1. Introduction

Criminal Justice System (CJS) responses to female offenders have been of particular concern in the UK since the Corston (2007) report highlighted the need for drastic changes. The report indicated that the CJS does not have the capacity to address women’s histories of victimisation, which often accompany their offending (Corston, 2007). Restorative Justice (RJ) has been described as a possible alternative disposal for female offenders because of its ‘holistic’ approach, taking into consideration the context of offenders’ lives while still being victim-led (Gaarder & Presser, 2006; Verrecchia, 2009). There have, however, been few studies examining female offenders’ experiences of RJ (Sherman, Strang & Newbury-Birch, 2008) barring some notable exceptions outside the UK where young male and female offenders’ experiences have been compared either qualitatively or quantitatively (Daly, 2008; Hayes, 2005; Maxwell, Kingi, Roberston, Morris, Cunningham, 2004; Rodriguez, 2007). Together, these studies demonstrate complex findings: RJ may reduce young women’s offending more than it does young men’s (Hayes, 2005; Rodriguez, 2007) but young women may report more negative experiences in RJ than young men do (Daly, 2008; Maxwell et al., 2004).
This article uses a mixed-methods approach to examine police uses of RJ, the differences in uses of police RJ for male and female offenders, and female offenders’ experiences of police RJ in one rural UK county from 2007 to 2012. This includes secondary analysis of administrative police data on 17,486 participants, including 2,586 female offenders, and narrative interviews with twelve female offenders, sampled from the database. The overall methodological approach is that of ‘new political arithmetic’, a two-step process involving analysis of big data followed by qualitative research in order to improve official work with vulnerable people (Smith, 2008: 335).
RJ, as a CJS intervention, has come far from the meeting between two young offenders and their victims in Ontario, Canada, in the 1970s (Zehr, 1990: 158). Forty years on, Aertsen, Parmentier, Vanfraechem, Walgrave and Zinsstag (2013: 2-3) have described it as,

a social movement with different degrees of self-criticism … [and] a field in its own right, looking for constructive ways of dealing with the aftermath of crime, while forming part of a wider socio-ethnic and political agenda.

