Prediction model - DV Knowledge

Violence prediction model prototype

The TACTICS project has developed a violence prediction model based on Estonia, analyzing the anonymous data on victims and perpetrators provided by the Women’s Support and Information Center.

The logistic regression model assesses the likelihood of a perpetrator’s prior convictions based on various socio-economic characteristics, such as age of victim and perpetrator, number of children, relationship and employment status, level of education, nationality, language of communication, perpetrator’s substance abuse and stepparenting role.

It is important to note that this model was developed using a very limited Estonian dataset. Due to the small sample size and the narrow geographical and socio-cultural scope, the results should be interpreted with caution. The findings may not be generalisable beyond the specific context in which the data were collected, and there is significant room for improvement through the inclusion of larger, more diverse datasets and further model refinement. As such, this model should be understood as a prototype with clear potential, rather than a final or definitive predictive tool.

Understanding the factors that increase the likelihood of repeated violent behavior can help predict the escalation of violence in families with specific characteristics, enabling earlier interventions.

Results indicate that older victims, Estonian perpetrators, and employed perpetrators are less likely engaged in repeated domestic violence, while step-parent perpetrators are more likely to have prior convictions.

Specifically, victims in older age groups exhibit

0 %

reduction in the likelihood of encountering repeated domestic violence.*

The 27% lower likelihood among older victims does not necessarily mean that violence happens less often to them. It may reflect that older victims report domestic violence much more rarely. Many elderly people depend on the perpetrator financially, physically, or emotionally, which can make reporting very difficult. In some cases, the violence may also have been happening for many years (even decades) and is therefore less likely to appear in official records as repeated recent incidents.

Furthermore, Estonian perpetrators are

0 %

less likely to engage in repeated domestic violence compared to their non-Estonian counterparts.

Additionally, employed perpetrators demonstrate a

0 %

lower likelihood of repeating domestic violence.

Conversely, individuals in a step-parenting role are

0 %

more likely to have prior convictions for violent actions.

Domestic Violence Prediction Model Prototype

Answer the questions below regarding your DV case to calculate the probability of repeated violence.

1) What is the age of the victim?

< 26 years old 26–35 years old 36–45 years old > 45 years old

2) What is the age of the perpetrator?

< 26 years old 26–35 years old 36–45 years old > 45 years old

3) Is the perpetrator representing a national minority?

Yes No

4) Are there children in the family?

Yes No

5) Does the perpetrator have a job?

Yes No

6) Does the victim have a job?

Yes No

7) Relationship status between victim and perpetrator

In relationship (partner) Divorced/separated Other (e.g., parent-child)

8) Is the perpetrator a stepparent?

Yes No

9) Has the perpetrator been previously convicted?

Yes No

10) Does the perpetrator have substance abuse issues (alcohol, drugs, gambling)?

Yes No

11) Level of the perpetrator's education

Higher education Secondary education Primary education

Co-funded by the CERV Daphne EC Program. Grant Agreement no. 101096908