Does prior history of domestic violence predict domestic

Does prior history of domestic violence predict domestic murder or other serious assaults? Sara Thornton, Thames Valley Police

The claim • There is a clear relationship between everyday…interpersonal violence…and…domestic homicide (Websdale 1994) • There is therefore an opportunity to intervene in escalating patterns of violence • Therefore we need to assess the risk based on predictors drawn from studies of domestic murders • Prediction can lead to prevention Conference’s name here 00. 00

The research questions • In how many cases was there a prior history of reported domestic violence? • How accurate were the risk assessments based on prior history in terms of false positives and false negatives? • Might a prospective denominator-based prediction produce more accurate forecasting than a numerator-based prediction? • Are the predictors for murder different from other serious assaults? Conference’s name here 00. 00

DASH high risk factors for potential murder victims • • • Fear Isolation Separation Conflict over a child Stalking Pregnancy Child abuse Frequency of abuse Escalation of abuse Conference’s name here • • • Jealousy Weapons Threats to kill Strangulation Sexual assault Other threats Mistreat animals Drug abuse Attempt suicide 00. 00

Domestic Violence Offences Offence Murder Attempt murder Manslaughter GBH with intent Total Conference’s name here 3 Years - Numbers 2007 to 2009 % of sample 13 11 16 14 1 1 88 74 118 100 00. 00

Gender and outcomes Murder Attempt GBH with All murder Manslaughter intent offences Female victims 11 15 (85%) (94%) 40 1 (45%) 67 Male victims 2 (15%) 48 0 (55%) 51 Total Conference’s name here 13 1 (6%) 16 1 88 118 00. 00

Risk of death per attack: 4 times higher for women victims Female Victims Male Victims Attacks = 67 Attacks = 51 Death = 11 Deaths = 2 Ratio = 1 in 6 Ratio = 1 in 25 Conference’s name here 00. 00

Victim prior contact with the police Victim has prior recorded domestic incident contact with police since 1/1/2000 Number % YES 53 44. 9 NO 65 55. 1 118 100 Total Conference’s name here 00. 00

Prior risk assessment Risk assessment for victim on most recent prior domestic incident (n=53) High Number % Valid % 5 4. 2 9. 4 Medium 21 17. 8 39. 6 Standard 21 17. 8 39. 6 Prior contact but risk assessment not known or not recorded 6 5. 1 11. 3 N/A as no prior domestic incident contact with police 65 55. 1 - 118 100 Total Conference’s name here 00. 00

How well did DASH predict? • None of the 13 murder victims had been assessed as high risk • Of the 6 murder victims who had been risk assessed following a previous incident, 1 was assessed as medium risk and 5 standard risk • No prior domestic incident contact with police for 7 of the murder victims • DASH missed 55% of attacks Conference’s name here 00. 00

Prior domestic incident contacts between victim & police Total number of contacts Number of victims Cumulative % 20+ 3 2 10 -19 5 7 2 -9 20 24 1 25 45 Sub total contacts 53 45 No prior Domestic Incident contact with police 65 55 Total 118 100 Conference’s name here 00. 00

Most recent prior victim contact with the police for domestic incidents (n=53) Nature of victim's most recent prior domestic incident contact with the police Number % Non-crime domestic incident 16 30. 2 Actual Bodily Harm 16 30. 2 Assault without injury 7 13. 2 Criminal Damage 4 7. 5 Threats to kill 4 7. 5 Harassment 3 5. 7 Racist incident non-recordable crime 1 1. 9 GBH with intent 1 1. 9 Arson with intent to endanger life 1 1. 9 53 100 Total Conference’s name here 00. 00

Offender criminal record Previous criminal record Number % of sample Suspect has prior criminal record 87 73. 7 [Suspect has prior criminal record for a violent offence] 70 59. 3 Suspect has no prior criminal record 27 22. 8 No record traced/missing data 4 3. 4 118 100 Total Conference’s name here 00. 00

Next steps • Collect data on the false positives: analyse MARAC watch list • Investigate potential denominators • Identify further studies which have assessed the effectiveness of DASH or its predecessor Conference’s name here 00. 00

Thank you