This transformation is in part due to the open-ended nature as to what RJ should look like (Marshall, 1996). Advocates of RJ have, for example, frequently stated the field should be open to ‘innovative’ uses (Braithwaite, 2002) in order to allow practitioners to grow their practice in unique settings (Marshall, 1999). Despite this freedom, there are, however, core principles; according to the United Nations, the most crucial are ‘(a) an identifiable victim, (b) voluntary participation by the victim, (c) an offender who accepts responsibility for his/her criminal behaviour, and (d) non-coerced participation of the offender’ (Dandurand & Griffiths, 2006: 8).
While RJ practices have remained diverse, including circles, conferences and victim–offender mediation in community, school and CJS settings (see McCold, 2006; Shapland, Robinson & Sorsby, 2011), the focus of this study is police-led RJ. Police-led RJ originated in Australia and developed from family group conferencing models (McCold, 2006: 32-33). Since then, police-led RJ has been used in the US, UK, Australia, Canada and elsewhere (see, e.g. Chatterjee & Elliot, 2003; Hoyle, Young & Hill, 2002; McCold & Wachtel, 1998; O’Mahony & Doak, 2004; Sherman, Strang & Woods, 2000). Supporters propose that police facilitators can ensure the growth of RJ in the CJS through their practice (McCold & Stahr, 1996) and may improve the relationship between the police and the community long-term (Alarid & Montemayor, 2012).
Evaluations of police-led RJ, however, have been mixed. One of the first police-led RJ programmes evaluated in the US concluded that victims and offenders viewed police officers as capable of running conferences with a range of benefits (McCold & Wachtel, 1998). This was after two trainings by the researchers; however, as they initially noted, police were not ‘restorative’ enough (McCold & Wachtel, 1998). An evaluation of the Thames Valley experiment in the UK involving interviews with participants and observations of conferences by researchers supported the findings that police officers could be good facilitators, although the authors identified a number of problematic areas (Hoyle et al., 2002). They found, for example, that individual police officers’ skills as facilitators differed and that police may have been more punitive towards young offenders than adult offenders (Hoyle et al., 2002). In their research on two police-led RJ programmes involving juvenile offenders in Northern Ireland, O’Mahony and Doak (2004) found evidence of net-widening, with RJ used for very minor offences. A 2012 report on the uses of RJ by police, probation and youth offending teams in six counties in the UK revealed some further concerns, including ‘widespread inconsistency in the implementation of RJ initiatives across the whole criminal justice system, with conflicting terminology and a lack of understanding of the principles of RJ’ (Criminal Justice Joint Inspection, 2012: 11). In sum, research on police-led RJ in the UK has found that as RJ for low-level offences is up to the discretion of, and facilitated by, individual officers, when their understanding of RJ is lacking, then this is evidenced in their practice (Criminal Justice Joint Inspection, 2012; Cutress, 2015; Shapland, Crawford, Gray & Burn, 2017; Stockdale, 2015). At its worst, this may include net-widening (Criminal Justice Joint Inspection, 2012; O’Mahony & Doak, 2004) and ignoring the ‘voluntary’ aspect of RJ (Shapland et al., 2017: 22). At its best, however, police-led RJ has reduced the criminalisation of children in residential care by over 50 per cent in some UK counties (Criminal Justice Joint Inspection, 2012: 25).
Of particular relevance to this study is research on female offenders. Women who offend are particularly vulnerable; they are more likely to have spent time in the care system due to higher rates of neglect and abuse (Arnull & Eagle, 2009; Sharpe, 2012; Williams, Papadopoulou & Booth, 2012), and they have higher levels of mental health problems than boys and men (Light, Grant & Hopkins, 2013). While the literature on the vulnerabilities of both women and girl offenders overlap, girl offenders are particularly at risk of being pulled into the CJS due to net-widening for a range of minor offences (Burman & Batchelor, 2009) but especially for traditionally ‘unfeminine’ offences such as assault (see review by Himmelstein & Bruckner, 2011). Nearly 30 years on from Chesney-Lind’s (1989) research describing the criminalisation of girls through statutory offences such as running away from abusive homes, links are still being drawn between childhood trauma and girls’ arrests. Sharpe (2012: 62), for example, through interviews with 52 young women involved in the CJS in the UK, makes connections between girls’ experiences of family conflict and convictions for ‘offending’ at home. On top of this, girls who offend often grow up in neighbourhoods with few prosocial peers and opportunities (Sharpe, 2012), leading them to join friendship groups where offending is normalised (Joe & Chesney-Lind, 1998; Sharpe, 2012) and to form ‘relationships’ with partners who exploit them (Cockbain & Brayley, 2012). Given the context of their lives and ‘poly-victimisation’ (DeHart & Moran, 2015), young women’s offending is increasingly depicted in the research literature as ‘normal’ or a display of agency (Batchelor, 2005; Henriksen & Miller, 2012; Sharpe, 2012) as are their ‘difficult’ attitudes (Alder, 2000; Sharpe, 2012).
In response to such research, feminist writers have advocated for RJ as a CJS disposal for female offenders (Gaarder & Presser, 2006; Verrecchia, 2009). Research on how they experience RJ, however, is scarce due to the low number of female offenders participating in RJ (Miles, 2013; Sherman et al., 2008). Daly (1996), (cited in Daly & Stubbs, 2006), in one of the few studies to examine gender across roles in RJ, for example, found that women most often participated as victim or offender supporters rather than as victims or offenders. Miles’ (2013: 33) interviews with twelve practitioners working with female offenders in the UK suggested that low participation in RJ may be due to some professionals’ concerns about female offenders’ vulnerability and their ‘manipulative’ behaviour.
Further, major evaluations of RJ rarely discuss differences in outcomes by gender (Elis, 2005; Daly & Stubbs, 2006). Two RJ studies have, however, shown some intriguing outcomes. In a re-evaluation of McCold and Wachtel’s (1998) Bethlehem, Pennsylvania study, for example, Hayes (2005) found RJ was more effective in lowering recidivism for young women than young men. A second American study (Rodriguez, 2007: 369), comparing juvenile offenders over 24 months after RJ and/or court, identified the same finding: girls were less likely to re-offend than boys, but girls who experienced RJ were less likely to re-offend within a two-year period than girls who experienced a court process for a similar offence.
Two qualitative studies from Australia and New Zealand, which include young female offenders’ own accounts of their experiences of RJ, add to the complexity. Maxwell et al. (2004) analysed family group conferences involving young offenders in New Zealand from 1998 to 2003, which were facilitated by youth justice coordinators and attended by police officers. The authors reported that unlike young men who rated their conferences positively, young women saw less benefit to the process and, specifically, mentioned feeling dissatisfied with their interactions with the police. The authors proposed that these differences in satisfaction and expectations may have been due to a combination of the girls in the sample being more vulnerable and having committed less serious offences, which they may not have felt warranted RJ.
Daly (2008) has, to date, offered the most in-depth analysis of young female offenders’ thoughts on RJ, as part of a larger study on young offenders’ participation in RJ. The South Australia Juvenile Justice Conferencing Scheme, in which the girls participated, involved facilitation by a youth justice coordinator who was accompanied by a police officer (Daly, 2001). Through researcher observation of conferences and interviews with girl offenders, Daly (2008) identified similar negative attitudes amongst young women to those reported by Maxwell et al. (2004). The researchers, for example, described the young women as ‘less often remorseful … more defiant and less likely to apologise spontaneously to victims’ than young men (Daly, 2008: 114). Daly further identified that violent offences involving other young women, whom the offenders had known prior to the assault, resulted in particularly difficult outcomes due to victim blaming. While as Daly (2013) suggests, such attitudes are understandable given women’s previous victimisation and relationships with their victims, victim blaming does not fit the ethos of RJ (Daly, 2013; Dandurand & Griffiths, 2006; Shapland, 2013).
A recent UK study, on behalf of the Restorative Justice Council by Osterman and Masson (2016), focusing on women only presents a more positive picture of RJ for female offenders. The authors interviewed fourteen female offenders between the ages of 15 and 60 who had mainly committed acquisitive offences. On the whole, the women said that RJ provided them with a sense of pride for taking part. This was in spite of some of them feeling ill-prepared for the conference and finding themselves outnumbered in the room on the day. The authors did not specify who facilitated these conferences or their level of RJ knowledge/training; recruitment took place through 75 locations providing RJ in the UK and the Restorative Justice Council’s professional network of practitioners.
Outstanding questions from the literature relate to the fit between the principles of RJ and the way the police practice it; how RJ is used for female offenders generally and by the police specifically; and finally, what the benefits/risks of RJ are for female offenders. This research sought to address some of these gaps through the following questions:

• How was police-led RJ used for female offenders in this county compared to male offenders?

• How did female offenders feel about their offences, victims and police-led RJ?

