Methodology behind Women Framed
How we scraped over 100,000 articles from Egyptian news outlets, then filtered them down to 455 stories covering violent crimes involving a female victim or perpetrator, across five outlets, applying Association Rule Mining to uncover gender-based framing patterns. Built on a proof of concept developed in partnership with the Egyptian Observatory for Journalism and Media.
Overview
Women Framed examines how women are represented in reporting on violent events in Egyptian news media, including whether they are positioned primarily as victims, perpetrators, or peripheral figures. The methodology matters because it connects newsroom language to broader patterns of gendered representation instead of relying on isolated examples. The project builds on a proof of concept developed in partnership with the Egyptian Observatory for Journalism and Media.
Data sources
We collected more than 100,000 incident-reporting articles published during the 16 Days of Activism Against Gender-Based Violence in 2024 from Mada Masr, Youm7, Al Masry Al Youm, Al Wafd, and Al Ahram. After collection, the corpus was narrowed to 455 stories specifically related to violent crimes committed against women or by women.
Method
The workflow combined automated filtering with human review to isolate the relevant corpus. We then applied Association Rule Mining (ARM) to headlines and article bodies separately, detecting and measuring the strength of recurring associations, so we could compare how framing patterns appear in display language versus full story text.
Limitations & caveats
The dataset reflects the five outlets included in the study, what remained available online, and the precision of the filtering stage. ARM can detect and measure recurring linguistic patterns, but it does not explain on its own why those patterns appear or how audiences interpret them.
Why this methodology matters
This approach gives journalists and researchers a structured, measurable way to audit recurring gendered language at scale. It also makes it easier to identify where representation patterns deserve deeper qualitative reporting and editorial scrutiny.