Slashdot Mirror
A Woman on Twitter is Abused Every 30 Seconds (fastcompany.com)
That shocking statistic comes from a study conducted by Amnesty International and AI software startup Element AI. From a report: In the study, called Troll Patrol, Amnesty International and Element AI looked at data from 288,000 tweets sent to 778 female politicians and journalists in the U.S. and U.K. in 2017. Using machine learning on the data, the group then extrapolated just how wide-ranging abuse toward women is on Twitter. The result: 1.1 million abusive or problematic tweets were sent to the women in the study during the year -- that's one abusive or problematic tweet every 30 seconds. And it's even worse for women of color -- and especially black women -- who were targeted more frequently than white women.
I just looked at the published methodology.
I detect some issues with their methodology. Definitions of what constitute "abusive" or "problematic" tweets are explained in Appendix E.
Their explanations of the categories do not seem very precise. They are vague enough to include lots of edge-cases. Their examples are clear enough, but a clear example does not exonerate a vague criterion. (Hypothetical example: "I am going to murder you tomorrow" is certainly an example of abusive and threatening speech... but what else is in included as threating and abusive speech? Counterexamples are also necessary, else "I hate that rabbit in the field behind your house" might also be considered abusive.)
But more to the point, a full 55% of the tweets they flag as "abusive content" fall under the category of "other"... and that's a huge red flag.
They describe "other" thus: There will be some tweets that fall under the âother categoryâ(TM) that are problematic and/or abusive. For example, statements that target a userâ(TM)s disability, be it physical or mental, or content that attacks a womanâ(TM)s nationality, health status, legal status, employment, etc.
Uh... yeah. "Some" is 55%. And not only is this one the majority, it is a particularly vague definition. For example, if someone replies to an irate tweet, "Calm down! I know the Irish are famous for temper, but cool down a little!" is that "targeting a woman's nationality"? According to their description, it could be.
But here is the biggest problem with this study:
Nowhere did they compare this to any other groups. They singled out famous women and women in politics, but they didn't compare against famous men and men in politics.
And since there is no such comparison, it's all pretty much meaningless.
What if famous men and men in politics are "abused" on Twitter every 31 seconds? Or every 28 seconds?
Without that knowledge, does the study really say anything other than "the subset of people we chose got 'abused' at this rate?"
So what? Nothing to compare it to makes it pretty much useless knowledge.
While what they say may be true,
Why limit research to one segment to create a self confirming headline?
To answer your question: because you can't study everything at the same time without making assumptions you can't justify yet. This happens all the time in social sciences. Before you can safely lump things together you have to study them separately.
On the face of it, attacks on women appear distinct in their character from attacks on men, although exactly how different is obviously possible to exaggerate. But before you object, yes, there have been studies that focus on male victims of social media bullying too, they just didn't get a mention here. Social media bullying is a hot research topic, but it's early days yet and because this is social sciences that means results are highly unreliable. This is largely because potential research populations tend to be mixed bags of apples and oranges.
So expect a lot of research looking at various target groups: men vs. women, straight vs. gay, or as in this case women in general vs. women of color. Really the best social science tends to take big homogeneous trends and tries to parse them into distinct pieces. For example the US economy *on average* resumed reasonable growth between 2010 and 2015, but a gulf emerged between large cities, where nearly all that growth took place, and rural/small town areas which continued to see job losses and falling labor participation rates.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
A Woman on Twitter is Abused Every 30 Seconds
And a Trump voter ever 30 milliseconds. So what else is new?
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
On the face of it, attacks on women appear distinct in their character from attacks on men, although exactly how different is obviously possible to exaggerate. But before you object, yes, there have been studies that focus on male victims of social media bullying too, they just didn't get a mention here.
Of course they wouldn't mention it. Because those studies show overall, men receive more online abuse. That wouldn't go so well for the desired outcome of this particular "study".
They explain their methodology and there is actually a textbox you can put your own test tweets into and it will give you an analysis and "problematic" score. It's right near the bottom of the page.
As for a the GP's question, they are not comparing to anything external. The study isn't making the claim that it's worse for women, merely that women get a now quantified amount of abuse and that it affects certain sub-groups more than others (women of colour, women with left leaning views/politics).
As a general point if men were getting as much or even more abuse, it wouldn't make any difference. Also studying how it affects women is helpful for men being abused too, as often the solutions are similar for both.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
A simple search turns up these:
https://www.abc.net.au/news/2016-05-27/research-claiming-women-are-half-of-online-abusers-flawed/7452568
https://phys.org/news/2014-10-online-abuse-affects-men-women.html
http://www.pewinternet.org/2014/10/22/online-harassment/
There is a lot of misinformation from all sides.. Anything to reinforce their specific ideas or points..
Try to keep an open mind and listen to both sides and maybe you learn something new from time to time.
You could google for yourself you know. This one aims at celebrities, but there's others that show the same trend for the general public. https://demos.co.uk/press-release/demos-male-celebrities-receive-more-abuse-on-twitter-than-women-2/
Also, fun fact, women abuse people on twitter at the same rate as men, https://qz.com/692902/on-twitter-a-study-says-half-of-all-sexist-abuse-comes-from-women/
Those damn women hating women!
It also very much depends on how you define attack. I would rather like to see the precise definition before rushing to any judgement.
Seven puppies were harmed during the making of this post.
Where exactly did this idea, that being offended by insulting language is abuse, come from? It's nonsense and we need to get away from it.