2. Methods

2.1 Sample

This study examined police uses of RJ for female offenders in one UK county from 2007 to 2012. The county was chosen because the constabulary had integrated RJ into their policing and had kept records of their cases.
According to the force protocol on RJ, which was made available to the researchers, RJ, in the form of either street RJ or conferences, was offered for low-level crimes and non-crimes; Antisocial Behaviour (ASB), defined as acting ‘in a manner that caused or was likely to cause harassment, alarm or distress to one or more persons not of the same household’ (Crime and Disorder Act, 1998: 2); and could be used as a crime prevention step for young people, including those under the age of criminal responsibility (10 in the UK). Street RJ was described as a meeting between a police officer and an offender, often immediately after the incident and/or at the site of the incident, and was deemed to be appropriate for minor offences frequently encountered by the police (shoplifting, criminal damage, assault) involving fewer than four people. These four persons could include victims or support persons, if appropriate. Conferences, led by police officers and attended by a greater number of people, involved more planning and were intended for more serious offending. Police-led RJ could be the sole disposal of any low-level offence for young people or adults; could be offered alongside Penalty Notices for Disorder and Reprimands for young people; and could be an element of conditional cautions, cautions and Penalty Notices for Disorder for adults (in the form of restorative approaches).
The research took place prior to the revision of the Victims Code of Practice when information on RJ became an ‘entitlement’ for victims (Crown Prosecution Service, n.d.; Ministry of Justice, 2015: 35). RJ was instead offered to victims if and when the police felt the case, and the offender, met the requirements of RJ. According to the force protocol, such decisions could be based on factors such as the offender’s age, previous offending, current offence and previous participation in RJ.
The researchers were provided with access to the anonymised police RJ database. Permission was also granted to sample female offenders between the ages of 18 and 30 from the database, with the assistance of a civilian administrator working for the police, for qualitative interviews.

2.2 Quantitative data set and variable coding

As administrative data is often not ‘ready’ to be analysed and may contain inconsistencies as well as missing data (Gorard, 2012; Smith, 2008), the data was cleaned and coded over a six-month period prior to analysis.
Once the data was prepared, two master files were created, one involving all participants and the other involving offenders. The following variables were derived for general participants (N = 17,486): gender, age, ethnicity, role in the process and relationship. Variables in the offender file (N = 7,030) included: gender, age, ethnicity, incident type, type of offence and type of RJ.

2.3 Qualitative participant recruitment and analysis

As the database had been anonymised, the researchers supplied the civilian administrator with ID numbers for female offenders aged 18–30 who had participated in RJ. These parameters were selected to capture a range of offending and RJ experiences. The researchers also supplied the administrator with text about the research for phone calls and letters. The administrator attempted to phone or send letters to 168 women. Of the women the administrator successfully reached via phone, four consented to speak directly with a researcher. Twelve women contacted a researcher by text or email, having received a letter. Out of these sixteen, twelve women were interviewed, reflecting a response rate of 7%.
The low response rate was in part due to participants’ contact details changing over the past five years, but may also have been due to a representative of the police contacting the women. The letter explained that the researchers did not work for the police; however, in the initial conversations between the researchers and the women, some expressed concerns that the police had contacted them. The women were reassured in person that the research study was independent from the police and that their views would be anonymised.
As literature on RJ highlights that many major life changes happen in young people’s lives, making it difficult to determine RJ’s impact, particularly when following up a few years after RJ (Hoyle et al., 2002; Maxwell et al., 2004), and as RJ is described as being about ‘storytelling’ (Umbreit, 1998; Gaarder & Presser, 2006), narrative interviews were chosen. Narrative interviews have in recent years become a burgeoning research method within criminology (Presser, 2009), producing novel insights into complex topics such as desistance (Maruna, 2001) and extreme violence (Presser, 2012; Sandberg, 2013). Because narrative interviews give participants the opportunity to guide the direction of ‘talk’ (Riessman, 1993, 2008; Wengraf, 2001), the researchers hoped the interview format would give women the freedom to explore a variety of topics that might be important to them, including offending, desisting and RJ.
The recruitment materials made clear that the researchers had an interest in women’s whole lives as context to participating in RJ. The letters read:

I’m looking for 20 young women between the ages of 18-30 who are willing to talk to me about their lives, and tell me their thoughts on young women and offending as well as restorative justice.

The interview followed formats recommended by Riessman (1993) and Wengraf (2001), beginning with an opening question of ‘Can you tell me about your life?’ and followed by prompts such as, ‘What were things like when you were young?’ The women differed in how much they spoke about various parts of their lives. Two women focused mainly on the present, their offence and RJ, while others spoke at length about their childhoods, making connections between this and later offending. The interviewer emphasised to the women that they could decide what they spoke about.
The women chose to be interviewed at home, coffee shops or at the university. The interviews resulted in detailed transcripts about the women’s lives, which were analysed thematically and narratively. Thematic analysis allows themes to emerge from the interview material (Braun & Clarke, 2006) while narrative analysis may include a focus on language within an excerpt of text or theme (Daiute & Lightfoot, 2004). Thematic and narrative analysis, used together, can provide a ‘top-down’ overview alongside in-depth, further analysis (see Shukla, Wilson & Boddy, 2014: 5). Qualitative analysis on the women’s experiences in RJ is presented in this article, alongside the quantitative findings. The qualitative and quantitative findings are then considered alongside each other in the discussion.

3. Quantitative findings

The administrative database provided an overview as to how the constabulary used RJ and for whom. As discussed in the methods section, however, the data contained a number of errors, missing data and inconsistencies. Importantly for this analysis, in terms of missing data, there were no significant differences in key offending variables for gender. Multivariate analysis is not presented in this article primarily because the data was not normally distributed and because of the reliability of some of the variables within the data set (i.e. large amounts of missing data). Outcomes of RJ are also not included, as the researchers only had access to notes on individualised RJ agreements rather than more general outcomes/disposals relating to the CJS. For these reasons, the quantitative section will provide an overview of police RJ through descriptive statistics as well as a non-parametric comparison of male and female offenders for key variables where data was sufficient.

