Paedophile hunters: An online ethnographic examination of cyber vigilantism

Research Institution / Organisation

University of Portsmouth

Principal Researcher

Dr. Andy Williams

Level of Research

Professional / Work-based

Project Start Date

January 2019

Research Context

As this is a recent phenomenon there is little research that directly examines paedophile hunter groups. The principal investigator (PI) has previously discussed the developments within technology (e.g. Web 2.0 or user generated content) as having a dramatic impact on the sexual grooming of children in terms of the policing of child sex offenders in the community and public fears around paedophilia (Williams, 2017). 

The aim of this research is to develop new knowledge and understanding within these areas. 

The PI has undertaken preliminary work on hunter groups whilst developing online research methods under the umbrella term of online ethnography. This work includes non-participatory observation of hunter groups’ online activities through the creation of diary observations, downloading ‘sting’ tubes (from YouTube and Facebook), collecting online public comments/opinions on the hunter groups, and collecting Facebook ‘newsfeed’ posts. An initial qualitative analysis using NVivo has been undertaken of 280 YouTube videos and approximately 56,200 public comments. The preliminary analysis indicates that the label of ‘paedophile hunters as vigilantes’ is reductive and unhelpful as this is a complex and multi-faceted social phenomenon that requires further in-depth investigation. More specifically areas uncovered that are fruitful for further analysis are: 


(i) the similarities and difference between various hunter groups (75+ groups known);

(ii) the nature of sexual grooming discourse; 

(iii) the impact hunter groups activities have on criminal investigations and criminal trials; 

(iv) the impact that public naming and shaming has on the groomers, their families and the wider community and; 

(v) the impact public manning on social media has on how risk is managed by probation and police services within the arena of public protection. 


The PI presented initial findings on area (i) at the Annual Conference of the British Society of Criminology, University of Birmingham, July 2018. 

The rationale for this study is multi-fold and based around the need for systematic academic research on the topic. Firstly, as this is a relatively recent phenomenon, which uses social media platforms to both catch and publicize alleged sexual groomers of children, there is little academic research on the types of grooming communication that take place, or what the public think about these self-appointed cyber activist groups. Secondly, there has been no research on the impact that such activities have on policing online child sexual offences or how the ‘evidence’ collected by the hunter groups is used or impacts upon a subsequent criminal investigation. Research findings and methodologies relating to similar digital behaviors and communities may brought to bear to this topic – namely that frame the digital radicalization of extremists. Finally, there has been no research examining the impact that this cyber activism has on how known offenders’ risk management by the probation and police services is handled once public naming and shaming has taken place, or what impact this has on offenders and their families or the fact that it then may be possible to identify previous victims of these individuals.


Aim: 
An exploratory examination of child sexual grooming cyber activism using a mixed-methods approach of online ethnographic (digital anthropology) and quantitative data to produce a trend, cluster and thematic analysis.


Objectives:

  1. Developing (proof of concept) a quantitative database for conducting trend and cluster analysis of social media data of hunter groups cyber activism and the public’s responses.

  2. Thematic analysis of social media data of hunter groups cyber activism and the public’s responses.

  3. Development of a research and training package for criminal justice professionals who deal with the aftermath and impact of online sexual grooming cyber activism.


Research Methodology

As this is an exploratory mixed-methods analysis of online activism and social media data, there is no hypotheses being tested. Instead, and following the tradition of ethnographic methods (Williams, 2014), specific questions will guide the research and will be developed using the ground theory approach (e.g. Braun & Clarke, 2006). Overall, the aim is to develop knowledge on this recent social phenomenon, to inform debate and policies regarding cyber activism and their engagement with communities and criminal justice agencies and provide information and training advice regarding online child sexual grooming and offending and public protection.   

This study will be analysing the hunter groups’ cyber activism by using social media data that will be retrieved from key sources such as Facebook, YouTube and Twitter. This is a mixed methods approach that uses online ethnographic methods to collect qualitative data such as public sting videos, Facebook news feed posts and comments and YouTube comments. This data will then be mined to produce a quantitative dataset for analysis and interpretation, which will sit alongside the thematic analysis from the online ethnographic material. Data will be analysed using text-mining method using thematic, topic modelling, clustering and social network analysis. Advanced data visualisation techniques will be used to map out the activities, online comments and their network. 


This research is interdisciplinary as it involves the broad topic area of criminology (cyber-crime and security, vigilantism and online sexual grooming of children), online qualitative methods (online ethnography) alongside the five-stage method of Ozcan and Islam (2017). The process can be summarized as follows: (1) data retrieval; (2) data cleaning and optimization; (3) data analysis; (4) data visualisation and; (5) data labelling and interpretation. Each stage is significant to the results, and for increasing validity and reliability.

Stage 1 is a continuation of the PI’s online ethnography that will continue to collect data on sting videos, news feed posts and public comments. Based on the PI’s preliminary search results, approximately 300+ YouTube and Facebook sting videos, 20,000+ YouTube and Facebook public comments and at least 250+ Facebook news feed posts will be examined (although this number could increase with additional research). This data spans at least the last 5 years but a clearer timeframe will be formalised later. 


Stages 2 to 5 takes place after data retrieval. Due to the large amount of data, various text mining and big data techniques are needed to merge the data and use dimensionality reduction for preparing the data to be examined. Subsequently the data will be cleaned and optimised, with the help of the relevant packages available in R language, Python and VantagePoint (software available within BaL). In data analysis stage, text mining, social network analysis and statistical analysis will be used to examine key patterns and relationships such as using TF-IDF, co-occurrence and centrality measurements. The results will be converted into the visual representations such as heat-maps and social network maps. After visualisation, special labelling techniques will be used to identify key themes, issues and their intersections. This will be completed through a semi-automated feedback mechanism. To interpret results further and identify the scope of fields experts will be used, allocating key words to each cluster. 

Finally, interpretation will be conducted using the results of the labelled visuals by examining key elements such as centrality measures. All the above relates to objectives 1 and 2.


Interim reports and publications

​Not available

Date due for completion

July 2020
Return to Research Map