Project
Defining intimidation: empirical evidence for an aggressive interpersonal phenomenon

Although intimidation is acknowledged as a universal behavioral phenomenon and a significant social challenge, the underlying psychological dynamics and principles are yet poorly understood (e.g., Lamontagne, 2010). Especially the insight in what scientifically defines intimidation, and therefore which psychological underpinnings and their mutual dependencies influence the effectivity of intimidating behavior, seems to be lacking. Previous studies, generally originating from the fields of Intimate Partner Violence (IPV) and Commercial Sexual Exploitation (CSE), typically reported two types of predominant intimidation strategies: physical (physical and/or sexual violence, e.g., Cleveland et al., 2003; DeGue et al., 2010) and non-physical methods of intimidation. The latter typically covering verbal aggressiveness (i.e., menacing and scrutinizing) and/or social isolation (e.g., García-Sancho et al., 2017; Zavala & Guadalupe-Diaz, 2018).

To enhance insights into the contributions of these psychological underpinnings (both independently and interrelated) on the aforementioned types of intimidating strategies, a Bayesian Network will be created on an extensive body of published statistics, primarily associated with Intimate Partner Violence (Kallen et al., 2021). This theoretical intimidation model might then be able to estimate the contributions of networks of pre-identified risk factors on the likelihood that i) one will engage in specific intimidating behavioral strategies; ii) one will be vulnerable for specific intimidation techniques; iii) intimidation will be effective, given the combination of certain characteristics being apparent of the instigator and receiver.

The phenomenon of intimidation has a unique complexity wherein it is highly likely that the underlying mechanics (and data) will be non-linear. In addition, the sort data that will be extracted from police files for the validation of components of the theoretical model is likely to consist out of non-random missing data. Subsequently, both complex issues call out for creative approaches with machine learning techniques (e.g., random forest, XGBoost) to fundamentally investigate the hypotheses within the project. It therefore seems that these types of approaches, in addition with the Bayesian networking based on published statistics, fit the format of psychometric methods and therefore an IOPS project.

Supervisors
Prof. Eric-Jan Wagenmakers
Dr. Ilja Sligte
Dr. Victor Kallen

Financed by
TNO and the Dutch ministry of Justice and Safety

Period
1October 2021 – 30 September 2024