3.1 RJ participants

Between the years of 2007 and 2012, 17,486 individuals (victims, offenders, support persons and others) participated in police-led RJ. This included 8,099 women (46.3%), 9,000 men (51.5%) and 387 persons of an unknown gender (2.2%). Of these participants, 69.6% (N = 12,165) identified as white British, 2.4% (N = 419) identified as from another white background, and 2.0% (N = 350) self-identified as minority ethnic. A further 0.5% (N = 79) were identified by the police as white and 0.0% (N = 5) were identified by the police as minority ethnic. 25.5% (N = 4,463) of the ethnicity data was missing. The ethnic make-up of participants reflected the population of a predominantly white county (ONS, 2012).
40.2% (N = 7,030) of the participants were offenders; 28.4% (N = 4,961) were victims; 0.0% (N = 8) were identified as both victims and offenders; 18.7% (N = 3,274) were offender supporters; and 5.8% (N = 1,011) were victim supporters. The remaining 6.9% (N = 1,202) were unidentified supporters and witnesses, appropriate adults or professionals such as interpreters and police officers. Due to data recording practices, it was not possible to identify for whom unidentified supporters and professionals were present (i.e. victims or offenders).
The majority of women in the database participated as offenders, followed by offender supporters, victims and then victim supporters. There were significant differences between men and women by role in RJ, (χ² (14, N = 17,486) = 3288.268, p = .000) and as found by Daly (1996, cited in Daly & Stubbs, 2006), women, in particular, were more likely than men to take on supportive roles (see Table 1).
While the database contained a category on the relationships between participants in RJ, the category was left blank in 73.0% of the cases and did not specify whether relationships were with victims or offenders. The category was, therefore, mainly helpful in providing a flavour of types of relationships between participants. Familial relationships were commonly recorded, for example. Women participated as mothers, step-mothers, foster carers, sisters, daughters, aunts, wives, partners and ex-partners as well as in other roles. Mothers were often present in RJ, including as victims, potentially posing challenges for police facilitators, as offences between parents and children have been noted in the RJ literature as particularly complex due to the unique power dynamics involved (Daly & Nancarrow, 2010).

3.2 Offender characteristics

A total of 7,003 offenders were included in further analysis, after including persons who were described in the notes as offenders but whose ‘role’ had been left blank and excluding offenders who declined or failed to complete RJ. Of these individuals, 37.0% (N = 2,588) were female, 61.0% (N = 4,271) were male and 2.1% (N = 144) were of an unknown gender. 74.9% (N = 5,248) of these offenders self-identified as white British, 2.7% (N = 189) as from other white background and 2.4% (N = 165) as minority ethnic. Additionally, 0.0% (N = 2) persons were identified by police as ‘black’ and 0.0% (N = 1) person was identified by police as ‘white’. 20.0% (N = 1,398) of the offenders’ ethnicities were missing. The difference between ethnicity type by gender was not significant, (χ ² (17, N = 6,859) = 15.638, p = .550).
The offenders ranged in age from 3 to 89; however, RJ was mainly used for young offenders. The mean age of all offenders in the sample was 19.98, with 62.1% (N = 4,124) of the cases involving young offenders between the ages of ten and sixteen (see Table 2). The mean age for male offenders was 20.31, and the mean age for female offenders was 19.49.

3.2.1 Incident and offence type

Incidents were classified in the database as crimes, non-crimes, local resolution, ASB, community resolution or traffic related. Of these categories, 77.9% of the incidents in the database were identified as crimes (N = 5,433), 8.5% as non-crimes (N = 592), 6.4% as local resolution (N = 449), 5.6% as ASB (N = 392), 1.1% as community resolution (N = 76) and 0.5% (N = 33) as traffic. 0.4% had no label and were recorded as missing (N = 28).
The database originally contained 677 offence types, which included misspellings. These were reduced to 95 with the aid of police-recorded offence types (Home Office, 17 July 2014). They included shoplifting and various types of theft; fraud; handling stolen goods; blackmail; burglary; arson not endangering life; assaults of varying types of severity; hate/race offences; indecent exposure; indecent assaults; sexual assaults, abuse of children through pornography; possession of cannabis; ASB; harassment; criminal damage; possession of weapons; traffic offences; and others. Additionally, there were nuisance activities or problematic behaviour to which the police had responded restoratively. These included bullying, school disputes, truancy, family disputes, minors misbehaving and threats.
As the majority of offence types contained very few participants, codes were merged to aid analysis. For example, all types of ‘assaults’ were merged; all ‘thefts’ (except shoplifting) were merged, and so on. Crimes and non-crimes were kept separate for the first few iterations of coding in the assumption that these incidents might differ. The result was two sets (crime/non-crime) of parallel codes involving similar incidents: violence, sexual offences, criminal damage, robbery, burglary, drugs, other theft, fraud, traffic, fear and provocation of violence, possession of weapons, harassment, etc. Due to the relatively small numbers of participants in the non-crimes compared to numbers of participants in the crimes categories, and given the researchers’ interest in types of behaviour for which offenders were referred to RJ rather than the precise criminal code, the crime/non-crime codes were merged for similar incidents and renamed to reflect this merging. For example, the new ‘intimidation’ category included harassment; fear, alarm or distress; as well as bullying and teasing, and ‘damage’ included a wide range of activities involving destruction and defacement of property, including criminal damage. Finally, in order to make meaningful comparisons between female and male offenders, incident types were synthesised to categories containing female offenders. Categories with few to no female offenders such as sexual offences, robbery, burglary and possession of weapons were coded as missing, leading to five final incident types containing both male and female offenders. These were: shoplifting, damage, violence, intimidation and theft.
There were significant differences between men and women by offence type, (χ² (4, N = 6,316) = 965.326, p = .000). Women were most commonly referred to RJ for shoplifting, followed by violence, intimidation, damage and theft. Men, on the other hand, were most commonly referred to RJ for damage followed by violence, shoplifting, intimidation and theft (see Table 3). Shoplifting was the only offence type in which there were more female than male offenders. The dominance of shoplifting as a reason for women to be referred to RJ has been found in other UK studies (Osterman & Masson, 2016).

