Testing a probabilistic model of desistance from online posting in a right-wing extremist forum: distinguishing between violent and non-violent users

ABSTRACT

Little is known about online behaviours of violent extremists generally or differences compared to non-violent extremists who share ideological beliefs. Even less is known about desistance from posting behaviour. A sample of 99 violent and non-violent right-wing extremists to compare their online patterns of desistance within a sub-forum of the largest white supremacy web-forum was analysed. A probabilistic model of desistance was tested to determine the validity of criteria set for users reaching posting desistance. Findings indicated that the criteria predicted “true” desistance, with 5% misidentification. Each consecutive month without posting in the sub-forum resulted in a 7.6% increase in odds of posting desistance. There were no significant differences in effects for violent versus non-violent users, though statistical power was low.

KEYWORDS:

• Life-course criminology
• extremism
• desistance
• violence
• online communities

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Research from other fields has similarly indicated the relevance of online behaviours for real-life outcomes Ahmad et al. (2019); Meshi and Ellithorpe (2021); Reinhard et al. (2020).

2. By former violent extremists, we refer to individuals who, at one time in their lives, subscribed to and/or perpetuated violence in the name of a particular extremist ideology and have since publicly and/or privately denounced violence in the name of a particular extremist ideology. In short, they no longer identify themselves as adherents of a particular extremist ideology or are affiliated with an extremist group or movement.

3. Data collection efforts followed the proper ethical procedures for conducting research involving human participants. Here, the former extremist was informed that their participation in the study was entirely voluntary. They were also informed that they had the right to decline to answer questions or to end the interview/withdraw from the study at any time. In addition, the former was informed that they would not be identified by name in any publication, and that all data collected from the interview would be de-identified for the purpose of ensuring participant anonymity. One in-person interview was conducted with the former in June 2017 and was approximately 10 hours in length. The interview was audio-recorded and transcribed.

4. There were concerns regarding the inherently biased nature of the sample due to selection into the sample based on one informant’s knowledge of user involvement in violent behaviour or lack thereof. As it pertains to the bacon function and limiting outliers in the data, there are some concerns that identified outliers may not actually have been outliers because users not included in the study may have also had a larger number of each type of post that would have resulted in there being no such outliers in a complete dataset comprised of all users. This may have resulted in differences in effects of interest identified here because there would be no need for management of outliers. However, a full understanding of the impact in this regard was beyond the scope of this study because of the inability to identify the violence status of all users in the sub-forum.

Additional information

Notes on contributors

Thomas W. Wojciechowski

Thomas Wojciechowski is an assistant professor with the Michigan State University School of Criminal Justice.

Ryan Scrivens

Ryan Scrivens is an assistant professor with the Michigan State University School of Criminal Justice.

Joshua D. Freilich

Joshua Freilich is a professor with the John Jay College of Criminal Justice.

Steven M. Chermak

Steve Chermak is a professor with the Michigan State University School of Criminal Justice.

Richard Frank

Richard Frank is a professor with the Simon Fraser University School of Criminology.