Abstract
This article addresses the issue of police officers’ use of discretion when responding to domestic violence. With reference to Ericson and Haggerty’s theory of risk-oriented policing, we collected data direct from information management systems in an English police force and conducted field observations with attending officers to explore the degree to which officers used discretion to interpret the national definition of domestic violence. We also considered how officers applied national standards for recording incidents and crimes. We found that considerable discretion was required to interpret the official definition of domestic violence, and that decision making in relation to recording or otherwise incidents and crimes of domestic violence was variable. Specifically, we found examples of domestic-related incidents not recorded as such, and examples of crimes either not or incorrectly recorded. The implications of these findings for policy and practice are discussed.
Introduction
The issue of officer discretion has been a recurring theme in academic research on the police since the first ethnographic studies of police work in the 1960s. Goldstein’s classic (1963) account of police discretion contrasted the ‘ideal’ of full enforcement of the law with the ‘reality’ that both police managers and frontline officers are required to use considerable discretion in discharging their roles. Klockars (1985) illustrated how laws are sufficiently broad to require interpretation by practitioners, and frequently overreaching, such that they must be applied with common sense in order that the legitimacy of the police role is preserved. The need for discretion is, then, in part a consequence of the frontline officer being in reality ‘primarily a “peace officer” rather than a “law officer”’ (Banton, 1964: 127).
Use of discretion can sometimes, of course, have a negative impact on police legitimacy, if the law is under or selectively enforced, or if procedures are not followed correctly. Indeed, contemporary discussions of police culture have tended to focus on the potential negative consequences of officer discretion; the concept of ‘cuffing’, whereby officers screen out crimes from official records or figures, ‘concealing the truth up [their] sleeves or cuffs’ (Young, 1991: 324), is a prime example. These largely negative connotations stand in contrast to how discretion is viewed in fields such as medicine, where use of professional judgement by practitioners is rarely questioned (Klockars, 1985: 94).
Nowhere has the question of police discretion been more hotly contested than in relation to domestic violence. Authors who studied police responses to domestic violence from the 1970s onwards (see, for example, Edwards, 1989; Kelly, 1999; Stanko, 1985) documented the problems posed by the interaction of police discretion and the uninformed, sexist attitudes of officers. Classic ethnographic studies of police work also revealed that many officers regarded domestic violence as a civil matter and not something that the police should be involved with, or at best as ‘rubbish’ work to be afforded a low priority alongside ‘real’, crime-focused activity (see, for example, Holdaway, 1983).
Changes to the administration of police work introduced since these studies have, however, had the potential to curtail the discretion of both police agencies and individual officers. Drawing on the work of Ulrich Beck (1992), Ericson and Haggerty (1997) applied the concept of the ‘risk society’ to policing in western democracies, suggesting demands, often from outside agencies, for knowledge around risk, and the development of systems to quantify, classify and communicate risk, had become the key principles around which police work was defined and organized. According to this thesis, police work has undergone a process of ‘Fordization’, making it similar in character to Weber’s conception of ‘bureaucratic “dehumanized” labour’ (Ericson and Haggerty, 1997: 37).
Alongside a greater focus on risk, public services including the police have been subject in the last 25 years to ‘new managerialist’ policies, characterized by target-driven performance cultures. In recent years, some commentators and police practitioners have suggested that a culture of ‘risk aversion’ has developed in UK policing (see Cockcroft, 2012; Heaton, 2011), which a major review of policing commissioned by the then Labour government suggested can ‘seriously dilute, or at worst remove, discretion or professional judgement’ (Flanagan, 2007: 8). This culture, it has been suggested, has developed in part as a result of ever more extensive scrutiny and inspection regimes by bodies such as Her Majesty’s Inspectorate of Constabulary (HMIC) and the Independent Police Complaints Commission (Heaton, 2011).
There may be some merit in the suggestion that risk and managerialist cultures have shaped the way police respond to domestic violence. Police forces in England and Wales work to an agreed definition of ‘domestic violence and abuse’, initiated by the Home Office: ‘Any incident or pattern of incidents of controlling, coercive, threatening behaviour, violence or abuse between those aged 16 or over who are, or have been, intimate partners or family members regardless of gender or sexuality’ (https://www.gov.uk/domestic-violence-and-abuse). Computerized systems of call logging, dispatch and incident recording have replaced more easily circumvented paper records. Perhaps most significantly, in England and Wales police forces have adopted ‘prima facie’ national standards for recording crimes, limiting potentially the scope for officers to cuff criminal offences committed in intimate relationships, and standards for recording incidents, restricting scope for ‘squaring away’ (taking no action in response to) behaviour that is abusive in nature but does not constitute a criminal offence. Yet recent ethnographic study of police work in England and Wales discovered themes relating to aspects of police occupational culture that were remarkably consistent with those that emerged from classic studies from the 1960s onwards. As well as the persistence of a masculine, crime fighting ethos (see Westmarland, 2001), Loftus (2009) witnessed examples of frontline officers minimizing the importance of reports of domestic violence and taking little or no action in response.
