Headline seems a bit misleading. "Half of online abuse" seems to suggest that this study is representative of the internet as a whole instead of just twitter. While the findings are still worthwhile, I think we'd see different results if the study could have included sites like 4chan, reddit, tumblr or even facebook. Of course, there are practical challenges to getting accurate demographic information on those platforms, but I suspect that fact also factors into the nature of abuse that persists on those sites.