High profile cases like rape threats on Twitter against a feminist campaigner make the headlines, but how widespread is abuse on social media? We try to provide some data
• Get the data
Rape threats made to feminist campaigner and freelance journalist Caroline Criado-Perez on Twitter have sparked a heated debate about the criminality of online abuse and a company's responsibility to tackle it.
The threats, which came after Criado-Perez ran a successful campaign to get a woman on a bank note, are the latest in a long line of online abuse cases to gain widespread attention.
The subsequent debate has focused on how much Twitter and other companies should do to tackle abuse of their users by other users.
There are three main camps, loosely 'do regulate', 'don't regulate' and 'can't regulate'. Some of the arguments are ideological, but others centre on whether it's feasible or not to regulate or moderate billions of messages.
While anecdotal evidence of horrific threats abound, there's little data on how common occurrences of abuse are and how widespread the problem is. We've tried to look at what's available to help inform the debate.
Hidden numbers
There's one immediate problem: no one systematically measures the frequency of abuse of the type that Criado-Perez received when Twitter users threatened to kill and rape her.
The UK communications regulator Ofcom only keeps statistics about online bullying targeted at children. The same is true of helplines and campaign groups like UK Safer Internet Centre, Internet Watch Foundation and BeatBullying.
That leaves us in a difficult position of having to collect together piecemeal statistics from individual specific sources that might indicate the scale of abuse in our online interactions.
2% of internet users
In 2012, the Home Office published a special bulletin titled 'Hate crime, cyber security and the experience of crime among children'. The numbers are slightly dated - they're from the 2010/11 Crime Survey - but they do ask internet users over the age of 16 about their 'negative experiences' online.
The statistic might seem surprisingly low. Just 2% of all internet users said they had experienced abusive/threatening behaviour in the past year - although that number was significantly higher for young people - 6% of women aged 16-24 and 5% of men aged 16-24. A higher proportion said they had seen upsetting images, though these weren't necessarily targeted at the user.
That doesn't seem to tally with the anecdotal evidence. Maybe a small number of users receive large volumes of abuse which skews our understanding of how prevalent this is. If that is true, the percentage of respondents saying they have experience online abuse would not indicate how much of that abusive material is online. People might see abusive material and be offended by it without specifically being the victim of it and not be counted within that 2%.
The other possibility is that it is our assumptions that need changing - online abuse may not be as widespread as it sometimes seems.
Convictions
Survey results are important because they show the prevalence of issues like online abuse that many victims may not report out of fear that it will not be treated seriously.
But offensive comments online are not outside the law. Last month, following a public consultation on the issue, the Crown Prosecutions Service issued guidelines for prosecutions involving social media communications.
Even without this guidance, prosecutions were taking place. Last year for example, a law student was sentenced to two years' community service after racially abusing a football commentator on Twitter and a teenager was jailed for 12 weeks for posting explicit comments and jokes about April Jones on his Facebook page. But, in the absence of this being a specific category in criminal statistics, these cases don't amount to indicators about the prevalence of online abuse.
Last year, statistics were released by 29 UK police forces after a freedom of information request. They showed that complaints to police about alleged crimes linked to the use of Facebook and Twitter had increased by 780% in four years, resulting in about 650 people being charged in 2011. They also showed that in 2012, there were 4,908 reports in which Facebook or Twitter were a factor.
Counting online references
We could try to look at tweets ourselves and come up with some data on threats online - but simple textual analysis can't capture sarcasm, double negatives or context and make it hard to accurately gauge sentiments.
There are some individuals that are trying to come up with more sophisticated methods. Among them is Sam Martin, a Digital Sociology student at Goldsmiths College who is researching abusive behaviour on Twitter.
Last year, he decided to track the reaction on Twitter after a Sheffield United player was jailed for raping a 19-year-old woman. Looking at 1,500 tweets in the space of two days, he used basic semantic analysis to identify the proportion of tweets that mentioned the word 'rape' with negatively (20%), positively (40%) and neutrally (40%) as well as the word 'victim' (68% neutral, 16% negative and 16% positive).
We can't say how reliable those numbers are but his research is very interesting for another reason. Martin distinguished between the number of Twitter users who mention a topic just once or twice and those that tweet about it 20 or more times (the pink circles below). What the animation shows is that infrequent users who don't stay long on a given topic far outnumber those that do (passionate campaigners and trolls both fall into the latter).
So what can we say?
So far, we haven't got any statistics from Facebook or Twitter themselves about the proportion of users that report abuse/are reported for being abusive or about the proportion of content that is reported as such. More transparency from social media companies would certainly help inform the debate, as would proper academic study on the prevalence of threats online.
We'll update if more information becomes available.
Can you do more with this data?
• Download the full spreadsheet
• Contact us at data@guardian.co.uk
• Follow us on Twitter
• Like us on Facebook
Twitter
Social media
Feminism
Women
Crime
Internet
Police
Facebook
Mona Chalabi
© 2013 Guardian News and Media Limited or its affiliated companies. All rights reserved. | Use of this content is subject to our Terms & Conditions