Unsocial Media: Tracking Twitter Abuse against Women MPs
By Azmina Dhrodia, Researcher, Technology and Human Rights
“I’ve had death threats, I’ve had people tweeting that I should be hung if ‘they could find a tree big enough to take the fat bitch’s weight’… I’ve had rape threats…and n*gger, over and over and over again”
Diane Abbott, UK MP and Shadow Home Secretary, July 2017
Diane Abbott’s testimony during a recent parliamentary debate about intimidation against members of parliament (MPs) was the first time many people heard the extent of sexist and racist abuse facing women in UK politics. But for women MPs who are active on social media platforms, these harmful encounters are something they must contend with daily.
At Amnesty, we’ve been investigating the extent of online abuse against women MPs active on Twitter in the UK through individual interviews and by using machine learning to detect abusive tweets sent to women MPs. The findings outlined in this post provide a detailed look at abuse on Twitter in the run-up to the 2017 election — in which Diane Abbott’s case stands out for all the wrong reasons.
The online abuse she and other women MPs experience sits in a wider context of pervasive and damaging attacks against women from all walks of life on social media platforms. For the last eight months I’ve been speaking to journalists, activists, bloggers, comic book writers, comedians and women active in all levels of politics and public life to hear about their experiences of abuse on social media platforms. I’ve had numerous long chats with women in cafés, parks, hotel lobbies, at youth centres or via Skype, and each time I hear the same message. Twitter can be a scary place for women online. Whether women use social media platforms as public figures or for personal use, the threat of abuse is all too real and it is having a silencing effect on women’s participation online and in the public sphere.
A Toxic Place for Women
While online abuse is certainly not limited to women in the public eye, women politicians face an extraordinary amount of abuse on social media. To understand just how much abuse women MPs face online we worked with a data scientist to analyse a sample of Twitter data from 1 January to 8 June, with a focus on the six weeks prior to the 8 June UK election. We wanted to understand how many of the tweets sent to women MPs were abusive, whether some MPs were targeted more than others and if there were trends in such targeting.
We also wanted to use Twitter’s own data to demonstrate the scale of online abuse and to show how discrimination against women doesn’t just disappear when you move into the digital world. Gender inequality in society exists both online and offline and people of all genders can experience online abuse. However, deep rooted and negative gender stereotypes against women also influence the way some individuals communicate online. This means that online abuse against women is often sexist or misogynistic in nature, and online threats of violence against women can be sexualized and usually include specific references to women’s bodies.
Our sample of 900 223 tweets between 1 January and 8 June was drawn from social listening tool Crimson Hexagon. Crucially, because we could only download historical Twitter data, our sample did not include tweets that have been deleted or tweets from accounts that were suspended or disabled. We made multiple requests to Twitter for access to a full data set covering the period of analysis but our requests were refused. This means we can only assume the true scale of abuse facing women MPs was even higher than our results show.
It is also important to note that any automated or semi-automated data analysis is imperfect. Further information on the figures used in this study can be found at end of this post including a link to our full methodology.
Despite the limitations in the data, the results of our study were striking. The analysis revealed the following:
· Diane Abbott received almost half (45.14%) of all abusive tweets in the run up to the Election
· Excluding Diane Abbott, black and Asian women MPs in Westminster received 35% more abusive tweets than white women MPs
· Online abuse cuts across party lines, affecting women from all UK political parties
Interviews with women MPs also highlight the serious psychological impact of online abuse. Overall, our study demonstrates just how much more work needs to be done by social media companies and governments to tackle this ever-growing problem.
1. Diane Abbott Receives the Most Online Abuse
Earlier this year, Amnesty International hosted a Hackathon with Accenture Digital during which participants were tasked with quantifying and analysing abuse against women MPs online. Over the two days, the high levels of abuse against Diane Abbott in the data set was flagged by almost every group of participants at the event. With these early findings, we were keen to find out whether a dedicated data scientist applying a more fine-tuned and tested methodology to the same data would produce a similar result. The detailed analysis confirmed these initial findings.
