Saturday, May 23, 2015

Three Hundred Thousand Stories

Content warning: sexual assault, mass shooting, profanity.

One year ago today, a 22-year-old named Elliot Rodger killed six people in an effort to “destroy all women because I can never have them. I will make them suffer for rejecting me. I will arm myself with deadly weapons and wage a war against all women and the men they are attracted to. And I will slaughter them like the animals they are”.

I do not want to write about Rodger. He deserves to be forgotten. I want to write about what happened afterwards, which is that Twitter exploded with the hashtag #YesAllWomen: thousands of people protesting sexual violence and sexism more broadly. Over the next month, I collected more than 1.13 million posts, more than 300,000 of them unique. I have studied many Twitter discussions, but I think this was the most extraordinary, an event unprecedented in history. Rape so often ends in silence: from Alaska to South Africa, college campuses to military barracks, the majority of sexual assaults go unreported, and the vast majority of assaulters walk free.

407 people told stories of being raped in my dataset. As I counted their stories and marveled at their courage I found myself wanting to reach out, like Darth Vader, and choke the life out of those who had hurt them. They were painting a picture that transcended any single tweet: who cared enough about sexual violence to speak up about it, how men and women had spoken differently, how the movement had evolved, who had tried to attack it. Because statistics can never replace individual stories, I also spoke to the most prolific tweeters, who ranged from scientists to sexual assault survivors, dominatrices to grandmothers. This is what they told me.


Three quarters of the tweeters whose gender I could identify from first name were women [1]; they were 10 times as likely as the average tweeter to describe themselves as feminists, and also more likely to describe themselves using words like “gender”, “equality”, “atheist”, “literature”, “violence”, “queer”, and “vegetarian”. To understand how their backgrounds influenced what they cared about, I made a visualization connecting tweets to the words tweeters tended to use to describe themselves. Here’s a screenshot; tweets are circles:
Screen Shot 2015-05-23 at 1.53.01 PM.png
You can see the interactive version and read the tweets here. It shows atheists retweeting "#YesAllWomen are treated like property in the Bible... but extremely valuable property! Like, worth as much as dozens of cows!" It shows men saying that they can do better. It shows gamers and video fans discussing gender inequality in online videos: "When a woman makes a video, most comments are about tearing apart her looks ... with a man, almost none."

But to me the most moving thing about this picture is its lack of pattern. Maybe you can see systematic differences in what yogis and daughters and actresses tweet, but as a statistician trained to find differences in data, I see instead a universal rage: all women face violence and misogyny. Perhaps their only common element is their courage: as their Twitter profiles attest, they are female, fabulous, and fierce.

Of course, there’s a lot more to someone than their Twitter profile reveals. So I wrote to the most prolific tweeters to learn more. Some of them had tweeted thousands of times, and had faced so much harassment for doing so that they were reluctant to talk to me. One, a former dominatrix, told me that need for money had driven her down dangerous paths, teaching her that “a group of men in the world and even some women expect there to be an underclass of people whose lives and bodily integrity just matter less than people's desire for entertainment.”

A second tweeter was a grandmother (never let it be said that they don’t use Twitter), who told me she got “granny-angry”. A survivor of sexual assault and domestic abuse who worked for two decades as a victims’ advocate, she spent a summer posting obsessively -- “I was determined the damned misogynists weren't going to hijack the tag” -- building support with the other frequent tweeters even as they were threatened and, occasionally, publicly identified online.

A third woman, a scientist, got angry at the “flood of trolls” saying “Rape claims are inflated” or “Be grateful you don’t have REAL problems.”

“I was so sick of seeing ‘cite your sources’ from men insulting women who had been raped...that I decided I’d cite ALL the sources,” she said. So she began tweeting out the conclusions of dozens of academic papers on assault.

“A man told me ‘shut up, no one cares,’ ” she said.  “I said ‘Obviously you do, or you wouldn’t be trolling this tag.’ In response, someone else told me to die and threatened to rape me. I reported it as a violent threat, only to have Twitter respond—two days later—that it wasn’t violating their terms of service.”

Interestingly, in spite of these attacks, the vast majority of tweets containing #YesAllWomen supported the hashtag, and most posters saw their number of Twitter followers increase. Still, I wanted to understand who, exactly, attacks a woman discussing her own rape, so I identified widely shared anti-#YesAllWomen tweets and studied the people who shared them. Like the pro-#YesAllWomen tweeters, most of these anti-#YesAllWomen tweeters had more followers a month later, probably indicating their different social circles. I contacted several of the attackers to determine their motivation, but they were not terribly interested in having a serious conversation with a woman studying a feminist tweet, and several preferred to harass me instead. So I resorted instead to my other strategy for getting to know people: I downloaded 1.5 million of their tweets. While I was tempted to homogenize the attackers, I found diverse clusters. Some were high profile, including a mixed-martial arts champion, a conservative commentator and comedian, and a Peabody Award-winning actor; the less high-profile tweeters included political conservatives and wrestling fans. One of the most prolific, who tweeted more than 1,700 times(!) in my data, highlights their diversity: she is, to quote her Twitter profile, “into Wrestling, Catfighting And Sexfighting”, “A Domme Who Conquors Her Slaves” and “Gay But...Do Not Hate Men!”. Ironically, she was upset about the same thing many pro-#YesAllWomen tweeters were: what she saw as double standards. For example, an attractive criminal, like Jeremy Meeks, got women asking to be raped, while a less attractive criminal would not; women who beat up men were powerful, but men who beat up women were abusers.

