Instagram Gender Portrayal Metrics
March 2024 - January 2025
🚧 This page is currently under construction.
Content and visualizations are still being finalized.
Introduction
During the depths of COVID, I spent some time scrolling through Instagram news accounts across the political spectrum. And a pattern caught my eye: most images featured men, and certain demographics were used in strikingly deliberate ways.
For example, some accounts used images of women of color to convey positive issues, while others used similar demographics in negative contexts.
These choices — whether intentional or not — shocked me because of how blatantly they associate certain demographics with specific connotations, influencing audience perceptions and beliefs.
We already know that text can reinforce gender biases, and recent research suggests these effects are significantly stronger in images (source). However, modern social media platforms, like Instagram and TikTok, use both combine both text and images into a single medium enabling a powerful more nuanced form of bias reinforcement. For example, ungendered captions such as "This is the pinnacle of humanity" convey wildly different gender messages depending on if they are paired with an image of a man or a women.
This realization led me to ask: ? And how do these patterns vary across the political spectrum?
Inspired by these questions, I set out to explore: How do Instagram news accounts portray gender through images and text?
Project Description
This project quantifies gender portrayal on Instagram news accounts by analyzing both the images and captions of their posts. I focused on a central question:
Does the emotion conveyed in a caption differ based on the demographics in the image?
For example, are captions more likely to express anger, joy, or sadness when images feature women compared to men?
To answer this, I did the following:
- 1. Scraped 150,000+ Instagram posts from 14 news accounts (e.g., @bbc, @cnn, @foxnews).
- 2. Used a sentiment analysis model (Hugging Face's BERT) to classify captions into emotions like joy, anger, sadness, and fear.
- 3. Applied object detection (OWL-ViT) to count the number of men and women in each image.
- 4. Combined this data to analyze emotional trends across demographics and political affiliations.
By connecting these dots, I uncovered patterns in how media portray gender—intentionally or not—across the political spectrum. For more details on the data and methods, check out About the Data.
Armed with this data, I analyzed three key aspects of gender portrayal:
Gender Representation:
How often are men and women featured in Instagram posts?
Emotion Trends:
What emotions are most commonly associated with posts featuring men versus women?
Combined Analysis:
How does the emotional tone of posts vary when images feature one man versus one woman?
Findings
Flow: 1. More men than women, 2. Very Different emotional profiles, 3. Gender is related to emotion.