Module 7 – Using Data & Digital Tools to Improve Responses to Domestic Violence

Welcome! By the end of this Module you will: 

  1. Understand the basics of predictive policing & its application in DV cases.
  2. Learn how criminal justice & multi-agency data can improve DV investigations & risk prediction.
  3. Explore current digital tools & their potential for enhancing DV responses.

Predictive Policing & Modern Risk Assessment

Computer-based Predictive Policing

Modern Risk Assessment in Criminal Justice

Main challenges

Enhansing DV Risk Assessment

Main Methods, Impact & Challenges

Importance of Data for DV Responses

Examples of different solutions

Successful Case Study from Spain: VioGén II System

Open Source Intelligence for DV Cases (Open Source Intelligence, OSINT)

Mobile Apps

Electronic Bracelets

From Reactive to Proactive & Predictive Policing

Traditional Reactive Policing:
Focuses on responding to crimes after they occur.

Proactive Policing:
Uses data to allocate resources (e.g., increased patrols in high-crime areas), deterring crime & improving response times.

Predictive Policing: Analyzes historical data to identify high-risk times & locations, revealing crime patterns & circumstances.

Key Benefits:

  • Optimizes police resources.
  • Enhances crime prevention strategies.
  • Enables more effective deployment of officers.

Examples:
Assaults after bar closures or burglaries during holidays.

Computer-based Predictive Policing

  • Computers help identify patterns in criminal justice data that would be difficult to spot manually.
  • Algorithms analyze variables (e.g., time, location, type of area) & determine their importance in predicting crimes.
  • With real-world data, algorithms adjust to improve accuracy & can uncover new predictors of crime.

Example:
Predicting robberies near a new bar may involve factors like closing time, neighborhood income, housing density, or nearby bars.

More data enhances prediction accuracy, aiding better police responses.

Modern Risk Assessment in Criminal Justice

Purpose:
Predicts the likelihood of crimes involving specific individuals to inform critical decisions, including sentencing, parole, probation, & protective orders.

Why It Matters:
More reliable predictions mean more effective resource allocation & improved protection for victims & communities.

From Traditional to High-Tech:

  • Traditional methods relied on simple, subjective scoring models.
  • Machine learning enhances accuracy by analyzing multiple variables & uncovering hidden patterns.

Practical Applications:

  • Categorizes risk levels (high, medium, low) for better decision-making.
  • Improves accuracy and fairness in assessing risks, benefiting both officers & public safety.
Source: Berk, R. A. (2021). Artificial intelligence, predictive policing, and risk assessment for law enforcement. Annual Review of Criminology, 4(1), 209-237.

Challenges of Predictive Policing & Computerized Risk Assessment

  • Accurate predictions don’t guarantee improved policing – it depends on how data is used & aligns with priorities.
  • Data flaws (e.g., underreporting or mischaracterization of crimes) can skew results, leading to over- or under-policing.
  • Privacy concerns arise from handling sensitive data.
  • There’s a risk of misuse, such as over-aggressive actions or suppression of dissent – think of the Sci-Fi movie Minority Report
Sources:
  • Berk, R. A. (2021). Artificial intelligence, predictive policing, and risk assessment for law enforcement. Annual Review of Criminology, 4(1), 209-237.
  • Grogger, J., Gupta, S., Ivandic, R., & Kirchmaier, T. (2021). Comparing conventional and machine‐learning approaches to risk assessment in domestic abuse cases. Journal of Empirical Legal Studies, 18(1), 90-130.
  • Messing, J. T., & Thaller, J. (2013). The average predictive validity of intimate partner violence risk assessment instruments. Journal of interpersonal violence, 28(7), 1537-1558.

Enhancing DV Risk Assessment

Purpose:
Assess danger levels to guide protective measures.

Methods:

  • Scoring Rule: Points-based thresholds classify risk.
  • Structured Judgment: Combines responses with professional judgment.

