6 charts: Twitter’s troll problem by the numbers

Trolls have always been a nuisance for Twitter. But in recent weeks, they became particularly problematic for the platform when Disney and Salesforce each pulled their sale bids due to concerns over bullying.

Although trolls are an obvious problem for Twitter, it’s difficult to quantify their pervasiveness. Tweeted epithets might be ironic, pop culture references or even terms of endearment in some cases, which makes tracking trollish behavior very difficult. While assessing the prevalence of trolls has its challenges, researchers have shown that misogynistic language is quite common, some Jewish journalists have received an overwhelming amount of verbal abuse and current events can trigger xenophobia on the platform.

Here are six charts that add some numerical context to Twitter trolling.

Misogyny
For multiple categories of hate speech, Brandwatch tracked how frequently a handful of common slurs were tweeted this past month. Misogynistic language was used in about 15,000 tweets per day, which was much more common than other epithets.

brandwatch-over-time
Source: Brandwatch

While many of these tweets probably did not involve overt trolling (and some tweets were likely sent by bots), the data at least show how prevalent feminine pejoratives are on Twitter. Brandwatch research manager Edward Crook said that misogynic language “is more normalized” in the English language, so it shows up more often in tweets than do other slurs.

On a more encouraging note, a four-year Brandwatch analysis found that in recent years, discussions about misogyny have become more prevalent than misogynistic insults. Interestingly, the misogynistic insults came from a pretty equal split of men and women.

brandwatch-misogyny
Source: Brandwatch

Anti-Semitism
Last week, the Anti-Defamation League (ADL) released a report that examined anti-Semitic tweets. The ADL found that from Aug. 1, 2015 through July 31, 2016, there were about 2.6 million tweets containing anti-Semitic language. The ADL also mined 19,253 “overly anti-Semitic tweets” that were aimed at 800 journalists. They found that a whopping 83 percent of these tweets were aimed at just 10 journalists, all of whom are Jewish.

As seen in the chart below, many of the anti-Semitic tweets aimed at these journalists were related to the presidential election. The word clouds show which words were used most frequently in a derogatory way.

adl
Source: ADL

Islamophobia
Think tank Demos found that world events such as the bombings in Brussels and Brexit led to spikes in discussions about Islamophobia in the U.K.

islamophobia-1
Source: Demos

Fortunately, the majority of the tweets were supportive and not derogatory. However, in July 2016, there were still 289 Islamophobic tweets sent per hour in the U.K.

islamophobia-2
Source: Demos

Correlations
Brandwatch also examined discriminatory language across states. It found that states with high levels of misogyny also have high levels of racism, and that states with more self-identified Republicans tend to have high levels of transphobia. Keep in mind that any correlation above 0.5 is generally considered to be a strong relationship.

us-correlation-table-pg43
Source: Brandwatch
https://digiday.com/?p=206157

More in Media

Digiday+ Research: Publishers expected Google to keep cookies, but they’re moving on anyway

Publishers saw this change of heart coming. But it’s not changing their own plans to move away from tracking consumers using third-party cookies.

Incoming teen social media ban in Australia puts focus on creator impact and targeting practices

The restriction goes into effect in 2025, but some see it as potentially setting a precedent for similar legislation in other countries.

AI Briefing: Amazon’s new Nova models boost AI model efficiency, accuracy and variety across AWS

One of the most buzzy debuts was Nova, a suite of six new AI models that include understanding and creating text, images and videos.