Are We Watching More Diverse Movies in the Digital Age? A Quantitative Analysis on the Consumed Diversity on MovieLens

Changes in how users rate movies on MovieLens is a snapshot of an individual’s lifetime movie consumption. This thesis studies the diversity of individual movie consumption by making use of the datasets about movies and users from MovieLens, a movie review website applying recommendation algorithms. Based on the Stirling Model and the Intra-List metric, a dissimilarity score is used as the indicator of diversity to measure the similarity between any two movies. The results show that users tend to watch less diverse movies, whether on an aggregate level or an individual level, even though we thought digitalization should have promoted a more diverse consumption. A decreasing trend in the overall diversity of individual movie consumption since 2017 is statistically identified with an interrupted time series analysis, which mainly results from a less diverse supply of new movies. Moreover, an individual’s movie consumption is observed to be less diverse as experience grows due to a stronger preference for personalized movies and new releases. The methods and findings in this study could help players in the movie sector enhance product differentiation and evaluate the effectiveness of a cultural law or policy in promoting diversity.