Impact:
Better tools ensure high-risk cases are identified, empowering officers to protect victims effectively.

Challenges:

  • Subjective scoring & inconsistent use reduce accuracy.
  • High-risk cases may be missed due to misadministration.

Technology Enhancements:

  • Machine Learning improves accuracy by analyzing criminal histories.
  • Combining tools like DASH with machine learning boosts violent DV recidivism prediction accuracy by 20%.
Sources:
  • Berk, R. A. (2021). Artificial intelligence, predictive policing, and risk assessment for law enforcement. Annual Review of Criminology, 4(1), 209-237.
  • Grogger, J., Gupta, S., Ivandic, R., & Kirchmaier, T. (2021). Comparing conventional and machine‐learning approaches to risk assessment in domestic abuse cases. Journal of Empirical Legal Studies, 18(1), 90-130.
  • Messing, J. T., & Thaller, J. (2013). The average predictive validity of intimate partner violence risk assessment instruments. Journal of interpersonal violence, 28(7), 1537-1558.

Importance of Data for DV Responses

DV often involves repeated, escalating violence, making context critical for effective police action (e.g., risk doubles by the 3rd call, triples by the 6th, & quadruples by the 8th).

Officers need comprehensive data, including non-criminal records, to improve responses.

Linked systems could provide access to criminal justice, social service, & health care data.

Examples: Patterns of injuries in medical records, victim support center reports, or substance abuse histories.

Examples of different solutions

VioGén Police Risk Assessment Protocol

Established in Spain (2007) to coordinate police responses to DV through the Internal Security Studies Group.

Involves multiple stakeholders: victims, aggressors, witnesses, technicians, & doctors.

Integrates National Police, Civil Guard, & local police forces.

In 2024 updated VioGen II System released.

Main Aims of the VioGen II System

Improved Coordination

Enhance collaboration between police, social services, & healthcare to provide a unified response to DV.

Accurate Risk Assessment

Refine risk tools to better predict violence recidivism & ensure proportional victim protection.

Personalized Protection

Tailor responses based on the evolving risk to each victim, adapting measures as needed.

Victim Empowerment

Provide clear information & personalized safety plans, ensuring victims understand their risk & protection options.

Data-Driven Decisions

Use advanced technology & algorithms to improve real-time decision-making & ensure accurate protection measures.

Guidelines of the Police Risk Assessment Protocol Include:

Source: González Álvarez, López Ossorio, Muñoz Rivas (2018)

Assessment of Risk of Recidivism of Violence (VPR) – Categories & Factors of Risk

The VPR indicators are grouped into 4 dimensions:

VPR – Risk Levels & Protection Measures

Cases are classified into five risk levels: Unappreciated, Low, Medium, High, & Extreme. Based on the risk classification, officers implement mandatory & optional protection measures, & these measures are designed to be proportional to the risk level.

Examples of protection measures include:

Low

Provide 24-hour support contact & periodic phone follow-ups.

Occasional surveillance & regular check-ins with the abuser.

Frequent surveillance & electronic monitoring of the abuser.

Constant victim surveillance & intensive control of the abuser.

Risk Evolution Risk Assessment Form (VPER) 4.0

The VPER includes 43 indicators:

34 for risk & 9 for protection, grouped into 5 dimensions:

  1. Severity of the incident
  2. Aggressor factors
  3. Victim characteristics
  4. Victim’s perception of risk
  5. Status of applied protection measures

Outcomes:

Positive Evolution: No new incidents.

Negative Evolution: New incidents, whether reported or not.

VioGén II System Gaps

While the VioGen II system has improvements over its predecessor, it still presents some potential gaps:

These gaps highlight areas for potential improvement, including enhancing flexibility, training, transparency, & communication within the system.

The Spanish VioGén system, while innovative in coordinating police response and providing risk assessments for gender-based violence cases, has notable gaps that impact its effectiveness. First, the system’s transparency and accountability are criticized. VioGén’s algorithm automatically assigns a risk level to each case, heavily influencing police protection measures without clear accountability. Police officers seldom deviate from the algorithm’s recommendations, which can limit professional judgment and lead to over-reliance on automated scoring, potentially affecting case-specific responses and victim safety.