3.2.2 Type of RJ used

The force protocol on RJ described the use of street RJ and conferences. The data set, however, indicated the presence of a further type of RJ: school RJ. Of these forms, street RJ was most common (53.4%, N = 3,743), followed by conferences (37.2%, N = 2,608) and school RJ (6.7%, N = 472). There was no significant relationship between gender and RJ type, χ ² (2, N = 6,737) = 2.698, p = .259, φ = .020.
Street RJ may have been chosen for the majority of offences to save police and participants’ time. However, this type of RJ may have had implications for how RJ was conducted. For example, a quick intervention at the scene of the offence would allow little time to prepare participants and limited the number of participants present.
In conclusion, while the data set presented some expected findings such as the dominance of women over men in support roles (see Daly, 1996 in Daly & Stubbs, 2006), it also revealed that women participated more often as female offenders (N = 2,588) than in any other role. Female offenders were white, had a mean age of 19.49; and were mainly referred to RJ for low-level offences, including shoplifting, violence or intimidation.

4. Qualitative findings

4.1 Sample description

The women interviewed ranged in age from 19 to 28, with an average age of 22. They had offended between one and five years prior to the interview. Six were one-time offenders, two had offended twice, and four had offended multiple times. Ten women self-identified as white British; one woman as white, foreign born; and one as British, mixed race. They all came from working class backgrounds. The highest educational attainment achieved by the women were apprenticeships and the lowest was leaving school at fifteen. Nine women were employed at the time of the interview. They were referred to RJ for violence, shoplifting, intimidation, damage and fraud. Seven experienced a conference (for violence, intimidation and damage), four experienced street RJ (for violence, shoplifting and fraud), and one was unclear about which she had experienced (for violence).
As expected, given the literature of risk and vulnerability associated with female offenders, and as also as specifically found by Miles (2013), Osterman and Masson (2016) and Daly (2008) in relation to female offenders in RJ, many had experienced victimisation. The most vulnerable had been exposed to domestic violence, abuse and neglect as children.
Women also reported having experienced bullying by peers, sexual assault and intimate partner violence (IPV). The effects of these experiences expressed themselves through mental health issues. Two women had been sectioned due to mental illness or suicide attempts. In contrast, three described ‘normal’, happy childhoods, with the most traumatic experience described as the offence, and subsequent police involvement, for which they were referred to RJ.

4.2 Defining RJ

As described in the methods section, eligible women were contacted prior to the interview and provided information about the research. Despite this process, many did not recognise the term ‘restorative justice’ at the interview when they were asked a variation of the following question, ‘I’m interested in what young women have to say about getting into trouble. Could you tell me how restorative justice came about?’
When the interviewer clarified what RJ could be, a meeting with a victim and/or a police officer after having offended, for example, some women said they attended mediation. Others described it as the disposal they were given because the police did not take their conflict seriously. Women who experienced street RJ sometimes described it as the time they had been ‘told off’ or ‘let off’ by the police.
Participants’ diverse understandings of RJ, and its place in the CJS, had implications for their expectations of the police and victims, as well as aspects of RJ such as ‘accepting responsibility’ and ‘voluntary’ participation (Dandurand & Griffiths, 2006:8). As the differences between women’s experiences in conferences and street RJ were distinct, the two will be described separately.

4.3 Conferences

The main themes across women’s experiences in conferences were: attending alone, mutual culpability, whose side are they on?, defiance and chaos, forced apologies, and leaving it.

4.3.1 Attending alone

While the quantitative data revealed a higher number of offender supporters than victim supporters in RJ, most of the women interviewed did not have a support person present, while their victims did, a finding echoed in other RJ literature involving female offenders (Osterman & Masson, 2016).
Women’s reactions to attending alone involved dismissing the need for support. Becky, for example, turned her lack of a support person into evidence of her self-reliance – a typical coping strategy for young people who have had few people to rely on during their formative years (Samuels & Pryce, 2008).

I don’t need to embarrass her by taking anybody else there who’s going to hear everything that’s going to be said. I’m a big girl. I don’t need somebody else to come with me and hold my hand. I’m quite capable of doing it.

4.3.2 Mutual culpability

As has been identified as common in the criminological literature involving women who have been violent (Arnull & Eagle, 2009; Sondheimer, 2001), and noted as problematic in RJ involving female offenders (Daly, 2008; Miles, 2013), the women interviewed only took partial responsibility for their offence. Similar to previous RJ research, all had assaulted or harassed women they knew as neighbours, classmates or former friends (Daly, 2008; Osterman & Masson, 2016) and all argued that their ‘victims’ deserved the offence (Daly, 2008; Miles, 2013). The reasons they gave included:

• The victim racially harassed the offender.

• The victim, who was dating the offender’s husband, confronted the offender when she was drunk.

• The victim ‘cheated’ with the offender’s boyfriend.

• The victim, when intoxicated, charged at the offender.

• The victim was a ‘slag’ and had words with the offender.

• The victim harassed the offender’s family member.

• The victim and offender harassed each other.

One of the eligibility criteria for offenders to participate in RJ is that they ‘accept responsibility’ for the offence (Dandurand & Griffiths, 2006: 8). Claiming mutual responsibility and blaming the victim during RJ, as these women described doing, therefore, violates a core value of the process and risks re-traumatising the victim (Daly, 2013; Shapland, 2013). How, then did they end up in RJ?
Katia was a one-time offender who assaulted a former friend after she ‘cheated’ with Katia’s boyfriend. Katia readily admitted to the police what she had done, but she also remained adamant she was the real victim, as she had been betrayed. Her motivation for attending RJ was to avoid arrest.

Although I know I done wrong. I said at the time, you know, ‘I know I done wrong.’… They knew I didn’t want to come to this meeting… They said to me, ‘If you don’t turn up we’re going to come to arrest you.’… so I just turned up just for the sake of not being arrested.

Katia may have been a participant who on the surface met the general criteria for participation in RJ; however, her narrative illustrates that admitting to an offence did not necessarily mean accepting responsibility for it, just as Daly (2008) describes. These subtleties may not have been picked up by the police officers. As the next two sections demonstrate, these subtleties had further implications on how she, and women like her, acted in RJ.

