An Evaluation of Gender Bias in 167K Movie Posters

Movie posters, as the front-line marketing tools, significantly shape audience expectations and perceptions, making their analysis crucial for understanding broader industry trends. This study uses the DeepFace facial recognition library to identify ‘gender bias’ in movie posters, which we define by the disproportionate prominence of male characters in terms of visibility, spatial position, and size, as compared to female characters. Employing novel analytical techniques, we examined a dataset of 167,322 movie posters and observed a prevalent male bias. This bias is consistent across genres, geographical regions, and decades, though variations exist: action and adventure genres display pronounced male bias, while romance genres show more balanced representations. Additionally, movies exhibiting male bias tend to achieve higher box office revenues and ratings. Our findings underscore the need for addressing gender representations in movie marketing materials and suggest avenues for future research in this domain. We conclude with a poster analysis of the 50 highest-grossing movies of all time, which shows that gender – and racial – representation is improving over time.