Are actors from minority backgrounds equally represented in movies? In this section, we investigate the importance of minority actors’ roles within movies. Even if the overall trend with the representation of minority actors increases within the movie industry, it is not clear whether or not these actors are being cast for significant roles.
To analyze this, we will see how many times each actor’s role is referenced within the movie’s plot summary, and therein assign them an importance score by calculating the percentage of times they mentioned in the summary vs the total number of times any character is mentioned in the summary. To accomplish this, the summary is first processed by a coreference resolution library so that different references, such as pronouns, to the same character in the text are taken into account. Afterward, their score is given based on the share they take compared to other characters.
In order to see how different ethnicities filled significant roles changed over time we then grouped character metadata from 1913 to 2013 by year of movie release and their broad ethnicity to find the average significance score per ethnic group per year. We normalized these scores and then visualized the change in the percentage of character significance share over time per ethnicity in the following line graph
Interestingly, at the beginning of the time-series, there were no significant roles held by cast members of ethnicities other than Caucasian or Latino. However, over time, the share of significant roles becomes much more diversified with each minority group taking a more equal share of significant roles. This trend makes sense as we have also observed a general increase in diversity in cast members.
To get a deeper understanding of the share of significant roles, we also employed bootstrapping to compute an average significance score for each minority group over the entire time period and visualized the results in the below pie chart
Black/African mean significance 95% CI: [0.04400008 0.04971878]
Native American/Indigenous mean significance 95% CI: [0.07710139 0.0840154 ]
Caucasian mean significance 95% CI: [0.10740498 0.11243097]
South Asian mean significance 95% CI: [0.09102008 0.10020505]
East Asian mean significance 95% CI: [0.06664806 0.07951001]
Latino mean significance 95% CI: [0.08583279 0.09424447]
Middle Eastern mean significance 95% CI: [0.09055271 0.11385254]
South East Asian mean significance 95% CI: [0.11471654 0.16075354]
Pacific Islander mean significance 95% CI: [0.0715678 0.14581974]
Interestingly, when analyzed over the entire time period, the distribution of significant roles appears to be much more even apart from Black/African actors.