4
views
0
recommends
+1 Recommend
1 collections
    0
    shares
      • Record: found
      • Abstract: found
      • Article: found
      Is Open Access

      Predicting Domestic Abuse (Fairly) and Police Risk Assessment Translated title: La predicción (equitativa) de la violencia doméstica y la evaluación policial de riesgo

      research-article

      Read this article at

      ScienceOpenPublisherPMC
      Bookmark
          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

          ABSTRACT

          Domestic abuse victim risk assessment is crucial for providing victims with the correct level of support. However, it has been shown that the approach currently taken by most UK police forces, the Domestic Abuse, Stalking, and Honour Based Violence (DASH) risk assessment, is not identifying the most vulnerable victims. Instead, we tested several machine learning algorithms and propose a predictive model, using logistic regression with elastic net as the best performing, that incorporates information readily available in police databases, and census-area-level statistics. We used data from a large UK police force including 350,000 domestic abuse incidents. Our models made significant improvement upon the predictive capacity of DASH, both for intimate partner violence (IPV; AUC = .748) and other forms of domestic abuse (non-IPV; AUC = .763). The most influential variables in the model were of the categories criminal history and domestic abuse history, particularly time since the last incident. We show that the DASH questions contributed almost nothing to the predictive performance. We also provide an overview of model fairness performance for ethnic and socioeconomic subgroups of the data sample. Although there were disparities between ethnic and demographic subgroups, everyone benefited from the increased accuracy of model-based predictions when compared with officer risk predictions.

          RESUMEN

          La evaluación de riesgo de las víctimas de abuso doméstico es crucial para poder ofrecerle a las mismas el nivel adecuado de asistencia. No obstante, se ha demostrado que el enfoque predominante en casi todas las fuerzas policiales británicas, que descansa en el uso de DASH (las iniciales en inglés del instrumento de evaluación de abuso doméstico, acoso y violencia por cuestión de honor), no sirve para identificar a las víctimas más vulnerables. En su lugar, este artículo evalúa varios algoritmos de aprendizaje automático y propone un modelo predictivo, usando como algoritmo con un mejor rendimiento una regresión logística con red elástica, que utiliza como fuente de información variables normalmente disponibles en los archivos policiales, así como en el censo de la población. Para desarrollar y evaluar este modelo usamos datos de un departamento policial responsable de un área metropolitana en el Reino Unido que incluía 350,000 incidentes de abuso doméstico. Nuestros modelos mejoran significativamente la capacidad predictiva de DASH, tanto para la violencia en la relación de pareja (AUC = .748) como para otras formas de abuso doméstico (AUC = .763). Las variables más influyentes en el modelo fueron medidas del historial delictivo y de violencia doméstica previa, en particular el tiempo transcurrido desde el último incidente. El artículo demuestra que el cuestionario DASH prácticamente no contribuye nada al rendimiento predictivo de nuestro modelo. El artículo también ofrece una evaluación del rendimiento en términos de equidad para distintos grupos étnicos y socioeconómicos en nuestra muestra. Aunque había disparidad entre estos subgrupos, todos ellos se beneficiaban de la mayor precisión predictiva resultante de usar nuestros modelos en lugar de las clasificaciones policiales basadas en DASH.

          Related collections

          Most cited references68

          • Record: found
          • Abstract: not found
          • Article: not found

          Regularization and variable selection via the elastic net

            Bookmark
            • Record: found
            • Abstract: found
            • Article: not found

            Scaling regression inputs by dividing by two standard deviations.

            Interpretation of regression coefficients is sensitive to the scale of the inputs. One method often used to place input variables on a common scale is to divide each numeric variable by its standard deviation. Here we propose dividing each numeric variable by two times its standard deviation, so that the generic comparison is with inputs equal to the mean +/-1 standard deviation. The resulting coefficients are then directly comparable for untransformed binary predictors. We have implemented the procedure as a function in R. We illustrate the method with two simple analyses that are typical of applied modeling: a linear regression of data from the National Election Study and a multilevel logistic regression of data on the prevalence of rodents in New York City apartments. We recommend our rescaling as a default option--an improvement upon the usual approach of including variables in whatever way they are coded in the data file--so that the magnitudes of coefficients can be directly compared as a matter of routine statistical practice. (c) 2007 John Wiley & Sons, Ltd.
              Bookmark
              • Record: found
              • Abstract: not found
              • Article: not found

              Fair Prediction with Disparate Impact: A Study of Bias in Recidivism Prediction Instruments

                Bookmark

                Author and article information

                Journal
                Interv Psicosoc
                Interv Psicosoc
                inter
                Psychosocial Intervention
                Colegio Oficial de la Psicología de Madrid
                1132-0559
                2173-4712
                20 July 2022
                July 2022
                : 31
                : 3
                : 145-157
                Affiliations
                [a ] orgnameUniversity of Manchester UK originalUniversity of Manchester, UK
                [b ] orgnameUniversity of Seville Seville Spain originalUniversity of Seville, Spain
                Author notes
                Correspondence: jmedina11@ 123456us.es (J. Medina-Ariza).

                Conflict of Interest: The authors of this article declare no conflict of interest.

                Article
                00002
                10.5093/pi2022a11
                10268549
                37361012
                9b492543-bc23-43de-bb40-3c71e5976fac
                Copyright © 2022, Colegio Oficial de la Psicología de Madrid

                This is an Open Access article distributed under the terms of the Creative Commons Attribution-Noncommercial No Derivative License, which permits unrestricted non-commercial use, distribution, and reproduction in any medium provided the original work is properly cited and the work is not changed in any way.

                History
                : 28 December 2021
                : 07 May 2022
                : 17 June 2022
                Page count
                Figures: 4, Tables: 5, Equations: 0, References: 57
                Funding
                Funded by: ESRC
                Award ID: ES/M01178X/1
                Categories
                Research-Article

                domestic abuse,risk assessment,machine learning,algorithmic fairness,police,violencia doméstica,evaluación de riesgos,aprendizaje automático,justicia algorítmica,policía

                Comments

                Comment on this article