Juliane Gerth

Assessing the discrimination and calibration of the Ontario Domestic Assault Risk Assessment in Switzerland.

Intimate partner violence (IPV) is a major public health issue; worldwide, almost 1 in 3 women is affected. Police involvement in IPV cases has substantially increased because of “proarrest” and “procharging” policies and the enforcement of laws protecting victims of domestic violence. In the course of these changes, several front-line instruments have been developed to structure police risk assessment and decision-making strategies in such cases. One of those is the Ontario Domestic Assault Risk Assessment (ODARA). To investigate its validity in a Swiss police setting, a total cohort of male IPV offenders was retrospectively assessed for a fixed time at risk of 5 years. The recidivism base rate was 32% when recidivism was defined as subsequent police-registered IPV. Although ODARA scores were significantly correlated with IPV recidivism, they showed poor discrimination and calibration. Despite comparable base rates of recidivism, the Zurich sample scored significantly higher on the ODARA than the development sample. This mismatch of the expected and observed recidivism rates resulted in an overestimating of risk, especially in the two highest risk bins. Several reasons for those deviations, such as level of intervention, victim’s reporting behavior, and the dynamic nature of IPV, are discussed.

Gewalt im Strafvollzug: Prävalenz, Erscheinungsformen und Risikofaktoren

When comparing different studies on prison violence, it must be taken into account that correctional facilities vary considerably among one another. Studies conducted in Switzerland have shown that every one inmate in four is violent at least once during incarceration. These violent acts differ in regard to manifestation and a specific characterization of violence according to type, target, implication and trigger appears to be necessary. Existing criminal prognostic instruments are not suitable for estimating the risk of violent behavior in prisons. Therefore, the development of specific models is essential. International studies have identified various risk factors. However, due to the aforementioned differences in facilities, these predictors can only partly be transferred to a Swiss setting.Beim Vergleich verschiedener Studien uber Gewalt im Strafvollzug ist zu berucksichtigen, dass sich Vollzugsinstitutionen stark voneinander unterscheiden. Studien aus der Schweiz ergaben, dass etwa jeder vierte Insasse mindestens einmal wahrend der Haft mit Gewalt auffallt. Dabei haben die Gewalthandlungen unterschiedliche Erscheinungsformen. Die spezifische Charakterisierung der Gewalt nach Art, Ziel, Auswirkung und Ausloser ist darum wichtig. Bestehende Instrumente zur Kriminalprognose eignen sich nicht zur Schatzung des Risikos gewalttatiger Handlungen im Strafvollzug, weshalb die Entwicklung spezifischer Modelle fur diese Fragestellung notwendig ist. In internationalen Studien wurden diverse Risikomerkmale identifiziert. Auf Schweizer Verhaltnisse lassen sich diese jedoch aufgrund der erwahnten Institutionsunterschiede nur teilweise ubertragen.