Police Attempts to Predict Domestic Murder and Serious Assaults: Is Early Warning Possible Yet?

Abstract

What facts known to police, if any, could have predicted the 118 domestic and family murders and near-murders in the Thames Valley (UK) police area over a recent three-year period? Thames Valley Police records were examined for 118 victims and 120 offenders in all 118 cases of “deadly” domestic violence: murder, attempted murder, manslaughter, and grievous bodily harm with intent in calendar years 2007, 2008 and 2009. Police National Computer (PNC) and other records of police contact with those persons prior to the cases identified for this prediction study were also coded, as well as the PNC records of a case-control sample of 150 male and 100 female offenders in less serious violent crimes. Local police risk assessments, including the DASH protocol, were coded for deadly violence cases with prior police contact. Risk assessments made during prior police contacts with domestic violence cases during the time period were coded for accuracy of forecasts. Characteristics of offenders and victims in deadly violence were also compared to characteristics of the case-control sample, calculating the relative-risk ratios between the two groups for the presence of factors that further research could explore as possible predictors of deadly domestic violence. In 55% of deadly violence cases, there was no prior recorded contact by police with the victim; a further 21% of victims had only one prior police contact. Not one of the 13 murder cases with any prior police contact had ever been assessed by responding officers as “high risk” based on DASH or its predecessor, for a 100% false negative rate of prediction on risk of murder and an 89% false negative rate for any deadly violence (only 6 cases with prior contact had been classified as high risk). In contrast, during the same time period, 2721 other domestic violence cases yielded “high-risk” assessments but were not actually followed by deadly violence. Yet the case control study found male offenders who committed serious domestic assaults were over three times more likely than other violent offenders to be have had markers of suicidal tendencies. Predicting deadly domestic violence based on intelligence from prior police contacts does not appear possible at present, given these findings that less than half of these cases had prior police contact, and that when contact did occur, the relationships were assessed by the DASH protocol or its predecessor as not high risk in 89% of those cases. A more promising predictor for male offenders in these cases, even without prior contact, may be a prior marker for suicide (ideation, threats or attempts), but use of that marker requires further research on the extent to which it can be identified before the deadly violence occurs, either from police contacts or other sources, so that more intensive efforts can be undertaken to prevent deadly violence in the truly high-risk cases.