4.3.3 Whose side are they on?

Shapland (2013: 66) has described the ‘core’ of RJ as ‘communication’ between victims and offenders. As this section illustrates, however, complex histories between offenders and victims; differences between ‘admitting’ versus ‘accepting’ responsibility; and participating in RJ to avoid arrest instead led to a focus on what the police did and said.
Vicky described being racially harassed by her neighbour before she assaulted her. She admitted to the police what she had done, but as her neighbour had engaged in racial harassment, she felt it was her right to act the way she did and believed the police would agree. In order to encourage the police to take her side, she reported being on her best behaviour during RJ.

I weren’t screaming. That was them. I thought, no. I’m going to be quiet, and you scream and shout and just show the police what you’re really like.

While Vicky said the police addressed the racial harassment in the meeting and included a ‘liaison officer for race’, she felt disappointed that RJ was the sole disposal for both offences and viewed it as a signal from the police that they had dismissed her victimisation.

The outcome weren’t really good because they never went to court for any of the racial abuse and I thought racism…I thought they took that seriously but clearly they don’t because they didn’t go to court or nothing or anything.

Lisa and her friend attended RJ as co-offenders after they were accused of harassing a female acquaintance. Lisa’s version of the story was that the three girls had a verbal disagreement and had traded insults since. Lisa knew the victim had phoned the police, but when RJ was the response, she thought the police had dismissed the victim’s story, ‘It went on so long that I think the police eventually realised that we weren’t actually doing anything but that she was trying to get us in trouble.’ Instead, she recounted with surprise that she was treated like a ‘criminal’ in RJ and was given no opportunity to tell her side of the story.

There wasn’t any, ‘What’s your side of the story?’… It felt like they were on her side. Everything that she said to them was real and they were trying to catch us out on stuff… we sat there feeling like criminals. That’s how it felt.

These examples show a significant gap between how the women saw themselves and their ‘conflicts’ and the way they felt they were viewed by the police and the CJS. As the section below will show, this mismatch sometimes led to the offender taking a deliberate and provocative stance against the police or the victim.

4.3.4 Defiance and chaos

Daly (2008; 2013) demonstrated that when female offenders refused to take full responsibility for their altercations with known female victims, outcomes in RJ were rarely successful. Like Daly’s participants, some of the women in this research undermined the process of RJ in order to maintain face, resulting in chaotic conferences.
Lisa, for example, reported that she and her co-offender deliberately laughed throughout the conference once it became clear that her version of what had occurred would not be listened to.

We went in and we sat there and laughed. We were laughing because it was that stupid.

Becky, who hit her husband’s girlfriend at a nightclub, refused to apologise to her victim, and proudly described herself standing up to the police when they insisted she should:

Eve: She said, ‘You haven’t said you’re sorry.’ I said, ‘I’m not sorry.’ I said, ‘And I’ll do it again.’

Interviewer: And that’s what you said to her at that meeting?

Eve: Yeah, I told her, and the police were telling me they could still arrest me, and I said, ‘Cool.’

These acts of defiance gave the women a feeling of control and added to the sense that RJ was an adversarial process. One participant, for example, described her internal monologue about the victim during RJ as, ‘You think you’re going to intimidate me, well, no, you’re not.’
Occasionally what began as defiance, or a lack of cooperation, evolved into a screaming match. Conferences that spiralled out of control were ended early by the police, and participants were directed to leave at separate times. The women interviewed said this not only made them feel that the conflict had, at least temporarily, become worse through RJ, but that the police had exposed themselves as out of their depth with these new practices:

They [the police] ended the mediation because I was bawling and I was crying and I just wanted to slap her to be honest whether the police was there or not. I just wanted to. I obviously, and well, I think the police sort of just said, because they’ve gone over a level they could obviously control which I think is silly… now every time I see her I just want to fucking hit her for it, excuse my language.

4.3.5 Forced apologies

Conferences that progressed less combatively were described as ending with a ‘forced’ apology. The academic literature on RJ makes it clear that apologies are not necessary to the process (Daly, 2013; Shapland, 2013). Policy literature in the UK, however, promises apologies to participating victims (see Home Office, 1997: 19, 33), which may explain the police’s efforts at encouraging apologies.
Women’s reactions to ‘forced’ apologies differed. Some women, like Becky, refused and said they were threatened with arrest by doing so. Others apologised but resented it. Katia, whose victim ‘cheated’ with her boyfriend, recalled RJ as one of the worst moments in her life because of this apology, ‘I haven’t had many bad moments in my life. That would’ve been one of them.’ While the literature suggests that the presence of police in conferences can reassure participants (Hoyle et al., 2002; McCold & Stahr, 1996), Katia’s story describes involuntary participation and insincere apologies as other potential consequence of police-led RJ.

4.3.6 Leaving it

Women who committed violence or harassment against known female victims tended to describe RJ negatively, citing that they felt they had been treated like criminals and had not been listened to. Surprisingly, however, these same women also reported that their conflicts with their victims resolved after RJ, regardless of how the conferences ended.
For some women, it was the disappointment in the CJS response and the fact that they felt the police had not been on their side, which encouraged them to ‘leave’ it.

If it happens again I won’t be ringing the police, I’ll just be leaving it because there’s no point. It doesn’t go anywhere and it just goes round and round in circles.

Conferences that ended due to screaming and fighting also de-escalated perhaps due to a similar realisation: the conflict would not resolve peacefully unless it was dropped completely.

That worked for me because she don’t, we don’t argue no more. Fair enough. We don’t argue no more so I suppose on a level that worked but on another level that is silly because she gone and told me things that were unnecessary whereas if we hadn’t had that mediation…I just walk past her and try to forget about it.

