The Media Propagates Stereotypes

The media holds tremendous sway over how individuals perceive and interact with various ethnic and gender-based groups. News articles have the power to shape people’s perceptions, beliefs, and actions in significant areas such as education, family, and politics. Understanding how the media portrays different groups, and whether these depictions are driven by stereotypes, is therefore crucial.

However, researchers face a daunting task in decoding the multitude of ways in which media outlets encode stereotypes, some of which may not be readily observable or quantifiable. To address this issue, recent studies have employed natural language processing techniques to analyze the language used by journalists in their reporting.

Unfortunately, the significance of visual imagery in shaping the slant and stereotypes of media outlets has been largely overlooked. Recent research from Bocconi University aimed to examine visual stereotypes that emerge in newspaper images.

Examining the media

The researchers have employed a custom-trained deep-learning model capable of identifying the identity features, such as ethnicity, gender, and race of people featured in images. By doing so, the researchers can automate the analysis and avoid the time-consuming and subjective process of manual coding. Additionally, this approach ensures consistency in classification that cannot be guaranteed with human coders.

The researchers focused on two prominent US news outlets, namely the New York Times and Fox, and analyzed over two million articles published on their web editions between 2000 and 2020, of which 690k included an image.

The study’s main focus was on occupational stereotypes portrayed in newspaper images. The researchers aimed to determine whether news outlets perpetuate common stereotypes regarding the occupational choices of specific groups. For instance, whether white men are often shown in managerial positions.

Uncovering biases

Combining computer vision and text analysis techniques, the researchers discovered that news article images perpetuate gender and racial stereotypes. In other words, occupations stereotypically associated with specific identity groups, such as female or black, are more likely to feature images of individuals from those groups.

To clarify, an occupation is deemed stereotypical if a higher proportion of a particular identity group works in that profession. For instance, even though the overall population of black people in the US might not necessarily be associated with “mail processor,” if a higher share of black people works in that profession than other groups, it becomes a stereotypical occupation for that group. As an example, the most stereotypically female job is “secretary.”

Notably, the researchers controlled for the true occupation shares in each profession, such as the proportion of black people working as mail processors. This allowed them to differentiate between stereotypes and genuine differences in representation of identity groups in various occupations in the US.

 

Facebooktwitterredditpinterestlinkedinmail