Another significant issue is the VioGén questionnaire itself, which is used to gather information during emotionally charged moments. Victims often report their experiences immediately after an incident, which can lead to responses clouded by trauma or confusion. This has led to concerns about the reliability of the input data, as many women struggle to provide accurate details while under distress. The system’s design also includes limited training for officers on effectively communicating the questionnaire’s purpose, leading to inconsistencies in how it’s administered and understood by victims.

Moreover, there is minimal communication with victims about their assessed risk levels, as many victims report they are unaware of their designated risk category or the corresponding protection plans. Addressing these transparency, consistency, and information-sharing gaps is crucial for VioGén’s improvement and better support for victims of gender-based violence.

Open Source Intelligence (OSINT) for DV Cases

  • OSINT involves gathering information from publicly available sources like social media, news articles, websites, & public records.
  • In DV cases, OSINT can help uncover patterns, track behavior, & gather additional insights to support investigations.
  • It allows law enforcement to access critical information without the need for special permissions, but it’s important to use OSINT responsibly & follow privacy & ethical guidelines.

Mobile Apps for Geolocation & Telematic Reporting in DV Cases

Mobile applications empower victims to discreetly report DV & provide authorities with real-time information for rapid response. Features typically include:

Emergency Alerts

Victims can send an SOS with geolocation to law enforcement.

Quick Reporting

Apps simplify reporting past or ongoing violence.

Resources

Many apps offer educational materials or connections to victim support services.

Benefits for Police:

  • Faster emergency response.
  • Improved situational awareness through geolocation.
  • Enhanced victim safety with discreet reporting methods.

Mobile Apps for Geolocation & Telematic Reporting in DV Cases

Examples Across Europe

Spain – AlertCops: Sends geolocated alerts directly to police for immediate assistance.

Serbia – SOS App: Multilingual reporting for emergencies, adapted for visually impaired users.

Montenegro – Be Safe App: Discreet alerts for victim safety.

Greece – Panic Button App: Emergency alert with geolocation sent to police.

Armenia – Safe YOU App: Sends geolocated alerts & connects victims to resources.

Electronic Bracelets: Enhancing DV Response

BENEFITS

  1. Increased Victim Safety: Alerts victims & police if restrictions are violated.
  2. Deterrence: Reduces risk of reoffending by holding perpetrators accountable.
  3. Efficient Monitoring: Saves resources with remote tracking.
  4. Real-Time Response: Immediate alerts enable swift police action.

PURPOSE

  • Tracks perpetrators to enforce restraining orders & protect victims.
  • Uses GPS or radio-frequency technology for real-time monitoring.

EXAMPLES

  • Spain: Tracks restraining order compliance.
  • France & Sweden: Used in high-risk cases to prevent violence.

CHALLENGES

  • False alerts or device tampering.
  • Requires reliable GPS infrastructure.
  • Privacy & legal considerations.

Key Messages

  • IT-based solutions help reduce subjectivity & improve accuracy.
  • By using data-driven risk assessment tools, we can better prevent recidivism & tailor victim protection to the evolving situation.
  • Data analysis & IT solutions enable faster information sharing & support decision-making.
  • Unified data systems & consistent application of protocols provide a solid foundation for collaboration & coordinated responses between different agencies.
  • Data usage must be secure, transparent, & purposeful to protect the privacy of victims & witnesses.
  • Adhering to policies & regulations ensures that data processing is conducted ethically & lawfully.
  • Technological solutions & risk assessment tools are constantly evolving; hence, it’s essential for police personnel to receive ongoing training & support in implementing new solutions.
  • Opportunities to enhance skills in data analysis & the use of digital tools improve the efficiency of the entire organization & enhance victim protection.

Self Assessment