4.4 Street restorative justice

Four young women experienced street RJ for fraud, shoplifting and violence. While street RJ could include victims, as per the force RJ protocol, none of the sessions the women described attending included victims. This meant that the whole experience of RJ for the women became about how the police responded to them. How individual police officers behaved towards the women could be more or less ‘restorative’ and, in turn, had an impact on the women’s view of themselves and the police.

4.4.1 The restorative officer

Restorative officers were described as those who listened to women’s explanations and accepted their apologies while also holding them accountable. When officers followed this formula, the women reported feeling that their ‘punishment’ (usually community service) was fair. For some participants, being believed and given a second chance, by not being arrested, had the added effect of changing their view of the police.
Rachel, for example, experienced RJ for shoplifting shortly after a sexual assault. As her family had been involved in offending throughout her childhood, and as her case had not gone to court, her view of the police was dismissive. The officer’s decision to offer her street RJ, however, surprised her and softened her perspective on the CJS.

I felt really appreciative about that… I had a bit more respect for the justice system I suppose really.

4.4.2 The unrestorative officer

In contrast, some officers conducting street RJ were described by the women as harsh and as refusing to listen to women’s explanations of their offending.
Anna, a first-time offender who shoplifted a winter coat, described being taken to a police station where she was shouted at.

They were shouting at me. And they were really, really ripping into me and I was a flood of tears… I can’t stand anyone shouting. I can’t stand violence.

As Anna had been sexually assaulted and then sectioned as a result of a suicide attempt following this assault, male aggression triggered her anxiety. The experience left her feeling humiliated, and her focus became on the unfairness of the police’s reaction rather than on her own actions, a common narrative resulting from feelings of shame (Tagney, Wagner, Fletcher & Gramzow, 1992).

5. Discussion

As the study was primarily interested in how police-led RJ was used for female offenders and they experienced these interventions, this section will focus on the intersection of the quantitative and qualitative findings in order to make recommendations for police-led RJ involving female offenders.
There were differences between the two sets of data. The mean age of women interviewed (22) was higher than that of the women in the database (19.49). The typical offence for a woman in the qualitative sample was violence, while in the quantitative sample it was shoplifting. Most of the women interviewed attended conferences, whereas male and female offenders in the larger data set predominantly participated in street RJ. One of the women interviewed was accompanied by a support person; however, offender supporters were more common than victim supporters in the quantitative data. Considering the differences in age, offence type and social isolation, it is possible that the interviewed women’s experiences of RJ differed from those of the ‘average’ female offender in the database. While a sample of twelve interviews is an acceptable number in narrative research, due to the in-depth nature of the analysis (Holloway & Freshwater, 2007), understandings of female offenders’ experiences in police-led RJ would have been strengthened by having a larger interview sample. Similarly, although there have been benefits from having such a large quantitative data set, having access to better quality quantitative data would have allowed for more complex quantitative analysis.
The context that the present quantitative data set provides, however, advances the relatively limited research on female offenders’ participation in RJ in terms of its complexity and range in practice, and the rich qualitative data provides insight into issues such as defiance and denial of responsibility not only for female offenders but potentially for other participants in RJ with prior, complex histories, including family members and neighbours. With this in mind, the findings suggest two main recommendations for practice for police-led RJ involving female offenders. These are:

1. the need for awareness of the backgrounds of female offenders in RJ, prior histories with victims, and potential opportunities and risks, given these prior relationships;

2. the need for clarity about and preparation for RJ with female offenders and caution when it comes to quick processing of offences through RJ, especially when participants have had prior relationships.

