An investigation into the perpetration of online sexual harassment: Who are the perpetrators, and what are the predictors of sexual harassment online?

This thesis investigates the pernicious problem of online sexual harassment (OSH), which can lead to financial and emotional consequences for victims online. The salient aspects of online spaces, such as anonymity and lack of geographic boundaries, enable perpetrators to find and harass victims. The research has often focused on victims, which means little is known about perpetrators of OSH, so this thesis addresses this gap by investigating the prevalence, motivations, and predictors of OSH perpetration. A systematic literature review and meta-analysis found that pooled perpetration rates for image-based OSH were 12%, while victimisation rates were 7% for image-based OSH and 27% for text-based OSH. The literature review also highlighted the lack of a tool to measure OSH perpetration behaviours, and this was addressed through the development of the ten-item OSH proclivity scale, which measured text and image-based OSH proclivity. The thesis also examined the motivations of OSH perpetrators and identified two broad groups of motivations, which were labelled as "malicious" and "insensitive." The insensitive motivations can include normalisation and minimisation of harm of OSH against victims, while the malicious group contained examples of revenge, control, or coercion of victims. An inductive content analysis of N = 496 messages and posts from OSH perpetrators further highlighted the different motives for text-based OSH, which consisted of threats to masculine identity and romantic aggression. A survey of N = 514 adults aged 18-70 years examined individual, social and environmental factors associated with OSH proclivity. This analysis found that OSH proclivity was higher for men and in younger age groups. A series of logistic regression models identified that significant predictors of OSH were imitation, definitions, sexting, psychopathy, and toxic disinhibition. A latent class analysis identified four distinct OSH behavioural classes of "malicious", "insensitive", "sexist", and "no proclivity". There were gender differences within the classes, with men more likely to be in the malicious and insensitive classes and women more likely to be in the non-perpetrator and sexist classes. The thesis concluded that educational awareness of OSH, a focus on safe sexting, and the removal of harmful online content would be useful for preventing OSH in online spaces.