In the six weeks prior to 8 June, Diane Abbott received almost half or 45.14% of all abusive tweets against women MPs included in our study. For the total period of analysis between 1 January and 8 June she received 31.61% or almost one-third of all abusive tweets. Not only did she top the list of MPs for most abusive tweets but she received 10 times more abuse than any other woman MP in the run-up to the Election and 8 times more abuse than any other woman MP during the entire period of analysis.
Our findings show that 5.8% or 8121 out of 140 057 of all tweets mentioning @HackneyAbbott were classified as abusive.
This means that Diane Abbott alone received more abuse during the total period of analysis (1 January to 8 June) than all women MPs from the Conservative and Scottish National Parties combined.
As the first black woman MP elected into Parliament, Diane Abbott is no stranger to targeted abuse. Thirty years after she was first elected, we met in her Westminster office, a small room overflowing with books and policy folders. On her desk, in between stacks of papers sits a framed photo of her with Nelson Mandela. Diane explained to me the difference between the abuse she faced back then and the abuse she faces now.
“When I was a new member of parliament, you might get one racist letter a week. But that was because if you were racist and you wanted to abuse an MP, you had to write a letter, you had to put it in an envelope, you had to put a stamp on it and you had to put it in the letter-box. Now, some days, we can get hundreds of items of abuse, depending on what happened the previous day.”
“It’s the volume of it which makes it so debilitating, so corrosive, and so upsetting. It’s the sheer volume. And the sheer level of hatred that people are showing.”
She also described what she believes to be the difference between genuine political debate or even criticism of her policies and the online abuse that she receives on a daily basis.
“I welcome scrutiny, and I welcome engagement, and I welcome debate. That’s why I was so positive about these online platforms. But the problem is when people are not engaging in debate or scrutiny but just showering you with abuse — [saying] that you are a nigg*r, that you are a prostitute, threats against your safety. It’s just abuse which has no political content and which actually people wouldn’t say in a meeting or to your face. I think the distinction between abuse and genuine political debate is, would they say it if they met you in the street? No, they wouldn’t.”
Upon hearing about the disproportionate levels of abuse against her, Diane Abbott told me,
“Although [I’m] not entirely surprised by the results of the research, it is truly sickening to see it in figures. These individuals believe it is acceptable to hurl this level of abuse directly at me, knowing there are little to no consequences to their actions.”
2. The More Identities You Have, The More Abuse You Face
The combination of racist and sexist abuse received by Diane Abbott is reflective of a wider trend of ‘intersectional discrimination’. That is, discrimination that targets an individual on the basis of their different identities.
Any analysis of online abuse against women should not be limited to only applying a gender lens to the data. When you are a woman with multiple or intersecting identities, your experience of the world is not just limited to your gender. Your race or disability or sexual orientation, for example, can have just as much of an effect as your gender — if not more — on how you are treated both in the physical and digital world. In the case of online abuse, women of colour, religious or ethnic minority women, lesbian, bisexual, transgender or intersex (LBTI) women, women with disabilities, or even non-binary individuals who don’t conform to traditional gender norms of male and female, will often experience abuse that targets these different identities.
Diane Abbott standing out in our analysis is an acute example of how intersectional discrimination works. The abuse that she faces is not just sexist and misogynistic; it’s also incredibly racist.
For example, she recounted to me the response she received after she spoke about her experience of online abuse in a recent parliamentary debate.
“My office got flooded with communications, both by letter and by email. People sent us emails and letters full of swastikas, people sent us postcards and letters with pictures of monkeys and chimps. People sent us hundreds of emails using the word nigg*r — that’s the sort of response we get. It’s highly racialised and it’s also gendered because people talk about rape and they talk about my physical appearance in a way they wouldn’t talk about a man. I’m abused as a female politician and I’m abused as a black politician.”
Our findings demonstrate intersectional online abuse against women MPs in more than one way. Despite representing only 8.8% of women MPs in Westminster, Asian women MPs were found to receive the most abusive tweets per MP.
