My name is Sam Martin, I am a CAST MSc Digital Sociology student at Goldsmiths College, University of London. The MA/MSc Digital Sociology course is an innovative new course which integrates sociological theory and methods with practical expertise in digital technology.

My research area studies abusive behaviour on Twitter in the form of tweets and re-tweets and whether this can change the behaviour of users and how they interact on Twitter.

This blog will be used to communicate the initial stages of choosing my research question, structuring my blog, setting up my research tools, analysis, and discussions around my research and findings.

A discussion of the key points of my research question and how I formulated it can be found under the research cloud below and throughout this blog.

Research Question

Word cloud of tweets from my @intothedigital Twitter account. Made with Twitterstats.com and Wordle.net


Over the last year, I have noticed more and more media reports on key public figures reporting abusive tweets posted on their Twitter accounts when they have been linked to or engaged in high profile and sometimes controversial activity, and people have responded by tweeted abusive comments to their Twitter page.

The usual pattern of response in these high profile cases has been for the abusee to post a reply to the tweet(s), for the abuser to respond in turn with more abuse, and the media picking up on and reporting this. In some cases, the initial response is a shocked, but moderated one, but in others, it can be aggressive rather than placating, with an ensuing war of words coming about.

Below is an example of media and user commentary on twitter abuse, re-tweeted through my research-based Twitter account:
An example of media and user commentary on twitter abuse.

Based on these occurrences and the interesting press coverage of these interactions, I wanted to do a study of how the average (rather than high profile) user responds to abusive tweets, and if this does in fact change immediate and future behaviour of the main Twitter user.

I want to start off by mapping the sentiment of tweets, and if possible the two-way interaction between abuser and abusee. Having done that, I would then monitor the account about 5 days after the episode to spot if behaviour had in any way changed – if the user had become more cautious in his or her tweets, had proceeded as if nothing had happened, or had later commented on the occurrence (for example if they had reported the abuser to Twitter, who had dealt with it, or if they had shut down their old account and setup a new one).

My Research Question is therefore:

On the Twitter platform, who abuses who, and how do they respond?

Ideas to consider within this question:

  • What is considered abusive behaviour on Twitter (words, aggression) – can we monitor this through Twitter scraping and sentiment analysis?
  • How do people deal with abuse (retweet with aggressive or placating response, report, no report, flaming-war)?
  • Are some areas/regions more abusive than others (this can be shown via a google map showing hotspots, via a Twitter Venn Diagram, or word-sentiment map)?
  • Can a study of abusive behaviour on Twitter help develop better filtering and reporting tools?

This blog will enable me to discuss and display my research process, methods and results.