Inter-ideological mingling: White extremist ideology entering the mainstream on Twitter

ABSTRACT

This case study explores the convergence of white extremist political ideology with mainstream political ideology on the micro-blogging platform Twitter – a phenomenon termed “inter-ideological mingling”. Exploring the spread of white extremism in the digital environment can provide insight into the growth of hate groups in the physical environment. A sample of 4800 tweets was examined through hierarchical cluster analysis and textual analysis. Several pieces of evidence were found supporting inter-ideological mingling. Cluster analysis shows that extremist terms are not isolated from terms found in mainstream political discourse. Textual analysis of individual tweets provides evidence for five strategies of inter-ideological mingling: joining, blending, piggybacking, backstaging, and narrating.

Notes

¹The digital environment is an all-encompassing term for the “social space produced through interconnected information and communication technologies” (Graham 2014). Twitter is a part of the digital environment.

²Microblogging is the process of composing small amounts of content online for public consumption. This process is usually facilitated through a web application such as Twitter.

³Twitter is a microblogging platform that allows users to publish content, called “tweets,” of up to 140 characters. These tweets may also include images and links to websites. Tweets are by default public; however, users can change the settings of their account to make their tweets private.

⁴Tweets denote the published content, as well as the act of publishing: Person A may be “tweeting” his reflections on person B’s tweet.

⁵For example, a user can tweet this phrase: “Don’t forget to vote this Tuesday. #votenow”. #Votenow acts both as a component of the tweet but also as a type of symbolic anchor, used by others. And so a second tweet may use the same hashtag, but in a different tweet or context. For example: “Why are there so many people talking on their cell phones at the polling station? #votenow”. A search for #votenow, would list both tweets.

⁶All analyses were done using the R statistical programs and packages for cluster analysis and tweet collection.

⁷The author has explored several extremist websites prior to this research. These websites include Stormfront (www.stormfront.org), American Renaissance (www.amren.com) and the Institute for Historical Research (http://www.ihr.org/).

⁸Interpretations of hashtags are from the context of the tweets in which they were found and from https://tagdef.com/.

⁹All cleaning of the data was done through the QDAP text mining package (http://cran.r-project.org/web/packages/qdap/index.html).

¹⁰QDAP’s common words are drawn from Fry (1997).

¹¹The majority of terms were removed by setting a level of sparsity for the matrix. The parameter selected was .98, meaning that terms were retained that appeared in at least 2% of documents.

¹²P2 was not one the final terms in the analysis. Also, two terms “amp” and “edm” could not be categorized. These terms are found on Twitter as both usernames and hashtags. It is possible that because of the R package used quotes are mistranslated into terms. For example, the term “amp” may be simply #.

¹³There are several ways of measuring similarity, or distance, between data points. The distance measure used here is Euclidean distance. Given a term-document matrix, where the terms are the words and the documents are the tweets, and the values in each cell are the number of times a term appears in a document, the distance between two terms is measured as: distance(x,y) = {i (xi − yi)2}½.

Additional information

Notes on contributors

Roderick Graham

Roderick G. Graham is an Assistant Professor in the Department of Sociology and Criminology at Old Dominion University. His current research explores the interplay between new media technologies and racial inequality.