Domestic violence is currently a high profile policy issue in England and Wales, exemplified by the Home Office’s announcement of legislation to criminalize ‘coercive control’, an insidious form of domestic violence that involves multiple controlling and abusive behaviours and has the greatest impact on victim well-being (see Myhill, 2015; Stark, 2007). Such a law could pose a challenge to frontline officers in terms of recording accurately a ‘course of conduct’, much of which may not fall into traditional categories of crime. The present study used field observations and data collected direct from force systems to explore the degree of discretion that exists in relation to classifying incidents as ‘domestic-related’ and recording offences as crimes. Following a review of literature on aspects of police occupational culture that may shape officers’ use of discretion, and the literature on police response to domestic violence, we present the finding that in practice considerable discretion is required to interpret the official working definition of domestic violence, and that decision making in relation to recording or otherwise incidents and crimes of domestic violence was variable. The implications of these findings for policy and practice are discussed.
Police Occupational Culture and Use of Discretion
Classic accounts of police work suggested officers exercised considerable discretion in their day-to-day role, and that frontline officers, in particular, were proficient at taking new policies and systems and adapting them to reflect prevailing cultural norms and ways of working (for reviews see Cockcroft, 2012; McLaughlin, 2007; Reiner, 2010). Classic studies also suggested a core set of characteristics shared by frontline officers. Though there has been considerable debate in the interim around the extent to which these core elements of police culture influence officers’ work and their interactions with the public (see Cockcroft, 2012; Waddington, 2012), it is possible factors associated with occupational culture may influence frontline response to domestic violence. Specifically, such factors could be seen as favouring a limited response.
Skolnick’s (1966) description of the police officer’s ‘working personality’ emphasized constant reference to danger, authority and efficiency. Domestic incidents are, first, viewed by officers as inherently dangerous, in terms of the potential for both physical confrontation and ‘in the job trouble’ (Waddington, 1999: 302), should an apparently innocuous incident escalate to a homicide. Second, victims of domestic violence may, for a number of valid reasons, choose not to engage with the police, and officers who expect members of the public to defer to their authority may view them as ‘uncooperative’. Finally, linked to this issue of cooperation is the issue of clear-up rates. Rightly or wrongly, many officers perceive that victims of domestic violence are particularly likely to withdraw a complaint, resulting in an undetected crime.
Other core characteristics of police occupational culture might be seen to influence officers’ responses to domestic violence. An over-focus on crime fighting and law enforcement may afford domestic violence a low priority, as much of the intimidation and controlling behaviour that characterizes more serious forms of abuse has not historically been criminal in a legal sense (see Stark, 2007). The masculine ethos that remains in policing may cause officers to perceive ambiguous situations from a male perspective, and Westmarland (2001) described a tendency for male officers to attempt to exert control over their female colleagues. Suspicion and cynicism are also regarded as core elements of police occupational culture, and an officer’s sceptical mindset combined with a tendency to deal with the scene as presented to them does not favour victims who are being systematically controlled and manipulated. Additionally, officers’ tendencies to differentiate between the ‘respectable’ and ‘rough’ elements of society (Shearing, 1981, cited in Loftus, 2009: 14) may lead victims with lower socio-economic status to be seen as ‘deserving’ of the abuse they suffer (Stanko, 1985: 114).
It is clear then that aspects of police culture described in classic accounts of police work have the potential to result in a poor response to domestic violence, if officers are able to exercise discretion.
The ‘Scientization’ of Domestic Violence
Developments in policing over the past 30 years have, however, had the potential to restrict the exercise of discretion by frontline officers and, by extension, any potential malign influence of occupational culture. In a process they termed ‘scientization’, Ericson and Haggerty (1997) proposed that police organizations sought technological solutions to problems associated with identifying and managing risk. These ‘risk communication systems’, they argued, curtailed use of discretion by frontline officers, who were required to defer to prescribed processes. Several such processes could impact on police response to domestic violence.
The first studies of police response to domestic violence highlighted, among other things, the poor standard to which domestic incidents were recorded. Although studies found outright cuffing of crimes or incidents such that they were never formally recorded to be ‘exceedingly rare … but not unknown’ (Edwards, 1989: 109; see also Hoyle, 1998), the downgrading or trivialization of domestic-related incidents was commonplace. In particular, incidents that involved physical violence were classified frequently as ‘disputes’ as opposed to assaults (Edwards, 1989: 110).
