhouding in, en de […]