Our analysis shows that Asian women MPs receive 132 abusive tweets per MP. This is 30% higher than white women MPs who receive 92 abusive tweets per MP. Given the disproportionately high levels of abuse against Diane Abbott, black women MPs were found to receive 2,781 abusive tweets per MP — but when Diane Abbott is excluded from the analysis — the findings show black women MPs receive 81 abusive tweets per MP.
Earlier this summer I spoke to Tasmina Ahmed-Sheikh, a former MP for the Scottish National Party and the first Black Asian Minority Ethnic (BAME) woman from Scotland to be elected to any Parliament whose Twitter account was also included in our analysis. Tasmina explained how she began experiencing online abuse during her campaign for the 2015 general election.
“When I was elected in 2015 and even during my election campaign, I found myself at the other end of horrific levels of abuse. And the question is: why might that be? Is everyone receiving the same levels of abuse? Is it women? Is it because I’m Black Asian Minority Ethnic (BAME)?”
When I asked Tasmina for examples of the type of abuse she faced she told me that somebody once tweeted her home address and postcode which then required police to patrol her house. The police have also advised Tasmina to designate a safe room in her house where her family should meet in case a threat made online should transpire. She also recalled a pile-on of abuse she received after appearing on BBC Question Time in November 2016.
“People said I should be grabbed by the pussy, that I should have items inserted into me…people were inciting others on Twitter to follow on the bandwagon and retweet [the abuse]. These people have been reported to the police, as have others who have suggested that my family should be bombed or my house blown up… ”
Tasmina wasn’t surprised about the high levels of online abuse experienced by Asian women MPs. She explained:
“I am from a Scottish Asian community. I am a Muslim. And I’m a Woman. So it’s everything. It has an exponential effect, so people will pile on for a variety of different reasons. Some of them because you are all of these things, and some because you are one of these things, or two of these things, which makes it so much more difficult to deal with, because you just wonder where do I start with this?”
3. Online Abuse Crosses Party Lines
Although our findings highlight the intersectional nature of online abuse against women MPs, it’s important to recognize that online abuse affects almost all women MPs and cuts across political party lines. Our analysis found that 2.85% of all tweets sent to women MPs between 1 January and 8 June were identified as abusive amounting to a total of 25 688 abusive tweets out of 900 223 total tweets.
Online abuse against women MPs on Twitter almost doubled in the run-up to the snap election with the days leading up to the election seeing the highest levels of abusive tweets at 5.03% in June.
Online abuse also cuts across political parties. Below is a breakdown of political party representation of women MPs in the UK as well as online abuse by party affiliation.
Since the disproportionate amount of abuse received by Diane Abbott skews the results, with her included Labour receive by far the most abusive tweets per MP. However, if excluded, SNP party members received the most abuse, with SNP women MPs receiving almost half of all abusive tweets in our study.
Although the levels of abuse may vary across political party and individuals, our findings show that no political party is immune to online abuse against women MPs on social media.
The cross-party nature of online abuse is further demonstrated by the fact that women MPs from the three largest political parties are represented in the Top 5 List of Most Abuse Received on Twitter. Diane Abbott along with Emily Thornberry (Labour), Joanna Cherry (SNP), Amber Rudd (Conservative) and Angela Rayner (Labour) made up the five women MPs receiving the most abuse in the six weeks prior to the snap election. Jess Phillips (Labour) and Anna Soubry (Conservative) also both topped the list for the highest number of abusive tweets during the wider period of analysis between 1 January and 8 June.
In December 2016, after receiving threatening tweets she subsequently reported to the police, Conservative MP Anna Soubry posted about her experience of online abuse. She stated,
“Like all MP’s I regularly and increasingly receive offensive emails, tweets and the like…I welcome forthright debate and do not object when constituents exercise their right to contact me in no uncertain terms. But I am no longer prepared to accept foul abuse, misogyny and hate….We must all reset the dial of debate to a more tolerant, better informed and honest level. If we don’t, those who revel in bully boy tactics will win and we run the risk of losing our hard won rights to free speech and democracy.”