In response to significant variations between forces in recording practices more generally, National Policing leads in England and Wales developed, with assistance from the Home Office, a protocol for recording incidents as crimes. The National Crime Recording Standard (NCRS), adopted by all police forces in 2002, was based on the principle that a victim report will be recorded as a crime, in line with the Home Office ‘counting rules’ for recorded crime, if, on the balance of probability, ‘(a) the circumstances as reported amount to a crime defined by law … and (b) there is no credible evidence to the contrary’ (Home Office, 2014). This standard is particularly relevant to domestic violence as, in the relatively common situation of a victim calling the police to report an assault by their partner and subsequently retracting the allegation, a crime should be recorded based on the initial report; the retraction, very possibly made under fear or duress, should not be regarded as ‘credible evidence’ that an offence had not occurred. 1 The NCRS was followed in 2006 by the National Standard for Incident Recording (NSIR). The NSIR provided a structure for classifying the wide range of non-crime incidents reported to the police, and contained a closing code for ‘domestic incident’ (Home Office, 2011). Finally, in 2009, National Policing leads endorsed a standardized risk model, which required attending officers to complete a risk identification form at every domestic incident they attended.
At face value, then, it would appear there are parallels between the current arrangements for policing domestic violence in England and Wales and Ericson and Haggerty’s description of risk-oriented policing. Police and partner agencies work to an agreed definition of domestic violence, officers’ recording of incidents and crimes is computerized and defined by national standards, and their response prescribed in part by a standardized national risk model. If Ericson and Haggerty’s thesis were correct, officers’ ability to use their discretion in defining and responding to domestic violence would be severely circumscribed.
The Present Study
This study concerns the scope for police officers to use discretion when classifying and recording incidents of domestic violence. We used data gathered from force systems to provide a snapshot of the recording and classification of domestic violence calls for service in a single week in a specific division of an English police force. Field observations were utilized to shed light on the processes that might influence whether specific cases are recorded as incidents or crimes. In particular, we considered how systems and/or the discretion of individual actors influenced the construction of occurrences as constituting domestic violence or otherwise, and as criminal offences or otherwise. This study addresses two key gaps in the literature. Barring work examining narratives used by officers in case construction (see Hester, 2012; Lea and Lynn, 2012), there has been little recent focus on how the police construct and administer domestic violence. There has also been a dearth of recent empirical work on police use of discretion, in the UK context at least.
Research Methods
The data presented herein were collected in ‘Southshire police’, a medium-sized police force in the south of England that covers a mixture of rural and urban areas. Domestic violence accounts for 6 per cent of calls for assistance in Southshire, and 8 per cent of recorded crime, which is in line with national averages (see HMIC, 2014a). Risk assessment by attending officers is compulsory for all incidents of domestic abuse, and completed risk forms are reassessed by central unit. Cases classified as ‘medium’ or ‘high’ risk are referred to a specialist unit for intervention.
Data were collected as part of the evaluation of a pilot project commissioned by the National Policing Reducing Bureaucracy Programme Board and evaluated by the College of Policing. In recognition that the standardized risk model may not be appropriate for some incidents classified as domestic violence under the official definition, frontline officers in a specific division (‘K’) were, for a six-month period, permitted to use their professional judgement as to whether to submit a risk identification form. The pilot did not extend to crime or incident recording – officers were expected during the pilot period to write up incidents and record crimes according to existing practice. Fieldwork for the pilot did, however, provide the opportunity to examine officers’ use of discretion more generally.
Data collection took two forms: field observations conducted while accompanying first response officers, and the transcription of information direct from the force’s ‘computer-aided despatch’ (CAD) and information management system (IMS).
Data collected from force records
When a call comes in to the Southshire police force control room, the call-taker attempts to ascertain what has happened 2 and deploys a unit to the scene if required (it is force policy that all domestic incidents are deployed to). A CAD log is created containing the call-taker’s transcription of what is reported to them by the caller, as well as a summary of what the attending officer(s) report from the scene. There may be some negotiation between the call-taker and attending officer(s) if their report differs from the nature of the original call. The incident is given an NSIR code. Incidents are transferred (‘dropped in’) to the IMS if they are crimes, or require further investigation or action by the police. As well as writing a record of what happened for incidents that are recorded on the IMS, attending officers in Southshire police are responsible for ‘criming’ – applying where necessary a crime code to an incident, in accordance with the NCRS.
All CAD logs generated in K division for a seven-day period from 17 February to 23 February 2014 (N = 511) were screened to see whether they were domestic-related. This week was selected as it was the first week of the wider pilot project and close monitoring of officers’ actions was required. Any domestic-related logs that had not been dropped into IMS were interrogated to see whether they had been allocated a ‘domestic incident’ closing code.
