Language on social media is honest, instant and, sometimes, inappropriate. And the Internet’s elephantine memory holds on to every last questionable comment.
Researchers at Humboldt State University used Twitter’s geo-coding to tag over 150,000 tweets from the better part of the past year to map what they call the “Geography of Hate,” an interactive database of hateful language all over the US.
Dr. Monica Stephens of Humboldt State began collecting tweets with the help of the University of Kentucky’s tweet aggregator the DOLLY project, beginning in June of 2012 and ending last month. The tweets collected contain 10 specific insensitive terms, and research assistants reviewed every single tweet to ensure the usage was intended to offend – blue represents "some hate," while red represents the "most hate."
Tweets fell in three categories for hate speech – homophobic speech, racist speech and derogatory tweets towards those with disabilities.
To protect privacy, the tweets were culled at the county level, with every county sharing a baseline national average of questionable language on Twitter.
While Travis County is a dark gray, indicating it produces no more hateful tweets than the national average, parts of Texas do contain pockets of prejudice. Those red pockets, however, shouldn't be taken to mean that every tweeter in those areas producing hateful tweets. The reason is in the numbers.
The study uses 150,000 tweets from across the entire country and it's not a per capita measurement. Some of the more "hateful" areas of the country are listed as such because of the tweets of only a handful of users, or in some cases only one. Keep in mind, 150,000 is a small swath of tweets compared to the average of 400 million tweets per day.
Still, the research does make for an interesting visualization of where some cluster of hurtful speech are appearing online. You can check it out for yourself by clicking here for the full map.