Labour MP Emily Thornberry was also not shocked when hearing about her high ranking in our analysis. She explains,
“I wouldn’t say I’m surprised because I think it tends to correlate to the media profile you have, and obviously the election campaign meant doing a lot of TV and radio, so perhaps it was inevitable. I guess other people may be surprised because the abuse is not something I ever talk much about in public.”
She adds, “On a practical level, the violent stuff and the death threats are just very time-consuming. There’s a big process to go through on each occasion with the police and the House authorities, there’s obviously extra security measures you have to put in place each time, and also if your kids see the tweets or it’s the first time for new members of staff, you have to do a lot of reassurance with them that they shouldn’t worry. Obviously you’re partly reassuring yourself as well.”
The Impact of Abuse on Women Expressing Themselves Online
For many individuals, social media platforms have now become the medium of choice to express abusive behaviour. Some people may think that online abuse is ‘less real’ than offline abuse, that online abuse can be easily ignored, or that women can simply ‘turn off’ these platforms — and by extension choose to stop receiving abuse. However, these responses not only deny women their right to freedom of expression, they fail to recognize that ‘turning off’ is not always an option for women who wish to (or need to) stay engaged online.
Women MPs, in particular, rely on social media platforms to engage with their constituents and the wider public outside of traditional office drop-in hours. As Tasmina Ahmed-Sheikh explains:
“Social media is hugely important tool… from a politician’s perspective or from anyone’s perspective who wants to get information out in public domain, it’s a great, great way of doing it”.
Nevertheless, online abuse or even the threat of it can drive women off platforms like Twitter and have a chilling effect on women speaking out online.
Ensuring that everyone can participate freely online and without fear is vital to ensuring that the internet promotes freedom of expression equally. Otherwise, the risk is that women will self-censor themselves online, refrain from engaging on certain subjects or in political activism, or will choose to leave social media platforms altogether.
The Psychological Toll of Online Abuse
While our findings demonstrate the scale of online abuse facing women MPs, they do not measure the psychological impact of that abuse. Research carried out by the Association for Progressive Communications, an organization at the forefront of addressing technology related gender-based violence issues, found that online abuse “can cause psychological and emotional harm, reinforce prejudice, damage reputation, cause economic loss and pose barriers to participation in public life, and may lead to sexual and other forms of physical violence.”
Samantha Silverberg, a Licensed Mental Health Specialist and Co-Founder of Online SOS , explains the extent of mental health implications on individuals experiencing abuse or harassment online:
“There is little research examining the psychological toll online abuse has on individuals. Anecdotally, we can see [the toll] when individuals are fearful of opening their emails, unable to return to work, or are making other changes to their daily lives based on fears related to the abuse.”
When I first asked Diane Abbott about how online abuse has personally impacted her, she did not immediately open up and spoke instead about the negative impact on her staff who often deal with the large volumes of abuse she receives. However, after a long pause, she admitted, “Online abuse does damage you, it damages your confidence and it corrodes your self-esteem”.
Tasmina Ahmed-Sheikh was emotional as she opened up to me about the implications of receiving torrents of abuse online.
“[Online abuse] is very difficult and very upsetting. It has been really difficult for me to deal with because I’ve sat and I have wondered why I am doing this and my family wonder why I’m doing this.
“I can be tough. I can be a mum and deal with this rubbish and still be right out there [again]. And externally, that’s what people see. Internally, it hurts a lot. It really, really hurts a lot. It’s personal. People are criticizing you, where you come from, your parents, what you believe in, the religion which you believe in, and for being a woman, in which you don’t have a choice… so that is really hard to deal with.”
“But you have got to stand up like I am doing now and not pretend you can cope when actually you are not.”
So What Should be Done?
Online abuse against women on this scale should not and does not have to exist on social media platforms. Companies like Twitter have a responsibility to respect human rights, which means ensuring that women using the platform are able to express themselves freely and without fear.