Data were collected from IMS for the same period. All occurrences (N = 432) were interrogated by the authors to identify cases that either definitely or could possibly have involved domestic violence, under the official definition. Some incidents were more equivocal than others. In such cases, the IMS history of the individuals or addresses involved was examined, and the original CAD log re-examined for evidence suggesting the present occurrence was likely to have involved domestic violence. Such cases were appraised by both researchers and incidents deemed relevant were included in the final sample, coded as ‘possibly domestic-related’. The first 24 hours’ occurrences were screened by both researchers to ensure inter-rater reliability for the remainder of the data collection.
Data analysis
Data were interrogated for emergent themes using the principles of grounded theory, a method where analysis is undertaken inductively and iteratively to produce theoretical inferences rooted in the phenomena observed (see Charmaz, 2014; Glaser and Strauss, 1967). We were interested particularly in incidents that were possibly domestic-related, the extent to which frontline officers applied discretion in determining whether these cases were inherently abusive, and whether they were recorded as such. Several coherent categories relating to use of discretion emerged; these are presented and discussed below.
In accordance with the ‘non-linear’ nature of grounded inquiry (see Charmaz, 2014), cases identified as domestic or possibly domestic-related in the initial data collection period in February were re-examined periodically over subsequent weeks. The purpose was twofold. First, this process of revisiting the data allowed emerging categories to be developed and refined. Second, the data evolved over this period; if further investigation occurred, or cases were reviewed by supervisors or specialist teams, the force’s own classifications of occurrences changed. A final review of occurrences was conducted in July 2014. 3
Data collected from field observations
As categories relating to officers’ use of discretion emerged from the data collected from force systems, field observations provided the opportunity to see whether the same phenomena could be recognized in practice. Fourteen shifts were observed between 3 March and 23 July 2014. The observed shifts were primarily late shifts, as these tend to have the largest volume of domestic incidents. The shifts were spread evenly across the five teams of response officers in the division, to ensure ‘range and diversity’ in respect of practice observed (McNaughton Nicholls et al., 2014: 254). The researchers took the role of ‘observer as participant’ (McNaughton Nicholls et al., 2014: 247), accompanying officers to all incidents – domestic-related or otherwise – to which they were deployed, observing and recording notes as unobtrusively as possible.
The data collected were a mixture of descriptive and focused observation. In the time between deployments, officers were probed on their general views towards domestic violence and responding to domestic incidents. Fieldnotes were recorded during the shifts. Detailed narrative accounts, comprising description, observer comments and subjective reflections (McNaughton Nicholls et al., 2014: 260), were written the following day.
The Influence of Officer Discretion on the Construction of Domestic Violence
There are several possible outcomes for calls to Southshire police which might comprise domestic violence, under the official definition. Such incidents may or may not be classified as a ‘domestic incident’, and may or may not be dropped into IMS. If they are dropped into IMS, they may or may not be classified as a crime. When interrogating the data for examples of discretionary decision making, four main categories emerged. The largest of these comprised cases where officers had used legitimately their discretion to interpret the official definition of domestic violence. The other themes related to traditional conceptions of officers’ use of discretion to in effect ‘square away’ or downgrade particular incidents.
Negotiation of definition and context
When responding to reports of potential domestic violence, call-takers and attending officers were required to interpret and sometimes negotiate the meaning and applicability of the official definition of domestic violence. Put simply: did the people involved fit the definition in terms of their relationship to each other, and was the behaviour reported inherently abusive? There were several cases where the dynamics turned out not to be abusive, such as teenagers play fighting, or driving their parents’ cars without being qualified to do so; low-level civil disputes related to property; and false or malicious allegations made by people with mental health issues. There were also cases involving for example disputes between ‘in-laws’ in very recent relationships, or nuisance communications between individuals’ current and former partners.
As the following example illustrates, some of these incidents may or may not have been abusive in nature: A member of the public called the police to report a man and woman ‘having a domestic’ in the street, with ‘lots of pushing and shoving’. CCTV showed a woman on the ground and then being held up against a wall by the male. Attending officers found ‘one very drunk female and husband trying to get her home’. They reported: ‘This is not a domestic. Female is very much in drink, and keeps falling over. Checked PNC [Police National Computer] and IMS for both male and female; no trace for either. Female unable to stand due to her intoxication and her husband was struggling to hold her up. On leaving the area, the female vomited and fell to the ground again. The husband was loving and supportive and showed a great deal of understanding to her plight’.
