This repository contains a replication study of the Demos report on gender and misogyny (https://demosuk.wpengine.com/wp-content/uploads/2016/05/Misogyny-online.pdf) and additional analysis.
The Demos report is pretty scant on technical details and methodology, so we have relatively sparse information on which to duplicate their analysis. Additionally, their comments about "Method 52" and the NLP pipeline for automatically inferring gender inspire some skepticism on my part. For a potentially inflammatory result like "the majority of misogynistic content on Twitter is written by women", the reader deserves more details about the process that came to this conclusion.
This work tries to replicate this result on a larger Twitter dataset (all of 2015) and a larger set of misogynistic content.