ABSTRACT
This paper presents findings from a comparative study of police decision-making. Interview data are used to revisit Hoyle’s [1998. Negotiating domestic violence: police, criminal justice and victims. Oxford: Oxford University Press] explanatory model of police decision-making in response to domestic violence. The analysis suggested Hoyle’s model remains relevant: officers were more likely to arrest if there was evidence of a criminal offence they perceived as serious (most often physical violence), and if they perceived ongoing risk of harm to the victim. The data did suggest official policies around presumptive arrest are now more salient, but that officers continue to use traditional ‘craft’ work to circumvent such policies in cases where they perceive the risk of further harm to be minimal. Implications for practice and debates around the efficacy of arrest are discussed.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1. National Policing leads are senior officers who take responsibility for specific areas of policing in England and Wales on behalf of the National Police Chief’s Council.
2. Serious Crime Act 2015 (s76). Available from http://www.legislation.gov.uk/ukpga/2015/9/section/76/enacted
4. In two cases, the victim attended a police station as opposed to making contact by telephone.
5. This outcome is not unusual in relation to domestic violence as a call may be made in response to a perceived crisis only for the victim to minimise or deny what has happened to the attending officer (see, for example, Kelly Citation1999).
6. The Home Office ‘counting rules’ for recorded crime, contained within the NCRS, are fairly prescriptive, but there remains some room for interpretation. Where it is suggested that a crime might or should have been recorded, this is the author's interpretation of the counting rules, drawing on previous conversations with the National Crime Registrar.
7. The most comprehensive meta-analysis of data from the arrest experiments actually suggested a modest deterrent effect, when pooling data across the sites (see Maxwell et al. Citation2002).