Fieldnotes from observations with frontline officers described the following incident. It is illustrative of a relatively common situation that reflects a ‘family dispute’ as opposed to a domestic violence dynamic as traditionally conceived. Additionally, as the individuals delivering/receiving threats were neither related directly or in an intimate relationship, the officers decided the incident was not covered by the official definition of domestic violence: A man called the police to report that his eldest son, while on the phone to him, had threatened to kill his other adult son’s girlfriend. When we arrived it became evident the family were in dispute over a number of different things and that the father had phoned the police more out of irritation that he had been pulled into the argument than because he felt the threats to be real. No one at the address was distressed, and the woman subject to the threats appeared unfazed. She confirmed she and her boyfriend’s brother do not like each other because they ‘rub each other up the wrong way’, but stated confidently she did not take his threat seriously.
That the negotiation and definition of context category was most prevalent is significant, as it suggests that officers are required to use discretion when responding to cases of domestic violence.
Incident cuffs
The cuffing of domestic-related incidents was observed as having possibly occurred at both the call-taking and first response stages. Cuffing is seen traditionally as downgrading the seriousness of a criminal offence, or concluding that the offence did not take place or is not worthy of police intervention. Although regarded primarily as a means for keeping crime rates artificially low, there is also an incentive for officers to cuff incidents that do not involve criminal offences. In Southshire, incidents for which there is no imperative to make further enquiries can be ‘filed at first submission’; the attending officer(s) are not required to write an enquiry log or, in the case of domestic violence, submit a risk identification form. The incentive for a call-taker to cuff a domestic-related incident is less clear, but could stem either from a lack of understanding of the dynamics of domestic violence, and/or an organizational imperative not to carry too many ‘open’ CAD logs.
Five CAD logs were discovered which, in the opinion of the researchers, may have been domestic violence and should have been dropped into IMS and investigated further, but were not. Two of these involved dropped 999 calls from addresses with warning markers for domestic violence and where the explanation on call-back was that children had made the call by accident. Another incident involved a report by a neighbour of a disturbance and a male threatening physical harm to a female. Attending officers reported that the male had been shouting at the dog. By coincidence, the lead author observed subsequently the female relating to this incident reporting for the first time the significant abuse she had suffered, and stating that, on one occasion, the police had attended. The remaining two incidents were treated as disputes over property. In one of these, there was a significant history of coercive and controlling abuse.
Certain characteristics of cases where no action was taken – such as a significant history of domestic violence, and/or disparities between the CAD log and the attending officer(s)’ account of the incident – were taken as indicative of an incident having possibly been cuffed. Among 30 domestic-related or possibly domestic-related incidents that were dropped into IMS, the authors felt that seven were either definitely or potentially cuffed. The following case shows how officers exercise discretion in relation to the amount of effort invested in following up an initial report: A neighbour reported a disturbance outside a specific flat; he had made previous reports of domestic abuse involving a couple and been told to call again if there were further incidents. A search of IMS for the address showed a female at high risk of domestic violence from a male. The attending officers reported that the flat was in darkness. No attempts were made seemingly to locate the occupants, and the incident was closed as ‘antisocial behaviour’.
During the shift observations, researchers observed directly one incident that might be described as a cuff: Officers were deployed to a report of a potential criminal damage. A man had come to collect some clothes from his ex-partner’s house while she was out; she had left them in a wheelie bin. This action had angered the male and he had emptied the bins and thrown them across the front lawn. One attending officer spoke with the woman by telephone, and it appeared she was asking about restraining orders. The officer advised her that was a civil matter and she should contact the Citizens Advice Bureau. When conferring, the officers clearly did not regard the incident as of any significance. One commented that he too would have been upset if his clothes had been left in a bin. No attempt was made to locate the woman and complete a risk identification form. This incident was not dropped into IMS. A subsequent search for the CAD log showed that the incident had been given a ‘for information’ NSIR closing code.
A review of IMS showed the woman in this case reported subsequently a history of very serious abuse leading up to this incident. It should be noted though that the observer did not regard the cuffing of this specific incident as malicious; rather, it appeared the attending officers simply did not regard it as serious, or appreciate the potential for ‘low-level’ incidents to form part of a wider pattern of harassment and abuse. It would appear though that there remains scope for domestic incidents to be cuffed ‘at source’ by some combination of the call-taker and attending officer(s).
Crime cuffs
The authors identified 10 incidents for which a crime might have been recorded according to the NCRS, but was not. Based on a re-examination of the original CAD logs and crime reports, the Southshire police crime registrar agreed that four of these cases should have been recorded as crimes (one assault occasioning actual bodily harm, two common assaults and a Section 2 harassment). The national crime registrar felt a further three crimes should have been recorded (two assaults and a harassment), illustrating that there is scope for differing interpretations of the NCRS and the counting rules (see also HMIC, 2014b).
These cases illustrated multiple ways in which discretion can be applied in the investigation and classification of incidents. One of the unrecorded assaults was observed by officers at the scene: A man and his adult daughter were arguing in the street. Officers on patrol observed the man push his daughter in the back of the head; he claimed this action was out of frustration. The officers stated the contact was very minor and that the woman said she did not feel as though she had been assaulted. The incident was classified as ‘other incident – non-crime’, but in accordance with the NCRS should have been recorded as a crime of common assault.
