'Geography of Hate' Maps 'Racist,' 'Homophobic' Tweets

May 14, 2013— -- The expansion of online social media sites has brought the emergence of the "plugged in" human. And all this online free flow has been grist for the mill for researchers, who've used it to gauge attitudes and how individuals interact around the world.

A new Twitter study by assistant professor Monica Stephens and Humboldt State University set out to measure hate via tweets in the United States. The result? A "Geography of Hate" map.

"This project began as a way to engage Humboldt State students around maps, big data and online social media," Stephens told ABC News. The results are displayed as an interactive, online "heat" map.

A drop down menu at the top of the page can filter the "hate" tweets by category: homophobic, racist, disability. These categories can be refined even further by such keywords as "homo," "queer," "cripple," "wetback."

Spots of the map that display the highest concentration of hate, according to the study's findings, glow bright red.

Identifying these hot points of "hateful" tweets required more than searching posts for keywords, Stephens said. The researchers examined pockets of tweets to determine negative or positive use of potentially explosive terms.

"The phrase 'dyke,' while often negative when referring to an individual person, was also used in positive ways (e.g., 'dykes on bikes #SFPride')," Stephens wrote in the project's blog. "The students were able to discern which were negative, neutral or positive."

To ensure the study was targeting concentrations of hateful tweets, and not just high totals, the researchers "normalized" the numbers to best equally represent large cities with more Twitter users against smaller towns, where there would be fewer people on Twitter, Stephens explained.

"These numbers are normalized by the amount of activity on Twitter, which counters the disparity in population. As Twitter is often used on smartphones and people tweet from counties where they don't live, it wouldn't make sense to normalize by population, but rather by the total number of tweets in an area," Stephens said.

"Wyoming had very few tweets overall and also had very few hateful tweets. It appears that the 'N' word was tweeted a small amount in four different counties of that state but not enough to create a cluster of activity."

READ: Twitter Study Measures Day to Day Happiness of Last 5 Years

When looking at the "heat" map of tweets, the study found more of a clumping of hot spots than an equal disbursement throughout the country. In the project's blog, Stephens wrote that "quite depressingly," there were a number of pockets of concentration that demonstrated heavy usage of the 'N' word.

"It's not surprising that different regions would have different attitudes toward certain groups," said Jay Van Bavel, an assistant professor of psychology at New York University who studies social and moral judgment.

"Each region has a set of social norms that influence what people feel about certain groups and how appropriate it is for them to express those feelings publicly," Van Bavel told ABC News.

Stephens cited academic studies that found regional characterizations similar to those on the "Geography of Hate" map. "The clustering of hate speech in certain areas corresponds to peer-reviewed literature about organized hate groups that implies they are more likely to exist in areas of low population, areas of lower property taxes and [that have] a larger number of African-Americans," Stephens said.

One 1999 study, "On the Geography of Hate," by Philip N. Jefferson and Frederic L. Pryor published in the journal Economics Letters, suggested that economic or sociological explanations for the existence of hate groups in a particular area were far less important than other historical conditions.