sheep floating tweets geolocated blog containing racist in preparation for the U.S. elections. How patterns vary from one state to another? • More data journalism and data visualization The Guardian

• Get Data

This article was published in the blog floating sheep, and was written by Matthew Zook, Mark Graham, Ate Poorhuis, Monica Stephens, and Taylor Shelton.

During the day after the 2012 presidential election was noted an increase in hate on Twitter in reference to the re-election of President Obama, as a chronicle of Jezebel (thank you Chris Van Dyke by brought this issue to our attention.) This is a useful reminder that technology reflects the society in which it is based, both good and bad.

information space is not separated from daily life and racism extends into the GeoWeb and helps contour shapes, and in turn, the data can be used to geoweb reflect the geographical distribution racist practice again in places that have been raised.

Using DOLLY
collected all geocoded tweets from the last week (from 1 November) with racist terms also refer to the elections in order to understand how these daily acts of overt racism spatially distributed . Given the nature of these search terms, we buried the details at the bottom of this message in a note. [1]

Given our interest in the geography of the information that we wanted to see how this kind of hate speech superimposed on the physical space. In addition there were 395 tweets per state hatred and then normalized by comparison with the total number of geocoded tweets out of this state in the same period [2].

We used a measure based location quotient (LQ) indicating the participation of every State in the election hate speech on Twitter the total number of tweets. [3] A score of 1.0 indicates that the state has roughly the same number of messages twitter hate speech that the total number of tweets. Scores above 1.0 indicate that hate speech is more common than all the tweets, which suggests that the State "twitterspace" contains more post-electoral racist tweets than the norm.

Therefore, these messages are fairly evenly distributed twitter? Or is it that some countries are more racist specializations in twitter? The solution shown on the map below (also available in an interactive way here) in which the individual location tweets (indicated by red dots) [4] are superimposed on the color-coded states.

yellow shading indicates states with a relatively small amount of hate tweets after the election (compared to general tweeting habits) and all states shaded green a greater amount. More green as the location quotient greater than hate tweets.

These are the conclusions of this analysis:

• Mississippi and Alabama have the highest LQ measures with scores of 7.4 and 8.1, respectively.

• Other southern states (Georgia, Louisiana, Tennessee) around these two central states also have very high scores and LQ are characteristic enough in the southeast.

• The prevalence of racist tweets after the election is not strictly a phenomenon southern North Dakota (3.5), Utah (3.5) and Missouri (3 ) have very high TC. Other states, like West Virginia, Oregon and Minnesota did not score as high, but they have a relatively high number of hate tweets to Twitter using generally suggest.

• The Northeast and the West Coast (except Oregon) have a relatively small number of hate tweets. • gray States did not hate geocoded tweets in our database. Many of these states (Montana, Idaho, Wyoming and South Dakota) have relatively low levels of use of Twitter too. Rhode Island has a greater number of geocoded tweets, but did not hate tweets that we could identify.

consider that we measure the tweets instead of users and therefore, a person may be responsible for many twitter messages and in some cases (particularly in North Dakota, Utah and Minnesota) the number of tweets of hatred is small and QL is high due to the relatively small number of tweets in general.

However, these results support the idea that there is a strong enough group of hateful tweets focused on the U.S. southeast, which has a much higher rate than the national average.

But also for anyone to become too complacent, the sad reality is that most states are not immune to this type of activity. Racist behavior, especially directed towards African-Americans in the United States, it is easy to find offline and in the information space.

The following table shows the values ??of location quotients hate tweets after the elections.

summary data

notes

[1] Using the examples of tweets, narrated by the blog Jezebel we chose the tweets containing the text "monkey" or "black", and also contains the text "Obama" O "re-elected" O "won." A rapid and disturbing results of the research showed that it was a good match for our target of hate speech related to elections. We ended up with a total of 395 on the worst tweets might imagine. And since we are talking about the Internet, this is not saying much.

[2] To be precise, a sample was 0.05% of all tweets geocoded November 2012 statewide aggregates.