In three other cases, assaults were not recorded despite a clear report of assault being made in the initial call. In these cases, the information presented to negate the initial report amounted to no more than the victim retracting the allegation when questioned by the attending officer(s). This rationale is clearly inadequate in relation to cases of domestic violence, where victims are frequently under duress to retract their complaint.
Harassment appeared to be an area where practice was particularly variable. The lead author observed officers deal very thoroughly with an allegation of harassment which was the first report from a long-term victim of coercive control. Abusive voicemails were transcribed and Facebook pages photographed to make out a ‘course of conduct’. Officers appeared happy in other cases to record a first report as ‘non-crime first stage harassment’, again with little verifiable evidence to negate a victim’s initial description of a course of conduct.
A further two incidents involved reports by third parties. In these cases, the force took a literal interpretation of the NCRS, concluding that if the victim was not aware that the person making the report was acting in their interest and/or did not confirm the report, then a crime would not be recorded. In one of these cases, a woman said she heard banging in the next room and her friend emerged and told her that she had been assaulted by her partner; the friend told attending officers she had not been assaulted and that a cut in her mouth had been caused by braces on her teeth. A review of the history of this case showed the man involved was a prolific offender and there was a considerable history of violence against the female, suggesting that ‘on the balance of probabilities’ an assault had occurred. In all of these instances, discretion appeared to have been used to the end that crimes were not recorded when they might have been.
The field observations also revealed instances of crimes that might have been classified differently. In the report of harassment mentioned above, and despite the officers dealing sensitively with the victim and initiating appropriate safety planning and specialist intervention, a crime of ‘putting people in fear of violence’ was recorded under Section 4 of the Protection from Harassment Act when, during the course of her interview, the victim revealed she was regularly assaulted physically and in one case choked by the male until she almost blacked out. In accordance with the NCRS, as these offences had not been reported previously the more serious offence of assault should have been recorded.
Quasi-cuffs
The category ‘quasi-cuffs’ describes cases where officers dealt – often thoroughly and satisfactorily – with one component of an incident, but ignored or failed to identify and address a possible domestic abuse element. Five quasi-cuffs were identified during the week-long data collection period. The following case was representative of those where there appeared to be an overlap between domestic violence and abuse of children: A female called the police in great distress to report that she had discovered her partner had been having sexual intercourse with her 16-year-old daughter without her knowledge, and throughout their relationship. While officers were interviewing the daughter, she disclosed that the male had subjected her mother to extensive domestic violence. Despite this disclosure, the sole focus of the investigation appeared to remain on the child abuse element, with no evidence to suggest that a risk assessment was conducted for the mother, or appropriate support offered.
The following fieldnotes were taken on the final shift of observations, and show officers squaring away an incident where domestic violence may have been present: A man called at 1am to report a disturbance outside the house opposite; one man was heard to say to another man ‘I will bite your throat out.’ On attendance, we found a woman and her neighbours sitting outside. The woman’s son had come home drunk, burnt some toast and set off a smoke alarm. When his neighbour offered to help get his mother, who had health problems, out of the address, the man became aggressive. The officers established there had been no physical violence between the neighbours. The caller also stated that the man had been aggressive towards his mother, but there was little attempt made to ask the woman about abuse from her son. A subsequent search of IMS showed two previous incidents involving the mother and son; the son was a heroin user who had also abused his ex-partner. When I asked the officers whether the incident might have been domestic-related, they replied emphatically ‘no’.
Taken together, our findings show that police officers are required legitimately to use discretion in the investigation and classification of incidents that may constitute domestic violence. While sensible and positive use of discretion was observed in several cases, there were also numerous examples of domestic-related incidents not being investigated thoroughly, or being misclassified or downgraded.
Discussion
This study found that police officers used considerable discretion when responding to incidents of domestic violence. Consistent with HMIC (2014a, 2014b), our data showed, in a single week, a dozen or more examples of domestic-related incidents that were ‘squared away’; a lack of thorough investigation of some reports of abuse, both current and historic; and under-recording of an average of one domestic-related crime per day. These findings are consistent with those of previous studies of police response to domestic violence, and also with research on sexual assault, whereby officers have frequently been shown to be sceptical of victims’ accounts and to ‘no crime’ initial reports without adequate supporting evidence (see Hohl and Stanko, 2015). In line with previous research, the present study suggests it is still possible for officers to minimize the seriousness of domestic-related incidents, or even make them disappear altogether. And in that sense, frontline officers, in particular, influence to some degree what domestic violence is and, consequently, how the police and their partner agencies respond to it.