Because men and women often have different perspectives on sexual assault, I also spoke to men who tweeted prolifically. Two said they were motivated in part by the experiences of women they cared about.

“I've had close friends come home crying from frat parties after a ‘rodeo’ gang rape,” one said. “I've worked jobs where I've seen women slapped on the ass, pinched, whistled at, preyed upon and cowed by management.  I have experienced fear reactions and trust barriers from women...once we're better friends they tell me about the times that they've been attacked.” A second male tweeter said he had been influenced growing up by his mother’s stories of being sexually abused. Both said that other men had questioned their masculinity for defending #YesAllWomen -- “calling me a ‘pussy’, asking me if I ‘had a vagina’ ” -- though one noted he got nowhere near as much abuse as the women.

Then I looked for broader gender differences. Looking at things retweeted mostly by men or women doesn’t reflect too well on the men. The most male-dominated tweets include charming observations like “#YesAllWomen should hit the weight room so they can avoid getting raped wit they weak asses” or “How many Women does it take to change a light bulb? 11, 10 to form a support group, and one to get her boyfriend to do it” [2]. But these tweets aren’t a fair summary of what most men think; rather, they reveal the things that women don’t think. In fact, four of the five most common tweets were the same for men and women.

It’s fairer to compare all tweets from men and women. For example, women use more words expressing fear and use more first-person language. Men use more profanity, and they also use it differently. When men say “dick”, it’s most frequently to retweet, “#YesAllWomen should suck dick on the first date”; for women, it’s “I'm not a tease for letting you buy me a drink and not going home with you. You offered me a drink, not your dick on a plate”. We can also find differences in how people treated those of the same versus the opposite gender. Men used more words expressing emotion, especially negative emotion, when talking to women. Women fired back: more than two thirds of tweets from women containing “part of the problem” were at men.

As I worked on this analysis, the laptop on which I stored the data became a grim weight; walking back to my room late each night, I shied away from the men I passed. But the weight also reminded me I did not walk alone: #YesAllWomen had revealed a vast network of people who shared common fears [3].

One of the tweeters said it best. A survivor of sexual assault who suffers from a severe anxiety disorder, she said she considered not speaking up because of backlash. I asked her why she did anyway.

Things like this help remind everyone who's had to face these types of oppression... that they are not alone,” she told me. “If my putting my neck on the line means that even one other person who's gone through these things feel just a little more supported, a little more safe, a little more like they can get up in the morning and feel okay with existing and that their life might just work out -- so be it.”

Thanks to the members of the #YesAllWomen movement for trusting me with their experiences, and to everyone who speaks out about sexism and sexual violence.


[1] Throughout, when analyzing gender differences, I use tweeter first name to identify gender, a method that works about half the time. It is important to remember that the other half may exhibit different statistical properties. Shoot me an email if you have thoughts on this.
[2] And oh, the many #YesAllWomen lightbulb jokes. A sampling:
How does a feminist change a lightbulb? She just holds on and the world revolves around her.”
“How many feminists does it take to change a light bulb? None, they can't change anything.”
“How many feminists does it take to screw in a light bulb? Nevermind that, the word screw promotes rape culture.”
The feminists countered with some of their own:
“How many men does it take to change a lightbulb? None because they get mad after it won't screw”
“How many ‘nice guys’ does it take to change a light bulb? No one knows, they keep blaming it for not getting turned on.”
“How many men's rights activists does it take to change a lightbulb? Well, not all of them." #NotAllMen #Butprobablyalotofthem”
[3] Indeed, I found some evidence that the movement gained strength as tweeters combined multiple ideas to make more powerful tweets. In several cases, there was an early tweet about how women had to behave and then a later, more widely shared tweet contrasting women’s experience with men’s.
Early Tweet
Later Tweet
Because you get a "rape whistle" when you start college #YesAllWomen
Because when girls go to college they're buying pepper spray and rape whistles while guys are buying condoms #YesAllWomen
The constant worry your skirt or dress is too short or maybe your bra strap is showing and you'll attract the "wrong attention" #YesAllWomen
a ‘cool story babe, now make me a sandwich’ shirt doesn't break the school dress code. a girl's bra strap does #YesAllWomen

The marketplace of ideas has become, perhaps, the fusion reactor of ideas, where memes combine with explosive power and supporters of a social movement draw strength from each other through digital synapses. But it is important to note that establishing causality is difficult -- how do we know if the second tweeter saw the first tweet? Still, the pattern is repeated and the explanation is plausible.


  1. Thanks for this. I appreciate you being willing to bear the "grim weight" of this project in order to provide a little insight for people like myself, who are relatively uninformed.

  2. Glad you liked it :) Much more credit really goes to the tweeters.

  3. This comment has been removed by the author.