A reliable risk assessment is needed during the process of issuing the protection order. The purpose of this study was to develop a lethal risk assessment tool for the relevant agencies while dealing with the marital violence. The marital violence cases received by Kaohsiung City and Kaohsiung County were used as samples, whereas the Danger Assessment (DA), Brief Conflict Tactics Scale, (Brief CTS), and a questionnaire were used as the assessment tools in this study. It was found that the correlation coefficients of DA and Brief CTS on lethal violence were. 364(p<.000) and .333(p<.000), respectively, and both of which reached moderate predictive accuracy. While DA was used to predict lethal violence through ROC (Receiver Operative Characteristic) curve, the area under the ROC curve was .718, which reached moderate accuracy, whereas, while using Brief CTS, the area under the ROC curve was .753, which reached lower-moderate but still satisfactory accuracy. To learn the variance of lethal violence explained by DA, it was found that the R-square was .3983. It was suggested that 50% and 75% (5 and 7 in DA scores) could be used as the cutoffs for dividing marital violence groups into three subgroups-low lethal risk abusers (5 and below in DA), moderate lethal risk abusers (6 and 7 in DA), and high lethal risk abusers (8 and upper in DA). How to modify the DA to improve its accuracy in Taiwan was also discussed in this study.
A reliable risk assessment is needed during the process of issuing the protection order. The purpose of this study was to develop a lethal risk assessment tool for the relevant agencies while dealing with the marital violence. The marital violence cases received by Kaohsiung City and Kaohsiung County were used as samples, whereas the Danger Assessment (DA), Brief Conflict Tactics Scale, (Brief CTS), and a questionnaire were used as the assessment tools in this study. It was found that the correlation coefficients of DA and Brief CTS on lethal violence were. 364(p<.000) and .333(p<.000), respectively, and both of which reached moderate predictive accuracy. While DA was used to predict lethal violence through ROC (Receiver Operative Characteristic) curve, the area under the ROC curve was .718, which reached moderate accuracy, whereas, while using Brief CTS, the area under the ROC curve was .753, which reached lower-moderate but still satisfactory accuracy. To learn the variance of lethal violence explained by DA, it was found that the R-square was .3983. It was suggested that 50% and 75% (5 and 7 in DA scores) could be used as the cutoffs for dividing marital violence groups into three subgroups-low lethal risk abusers (5 and below in DA), moderate lethal risk abusers (6 and 7 in DA), and high lethal risk abusers (8 and upper in DA). How to modify the DA to improve its accuracy in Taiwan was also discussed in this study.