In relation to these current findings and recommendations, the evidence-base for RJ has suggested for some years that RJ is most effective for offences involving personal victims who have experienced significant effects related to the offence (see Hayes, 2005; McCold & Wachtel, 1998; Sherman & Strang, 2007). Female offenders’ participation in violent offences, common in this interview sample, fits with this evidence-base. Other research on female offenders in RJ, however, demonstrates that these types of conferences are often unsuccessful because participants’ prior relationships lead to victim blaming and a denial of responsibility (Daly, 2008; Daly, 2013; Miles, 2013). The qualitative data in this study corroborates these findings about the impact of often complex relationships between female offenders and victims, a complexity also evident from the quantitative data on the range of circumstances in which RJ is used.
The narrative methodology used in the qualitative strand of this research suggests that the difficulties female offenders, who are referred to RJ for violent offences involving persons known to them, face in RJ may begin with the volatile context of their lives where aggressive responses to conflicts are ‘normal’ and where their relationship with the victim is viewed as an additional stressor (Batchelor, 2005; Henriksen & Miller, 2012; Sharpe, 2012), leading to feelings of ‘mutual culpability’. Depending on their backgrounds, they might further bring with them into RJ previous experiences of being misunderstood, mistreated and unsupported (‘attending alone’).
RJ, as a CJS disposal, was unclear or alien to the women interviewed in this sample. The sense that they were given a ‘softer’ disposal made some feel grateful to the CJS, others not taken seriously enough, and the rest confused as to whether they had committed an offence. Some believed they could use RJ to convince the police of their innocence (‘whose side are they on?’). When they instead found themselves to be ‘offenders’ and ‘forced’ to apologise, some disrupted conferences with ‘defiance and chaos’ (see also Daly, 2008; Maxwell et al., 2004). These narratives are not only consistent with the backgrounds of female offenders and their ‘coping’ strategies (Sharpe, 2012) but also consistent with experiences of shame, which research demonstrates leads to victim blaming and aggression as a form of self-protection (Tagney et al., 1992).
As research has previously suggested, female offenders who have committed personal offences against known victims may pose challenges to the values of RJ, particularly ‘responsibility’ (see Daly, 2013; Shapland, 2013). To break the cycle of poor outcomes in RJ for female offenders fitting this profile, police may need to be clear with them about what RJ is not only as a process but also as a CJS disposal, what will be expected of them in RJ, and what the alternative to participating in RJ is (see also Daly, 2008; Osterman & Masson, 2016). Police may also need to be conscious of the authority they hold in RJ and may need to shift the focus away from themselves and onto the ‘relationship’ between the offender and the victim (Zehr, 1990). All this might reduce feelings of ‘forced’ participation and insincere apologies, which, in turn, benefits victims.
Police facilitators may also need to be aware of female offenders’ complex lives, particularly their victimisation and their relationship with the victim (see also Daly, 2008; Osterman & Masson, 2016). This study has shown that when some female offenders felt they could speak about the context of their offending and were heard, RJ contributed towards changing negative views of the police and the CJS. The direct value for RJ was that these same women spontaneously apologised, cooperated and accepted the consequences of their actions. Allowing some part of the process of RJ to be about ‘storytelling’ (Gaarder & Presser, 2006; Umbreit, 1998) for female offenders may also permit joint-working with voluntary agencies, particularly those dealing with sexual assault, IPV and mental health, which, in turn, might help break women’s offending pathways in a meaningful way (Osterman & Masson, 2016). This can be accomplished for most participants through careful preparatory work, which, for years, has been identified as crucial in restorative processes involving victims and offenders (Umbreit, 1998).
The quantitative data indicated a preference in using street RJ. Quick processing of offences is a clear benefit for the police as well as to some participants. The qualitative interviews, for example, suggested that street RJ could produce excellent outcomes for female offenders when officers acted restoratively. Osterman and Masson (2016) have advocated for the increased use of RJ for female shoplifters, given the positive outcomes in their research. This research suggests street RJ might be a good option for women who have shoplifted or committed criminal damage, provided officers permit women to explain the context of their offending, thus allowing officers to pick up on any welfare need. This study, however, also shows that quick processing, whether through street RJ or through limited preparatory work, may not be appropriate for interpersonal offences involving victims that are previously known to the offenders. Quick processing may have partly been behind some of the participants’ feelings of being ill prepared for RJ (see also Hoyle et al., 2002; Osterman & Masson, 2016).
This study also goes some way to explain the paradox of why RJ seems to ‘work’ for women, in terms of desistance (Hayes, 2005; Rodriguez, 2007) but lead to negative feelings (Daly, 2008; Maxwell et al., 2004). Women in this sample felt the need to ‘let go’ of conflicts with the victims for a variety of reasons including ill health, work, motherhood or moving away. RJ, however, also encouraged them to handle the conflict themselves in the future rather than involve the CJS, which may be an aim of RJ (Christie, 1977). RJ also made some participants see their conflicts or victims in a new light (whether as unreasonable, dangerous, risky). Women may want to leave offending behind them and can develop the agency to do so (see Giordano, Cernkovich & Rudolph, 2002). RJ may be a vehicle to start this process, regardless of its quality, but more research is needed in this area.
This case study, therefore, does not suggest that the police are inappropriate RJ facilitators. Indeed, women’s experiences with ‘restorative’ officers show that individual police can do a great deal in changing how marginalised individuals view the police and the CJS. The high number of female offenders processed through RJ in this county and the variety of offences committed by them shows a commitment to alternative disposals for female offenders as the Corston (2007) report suggested. From the women’s narratives in the qualitative study, however, and as found in previous research, there seemed to be variation in the type of facilitation and RJ they experienced, including in its quality (Cutress, 2015; Criminal Justice Joint Inspection, 2012; Shapland et al., 2017; Stockdale, 2015) alongside general confusion for offenders over the place of RJ in the CJS (Criminal Justice Joint Inspection, 2012). What this case study, therefore, points to is that ongoing training in RJ is important for police facilitators in order for them to adhere to the core values of RJ, gain more tools to tackle particularly complex conferences, and to safeguard both victims and offenders, as Shapland et al. (2017) have suggested.

6. Conclusion

Policy literature in the UK has proposed a restructuring of how female offenders are processed by the CJS, given female offenders’ backgrounds of abuse, general vulnerability and mainly low-level offending (Corston, 2007). For decades, RJ has been suggested as a possible CJS alternative for women who offend (Gaarder & Presser, 2006; Verrecchia, 2009); however, the research literature thus far has identified that relatively few female offenders participate in RJ, leading to a lack of awareness to how female offenders experience, or benefit from, these processes (Miles, 2013; Sherman et al., 2008). A handful of quantitative studies involving young female offenders have suggested increased desistance compared to the CJS (Hayes, 2005; Rodriguez, 2007); however, qualitative studies involving young female offenders suggest that they may find RJ very difficult (Maxwell et al., 2004; Daly, 2008).
While RJ exists across several branches of the CJS system in the UK (Criminal Justice Joint Inspection, 2012), this study supports the idea that RJ by the police as a front-line and first response might allow more cases involving vulnerable female offenders to be diverted from the CJS altogether. Quantitative data from the study demonstrated that the police used RJ for a wide range of offences committed by women, including acquisitive offences and violence, although the majority of female offenders were referred to RJ for shoplifting. RJ was made available to both young and adult female offenders, with a median age of 19.49. The constabulary conducted conferences and street RJ, but quantitative data showed a police preference for street RJ, a form of RJ consisting of a ‘restorative’ conversation between a front-line police officer and offender(s), often at the scene of the offence.
The qualitative data suggested that when female offenders perceived police officers were acting restoratively, through listening to them and their reasons for offending, they viewed the police and RJ positively. The qualitative data, however, also suggested that police-led RJ for violent offences involving female offenders and victims who had known each other previously to the offence, were experienced as particularly difficult by women and sometimes compromised values of RJ through female offenders claiming mutual culpability and feeling as though their participation and apologies were not voluntary, given the police presence.
This study supports the need for training for police officers in RJ with female offenders, in particular how to handle high emotions and complex cases and how to provide careful preparation for female participants prior to participating in police RJ. The study also demonstrates the value of combining quantitative and qualitative data in this area of research and practice, where large-scale data across time and detailed narratives of women’s experiences are both necessary.