ALL GOOD THINGS MUST END, CATS FANS!

(the unified scale shows compared to the entire world, the decile scale only accounts for what's in the map. you can access all maps at: assets.aeaweb.org/asset-server/files/22622.pdf)

Ok, now: is the study accurate?

Who wrote it: Two Meta researchers 🚩worked with a business school to write this paper (Associate Professor of Finance, predoctoral fellow in Finance, Professor in Finance 🚩). I would have expected to see Mass Media studies, Gender Studies, Sociology, and/or Computer Science. These people aren't unqualified for this research, but I would want a thoroughly sourced introduction and discussion to make sure they had a good understanding of the field (this was not present in the paper. 🚩5/12 of the citations were from the same set of authors, 3/12 were statistical data, 2/12 heavily rely on citations from this set of authors for the point 🚩🚩🚩, the remaining 2/12 were used for the first sentence of the article 🤡🚨)

Publishing: Peer-reviewed? No. Journal? No. Papers and Proceedings of the American Economic Association. Okay, so this was a paper presented at a conference 🚩. So we know the review process for this was less strict, and honestly from the wrong field. (The citations reveal why this may be true: the two professors are trying to promote economic assessment of social media websites as a valid path of economic study. Social Interactions, Resilience, and Access to Economic Opportunity: A Research Agenda for the Field of Computational Social Science 🚩🚩. If you ask me, it seems like FB is paying to try and legitimize this kind of research)

Method: A good number of flaws.

Ignoring the bottom 25% of users by friend count "to focus on active users". (To me, this excludes the group of people most likely to have created their friends list to only include their real friends)

Including the top 25% of users by friend count. (scammer/bot data definitely included 🚩)

On that front, ignoring total friend ratio (the actual number we see is men's friend gender ratio of their top x friends compared to women's friend gender ratio of their top x friends).

Using FB's secret method for assessing friendship level 🚩🚩. This is critical, because they use it as an objective measure of friendship, but we don't know what determines this assessment. Most likely, it's based on how much interaction happens between a dyad, which skews it based on activity level (someone always on FB is more likely to have someone else who is always on FB as a "closer friend" than someone rarely on FB, regardless of actual friendship level). ⚠️ This makes any possible outcome circumstantial ⚠️ and is one of the things a seasoned Mass Media researcher would have told you.

Assuming "Facebook friendships generally reflect real-world social ties". They cite this one! They only cite papers by this group of authors, and a paper that cites papers by this group of authors 🚨.

They only include users active in the month before January 9, 2025. ⚠️ Why do they need for two separate ways to only include active users ⚠️ (this is the kind of data exclusion you see when someone is trying to make their data fit a certain outcome 🚩)

They find their friendship data to be valid because it "correlates strongly with gender differences in labor force participation" 🚩. How are these two things connected in a way one helps prove the other? ⚠️ Unaddressed assumption ⚠️ (this is the certain outcome).