By Ellen Airhart
On October 10, 2015, Whitney Beall was arrested after broadcasting a video on Periscope, Twitter’s live-streaming app. The footage featured her driving, slurring her words, and making statements that suggested she might have had a few alcoholic beverages (“I'm driving home drunk”), according to ABC News. Police used the Periscope video to track Beall down and arrest her.
It was a striking example of how social media could be used to enforce the law in an individual case. But scientists are also harnessing social media to learn about potential wrongdoing on a larger scale. At the Social Data Science Lab at Cardiff University, they are using large quantities of Twitter data to analyze crime trends and test theories about where crime might occur in the future. “The growth of social media has presented potential opportunities to address criminological questions at hitherto unrealizable scale and speed,” says Matthew Williams, the lab director, in an email.
Historically, legislators placed crime-fighting resources in areas where they noticed patterns of crime. The broken windows theory, which posits that smaller crimes and vandalism normalize larger crimes, was deeply embedded in the American psyche after it was published in 1982, although it has not always been supported by research. The Cardiff lab wanted to test this theory in the Twittersphere. They tracked neighborhood maintenance requests that Twitter users sent to local city council members and found an association between these mentions and crime rates in various London boroughs, consistent with what the theory would predict.
Twitter, of course, is much more than a haven for anxious citizens concerned about the shape of their neighborhoods. Of relevance to the Cardiff team’s efforts, it’s also a place where people with hateful attitudes find each other. After the 2013 terrorist attacks in London, the researchers decided to use social media to study potential backlash against Muslims. They expected hate speech to reverberate widely after the attacks. However, that was not what they found. “The hate did not propagate in terms of size (via retweets),” Williams explains. “Instead, it was produced and shared amongst a small group of like-minded individuals that associated with right-wing groups.”
They have taken this research on hate speech stateside. In 2016, the Cardiff lab began a study in Los Angeles that is looking into the relationship between racist tweets and offline violence. The study won’t conclude until 2019. “If we find an association, the project will provide some evidence that social media may offer an alternative source of information on the hate crime problem in the U.S.,” Williams says.
There are limitations to this approach. For example, no social media site provides clean data—pure information without outliers, duplications, or irrelevancies—which means a lot of added work for the researchers. Facebook data is more expensive and harder to come by, and comparable Instagram numbers are nonexistent. The Cardiff lab and other researchers studying crime and social media usually use Twitter, which offers 1 percent of all public tweets for free. While law enforcement officials looking for private Twitter information would need a subpoena, court order, or emergency access request, according to Twitter guidelines, Twitter users consent to have researchers use their public tweets by accepting Twitter’s privacy agreement. (“Of course, nobody reads the privacy agreement,” notes privacy specialist Bruce Schneier.)
Despite the limitations, researchers are excited to seize on the potential of Twitter and other platforms to help them better understand and predict patterns of criminal activity. “Big social data are transforming the way social scientists approach research designs,” says Williams. “Any study of contemporary society must consider the role of the Internet in everyday life.”
Ellen Airhart is an editorial intern at Psychology Today.