Ericson and Haggerty (1997) painted a picture of police work in western democracies governed primarily by restrictive rules and formats for communicating risk. This account questioned traditional sociological accounts, which emphasized the primacy of the frontline officer in determining both what gets done and how. There are, however, several reasons why discretion remains integral to police and particularly frontline officers’ handling of domestic violence. The official definition of domestic violence, for example, is extremely broad and perhaps overreaching (for a critique, see Kelly and Westmarland, 2014). Indeed, some officers in the study area appeared to differentiate between ‘genuine’ domestics and those that ‘hit the definition’, an example of the latter most often being ‘two brothers having a fight’. The Home Office acknowledges the need for forces to interpret this non-statutory definition, 4 and HMIC (2014a: 37) suggested, following an inspection of forces’ handling of domestic violence, that officers require ‘greater discretion’ to ensure that the police response is ‘targeted to address the particular risk that they find’. These official positions stand in contrast to notions of officers as process-driven automatons, and demonstrate the difficulties associated with squeezing discretion out of practice through rules and systems.
Additionally, although the NSIR and NCRS are prescriptive, they, as well as the official definition of domestic violence, are open to interpretation; and the greater the number of rules and processes, the less likely they are to dovetail. An example concerns ‘civil disputes’, which NSIR guidance (Home Office, 2011) suggests should be given a domestic closing code if they occur within a relationship. Research has shown consistently that civil issues such as child contact are used as a means of continuing to exert control by coercive ex-partners (Coy et al., 2012). Officers and staff should realize that in this instance the official definition of domestic violence overrides the NSIR guidance, 5 but our data suggested that in some instances call-takers and officers focused on the nature of the specific incident, and failed to situate it in a wider pattern of abuse.
Clearly, then, there is a requirement for police officers and staff from the control room, through the frontline and up the supervisory chain to use their discretion when interpreting and classifying cases of domestic violence. Our study suggests this discretion is not always used in an appropriate way. This situation is deeply problematic, and not primarily in relation to the ethical issue of the accurate recording of crime. The way in which domestic incidents and crimes are classified has a direct impact on the level of intervention that a victim can hope to receive: Southshire police uses the number and severity of previous occurrences as one criterion for assessing risk and referring cases for specialist intervention. Inconsistent and inadequate investigation and recording of incidents and crimes prevents robust documentation of the context and history of abuse. Opportunities for early intervention and ongoing management of risk will be missed.
Having established that discretion is not always used in a desirable way, the pertinent question becomes ‘why?’ We feel it is too simplistic to point solely to domestic violence being a crime committed primarily by men against women and investigated primarily by men. Indeed, many officers in K division displayed highly risk-averse attitudes to dealing with domestic abuse, stating that nobody wants to be seen to ‘drop the ball’ in such a high risk area. It is likely, then, that wider systemic and organizational factors are also at play (see also Hoyle, 1998).
One issue is the relative priority afforded to domestic violence at the organizational level. As Klockars (1985) notes, in addition to individuals, discretion is sanctioned and exercised by administrations through their policies and allocation of resources, which influence police behaviour and decision making. Loftus (2009: 129) suggested domestic violence was still ‘marginal to what many celebrated as meaningful police work’ in the force she observed, and HMIC (2014a) highlighted the low priority domestic violence continues to be afforded relative to ‘acquisitive’ crime. Domestic violence is an area where it can be difficult to get a ‘result’, as many victims, for a variety of reasons, do not support a prosecution. HMIC (2014b) did not rule out targets for reducing and detecting crime as one explanation for forces’ under-recording of crime.
The issue of resourcing should also be considered. On more than one occasion, the authors saw frontline officers writing up incidents long after their shift had finished. Now that officers are not paid routinely for overtime, there is little incentive for them to investigate and document thoroughly a case they, perhaps erroneously, regard as ‘low risk’. Officers have always been acutely aware that ‘making something into a crime requires work’ (Sumner, 1994, cited in Cockcroft, 2012: 19), and making something into a domestic incident also creates work, in the form of a 27 question risk identification form that most officers we spoke to felt was not appropriate for some incidents (see also Loftus, 2009).
Perhaps, though, the most important factor shaping the way officers exercise discretion in relation to domestic violence is their knowledge of the issue. It was evident that some lacked the nuanced understanding of the dynamics of coercive controlling abuse that would enable them to respond consistently and effectively. The term ‘cuff’ is generally used in a pejorative way, but our data suggested much of the poor practice relating to recording domestic violence was not malicious. In some instances, it appeared the officers involved simply regarded a particular incident as being of no risk. While in certain cases that might be true, in other cases such apparently minor incidents are symptomatic of a course of conduct, the impact of which is cumulative for the victim.
