Removing Biases in Communication of Severity Assessments of Intimate Partner Violence: Model Development and Evaluation

BACKGROUND: To support a victim of violence and establish the correct penalty for the perpetrator, it is crucial to correctly evaluate and communicate the severity of the violence. Recent data have shown these communications to be biased. However, computational language models provide opportunities for automated evaluation of the severity to mitigate the biases. OBJECTIVE: We investigated whether these biases can be removed with computational algorithms trained to measure the severity of violence described. METHODS: In phase 1 (P1), participants (N=71) were instructed to write some text and type 5 keywords describing an event where they experienced physical violence and 1 keyword describing an event where they experienced psychological violence in an intimate partner relationship. They were also asked to rate the severity. In phase 2 (P2), another set of participants (N=40) read the texts and rated them for severity of violence on the same scale as in P1. We also quantified the text data to word embeddings. Machine learning was used to train a model to predict the severity ratings. RESULTS: For physical violence, there was a greater accuracy bias for humans (r(2)=0.22) compared to the computational model (r(2)=0.31; t(38)=–2.37, P=.023). For psychological violence, the accuracy bias was greater for humans (r(2)=0.058) than for the computational model (r(2)=0.35; t(38)=–14.58, P<.001). Participants in P1 experienced psychological violence as more severe (mean 6.46, SD 1.69) than participants rating the same events in P2 (mean 5.84, SD 2.80; t(86)=–2.22, P=.029<.05), whereas no calibration bias was found for the computational model (t(134)=1.30, P=.195). However, no calibration bias was found for physical violence for humans between P1 (mean 6.59, SD 1.81) and P2 (mean 7.54, SD 2.62; t(86)=1.32, P=.19) or for the computational model (t(134)=0.62, P=.534). There was no difference in the severity ratings between psychological and physical violence in P1. However, the bias (ie, the ratings in P2 minus the ratings in P1) was highly negatively correlated with the severity ratings in P1 (r(2)=0.29) and in P2 (r(2)=0.37), whereas the ratings in P1 and P2 were somewhat less correlated (r(2)=0.11) using the psychological and physical data combined. CONCLUSIONS: The results show that the computational model mitigates accuracy bias and removes calibration biases. These results suggest that computational models can be used for debiasing the severity evaluations of violence. These findings may have application in a legal context, prioritizing resources in society and how violent events are presented in the media.