Twitter themselves acknowledge this. In 2015, Twitter announced its plans to get tough on online abuse, with the company’s General Counsel, Vijaya Gadde, declaring in an op-ed that “freedom of expression means little as our underlying philosophy if we continue to allow voices to be silenced because they are afraid to speak up”.
And yet this is exactly what is happening. Online abuse flourishes on the platform, making it a toxic space for women. A recent study published on 22 August by the Fawcett Society and Reclaim the Internet reviewed a range of abusive and violent content on Twitter which was then reported to the platform by anonymous accounts. However, one week later, the reported posts remained on Twitter and the accounts which reported the abuse received no further communication from the platform.
Twitter, first and foremost, must enforce its own policies on hateful conduct and abuse. Much of the abuse detected by Amnesty contravenes Twitter’s own Terms of Service and Community Standards. Yet, months later, the abuse remains on the platform. Although the company states that ‘it doesn’t tolerate behaviour that harasses, intimidates or uses fear to silence another person’s voice’, this is exactly what is happening to many women who use their platform, including women in the highest levels of government.
In July, Twitter stated that they are now taking now taking action on 10X the number of abusive accounts every day compared to the same time last year. But it’s clear that Twitter needs to invest more resources to enforce adequate and transparent reporting mechanisms that users have confidence in utilizing. Training about the various ways online abuse manifests on the platform, including specific attention to gender and other identity-based abuse, should be mandatory for all staff members, including developers, researchers, and especially moderators. The company should also record and publicly share disaggregated data about the levels of abuse and their response. Twitter needs to be transparent about the resources it’s investing into tackling online abuse.
The UK government has an obligation to protect women from human rights abuses and has taken some steps to address violence and online abuse targeting women on social media. The Committee on Standards in Public Life will soon hold a hearing about online and offline intimidation of Parliamentary candidates during the 2017 election period. The hearing is part of a Parliamentary Enquiry launched by Prime Minister Theresa May to investigate both online and offline abuse against all MPs and other standing candidates.
The enquiry presents an important opportunity to call on social media companies to increase transparency in reporting mechanisms and in resources dedicated to ending online abuse — particularly online abuse against women — on their platforms. The enquiry must also ensure that an intersectional gender lens is applied to any recommendations which recognizes the different ways that women, and especially women with different identities, experience online abuse.
The UK government must also exercise caution. Introducing legal sanctions on companies that fail to remove content are both dangerous and unnecessary. The UK government must commit to tackling the source of the problem and invest in programmes that challenge negative gender stereotypes of women in society which manifests as misogyny and abuse online. The government must also ensure that any response to this issue recognizes that online abuse against women is an extension of existing offline discrimination and abuse against women. A failure to do so will have serious consequences for women now and in the generation to come.
Nearly ninety years after women won the right to vote, there is a real danger that the high levels of online abuse against women MPs will have a chilling effect on women taking part in public life — particularly women of colour. This is not only detrimental in terms of the possible long-term effect on the representation of women in politics in the U.K but also continues to deepen societal inequality between genders.
“I think the online abuse I get makes younger women of colour very hesitant about entering the public debate and going into politics. And after all, why should you have to pay that price for being in the public space?”
(Diane Abbott, Amnesty International Interview, July 2017)
Notes on Figures Included in This Study
Our analysis found that between 1 January and 8 June 25 688 tweets out of 900 223 total tweets were identified as abusive. The true positive rate, i.e. the tweets labelled as abusive by the model that are truly abusive, is around 64%. The remaining tweets are false positives, i.e. not truly abusive. This means that we expect the number of truly abusive tweets within the 25,688 tweets to be 16,440 (= 0.64 X 25,688 ).
However, a large number of abusive tweets are not captured by the model. Only about 44% of truly abusive tweets are detected and the rest are false negatives, i.e. tweets not labelled as abusive but are truly abusive. This means that our estimate of the actual number of abusive tweets out of the 900 223 total number of tweets is around 37 364 truly abusive tweets (= 16,440 / 0.44). For the purpose of this publication, we have used the 25,688 figure.
You can find our full methodology here.