Some officers had an overt focus on criminal offences and physical violence, and although some recognized non-criminal and non-physically violent forms of abuse, it was not clear this knowledge stretched to how these behaviours might affect the behaviour of victims over the course of a relationship, and, consequently, how they ‘present’ at the scene. This finding is consistent with HMIC (2014a: 9), who concluded ‘officers are often ill-equipped to identify dangerous patterns of behaviour … in particular where there is no overt physical violence but instead there is psychological intimidation and control’. There was, both in officers’ written narratives and in what we observed on patrol, too frequently a tendency to respond to what they were presented with, as opposed to considering the wider context. Waddington (2012) highlights how this ‘incident-driven’ approach is partly a response to what is required by the wider criminal justice process.
What, then, is the solution to the problem of inaccurate or inappropriate use of discretion? Ericson and Haggerty suggest police agencies’ initial response to continued uncertainty is to introduce ever more complex and prescriptive systems and processes. But even in describing such systems, they hint at problems inherent in their thesis. Specifically, ‘the rush for technological solutions to problems of uncertainty often has the ironic consequence of increasing uncertainty’ (Ericson and Haggerty, 1997: 34). The definition of a problem might be too broad to permit focus on cases of greatest risk; systems for classifying information may require interpretation and be in conflict with one another. Ultimately, ‘the fact that communication formats are scientized to effect closure and certainty does not mean that they are always fully understood and accepted by the police officers who use them’ (Ericson and Haggerty, 1997: 385).
The latter point is crucial. According to Ericson and Haggerty (1997: 83), ‘discourse’ is ‘the institutional construction of knowledge which takes place within a social organisation of … people, rules, formats, and technologies’; it provides a ‘shared understanding’ and knowledge of an issue. The wider policy discourse around domestic violence is, however, complex, with even academic experts disagreeing on its prevalence and how best to respond. In part, these disagreements are bound up in debates about how to define ‘domestic violence’. We suggest that if officers had a better understanding of coercive control specifically, they would be better placed to exercise discretion in a positive way. While training clearly will not solve all the issues surrounding police response (Waddington, 2012), until officers are able to see beyond a ‘verbal only argument’ to the continuous and cumulative nature of coercive control they will fail to recognize the importance of identifying and recording accurately apparently low-level incidents of domestic violence.
As the sample size used for this research (N = 511) was limited to the workload of one specific division in one police force over the course of a single week, the results cannot be taken to be representative of practice across even the Southshire police force area. Rather, the data are used to exemplify the complexity involved in responding to domestic incidents and the requirement for officers to use discretion. The findings concerning the under-recording of crimes are however consistent with much broader examinations (see HMIC, 2014b). As with all research that uses qualitative and especially observational methods, there was a degree of subjectivity involved in data collection and analysis. The potential for personal attitudes and orientations to affect the interpretation of data was minimized by having only two researchers working closely on all aspects of the data collection and analysis, and reviewing each other’s interpretations and decisions.
Concluding Remarks
The issue of discretion remains pivotal to understanding the way police work is operationalized. Consistent with Goldstein’s classic discussion, police leaders and managers continue to perform a ‘double shuffle’, accepting universal and officially imposed processes while recognizing that those processes are unworkable without the application of discretion. Frontline police officers are and always have been the embodiment of this contradiction, their operational terrain continually contested. Discretion does not sit comfortably with ‘one-size-fits-all’ processes, and this notion may help explain why many frontline officers we observed expressed the view that their discretion had been removed, despite retaining the scope to make discretionary judgements.
In England and Wales, the advent of the College of Policing has given fresh impetus to the development of a professionalized model of policing. Under such a model, Klockars (1985: 119) suggests discretion should be recognized as an inherent and desirable element of police work whereby ‘we … trust police and train them to use their power wisely in the same way we … train and trust other professionals in whom we have no choice but to entrust equally impressive powers’. To that end, further in-depth study of police discretion is required to underpin a professionalized model, to tease out the likely complex set of individual, organizational and external factors that shape police decision making as ethical, unethical, informed or uninformed.
Research in the health sector has underpinned nuanced debate around the issue of discretion. Cheraghi-Sohi and Calnan (2013) showed that a centrally imposed care quality framework was welcomed by doctors as it was evidence-based and allowed them to retain ‘task discretion’. Many police practitioners are broadly sceptical of research evidence and guidance, at least initially, and the challenge for police policy-makers will be to develop standards that allow officers to feel trusted to make professional decisions based on robust evidence and practice. Ultimately, and in line with research on ‘organizational justice’ (see Bradford et al., 2014), well-trained, knowledgeable officers who are empowered to make professional judgements are likely to internalize the desire to respond appropriately more than officers beholden to processes they perceive in some instances as disproportionate and unfair. A truly professionalized model of policing would, then, have discretion as the ideal, as well as the reality.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.