Repeated and Systematic Intimate Partner Violence in Rural Areas in Sweden

Violence against women lacks geographical boundaries, although research demonstrates higher rates of such violence in rural areas compared to urban areas. The repeated and systematic intimate partner violence (IPV) is especially problematic in isolated areas. This study aims to investigate how repeated and systematic IPV, was handled by the criminal justice system in rural areas in Sweden and how risk and victim vulnerability factors were related to recidivism in this longitudinal prospective study. The sample consisted of alleged perpetrators of repeated and systematic IPV who had been either reported, charged, or convicted of repeated and systematic IPV defined according to the Swedish Law Gross Violation of a Woman’s Integrity targeting such violence, in two rural Swedish police districts during 2011–2014 (N = 258). Results demonstrated that 30% of IPV perpetrators were charged with the Gross violation offense and 5% were charged for other IPV-related offenses. The conviction for the Gross violation offense was 11% and 24% for other IPV-related offenses. 56% were not charged or convicted of any IPV-related offenses. Perpetrators convicted of the Gross Violation offense were more likely to receive longer prison sentences than perpetrators convicted of other IPV-related offenses. Victim cooperation in the police investigation increased the likelihood for prosecution with 7.3 times and for a conviction with 6.1 times. In terms of recidivism 24% engaged in IPV towards the same victim and another 27% recidivated into general criminality. Recidivists had higher summary risk ratings and more individual risk factors than non-recidivists, such as general criminality, employment problems and mental health problems, and victim vulnerability factors including personal problems. To reduce re-victimization, risk and vulnerability factors and supporting victims to cooperate in the police investigation should be considered when forming risk management strategies to protect victims of repeated and systematic IPV in such rural areas.

Drawing on the more severe forms of IPV reported to the police in rural and remote areas (Edwards, 2015;Peek-Asa et al., 2011;Strand & Storey, 2019), it is important to examine the criminal justice system's response to this violence since they play a vital role in preventing further re-victimization. This investigation involves studying the criminal justice system process from reporting to the police, the subsequent decision to prosecute or not, to the potential conviction of the perpetrator. To this end, research has primarily examined large-scale government statistics on reporting to the police and prosecution and sentencing at an overall group level, resulting in limited information on how severe IPV cases move through the system. Furthermore, previous research and evaluations of the specific Swedish law for the most severe and repeated forms of IPV, namely the law of Gross Violation of a Woman's Integrity (Swedish Penal Code, Law 2013:367), have neglected to study the characteristics of both victims and perpetrators of this offense. Thus, there is limited knowledge concerning the characteristics of perpetrators and victims, the risk for future IPV, and actual recidivism. These limitations are of concern as the perpetrators of Gross Violation of a Woman's Integrity are, by definition, engaging in repeated IPV with severe consequences for their victims. Thus, research that attempts to identify characteristics and risk factors related to recidivism in these high-severity IPV cases is needed to better protect the victims, as well as heed the call to increase the knowledge of crimes and policing in rural contexts (e.g., Lindström, 2015).
The aim of the present study was twofold; first, to explore and examine rates of reported offenses, charges, convictions, sentences, and victims' cooperation in cases with male perpetrators reported for repeated and systematic IPV according to the law of Gross Violation of a Woman's Integrity in rural areas in Sweden. Second, to compare recidivists with non-recidivists in terms of the perpetrator and victim demographic characteristics, criminal justice system outcomes, risk factors, victim vulnerability factors, and summary risk ratings for IPV.
This study was carried out in Sweden, which has a total landmass of 407,310 km 2 . In 2015, when the data for the current study was collected, the population of Sweden was 9,851,017, meaning that on average there were 24.2 individuals per km 2 (62.7 individuals per mile 2 ). Naturally, individuals are not spread evenly throughout the country, with most people living in the three major cities: Stockholm, Gothenburg, and Malmö. The most scarcely populated areas are in the North, where this study was conducted.
This paper first provides a framework of the study (i.e., the rurality context), followed by an overview of the theoretical background of the law of Gross Violation of a Woman's Integrity, the methods and materials, study results, and a discussion of the results and implications for policy and practice.

Theoretical Background
IPV reported to the police is generally more severe in nature and has often occurred repeatedly and systematically over a long time before being reported. In such cases, according to the theory of normalization of violence, the understanding of violence is redefined by the victim. As such, less severe violence will not be seen as violence (i.e., normalized) as the violence progresses and escalates in the case. Thus, less severe violence will become a normal part of the relationship, and, after a while, only the most severe forms of violence will be defined as violent (Lundgren, 2012;Pornari, Dixon & Humphreys, 2013). One way to escape this negative spiral is for the victim to seek help from others. A turning point for victims is when they realize that the abuse might escalate into lethal violence, at which point they are most likely to leave their abuser. In line with the normalization process, research has shown that women threatened and abused by an intimate partner are at an elevated risk of being victimized by even more severe violence (Alsaker et al., 2011;Kropp, Hart & Belfrage, 2010).
IPV lacks sociodemographic boundaries, meaning that women subjected to violence may be well educated and have social support from an extensive social network of friends and family (Garcia-Moreno et al., 2013). Still, the perpetrator gradually isolates the victim physically, psychologically, and socially until their only contact is with their abusive partner. Furthermore, one could also argue that isolation, in its different forms, is amplified in rural and remote areas where the victim is already geographically isolated. For example, victim support services are generally less available and accessible in rural and remote areas, which makes it more difficult for the victims to disclose their victimization (Edwards, 2015). Additionally, the disclosure of IPV victimization among victims living in rural and remote areas has also been linked with psychological and emotional barriers, uniquely associated with such geographical settings. In smaller communities where 'everyone knows everyone', victims report facing feelings of embarrassment both individually and for their family if they report to police and thus make the violence publicly known (Edwards, 2015;Logan, Shannon & Walker, 2005). In turn, the geographical isolation of victims living in rural and remote areas enables the perpetrator's unchallenged use of IPV (Schwab-Reese & Renner, 2017). Thus, the normalization process of violence, including the perpetrator's controlling and coercive behaviour that can impact every aspect of the victim's life, emerges in this isolation, where the man controls the woman physically, psychologically, and socially. It should therefore come as no surprise that some studies have reported higher rates of re-victimization among victims living in rural and remote areas, compared to victims living in urban areas (Logan et al., 2005;Peek-Asa et al., 2011), although there are studies showing similar rates (Petersson & Strand, 2017).
Furthermore, DeKeseredy and Rennison (2020) describe in their theoretical work on male violence against women in rural areas that patriarchal forms of organizing society are common in rural areas and play an important role in sustaining negative attitudes towards women, and thus increase the risk for further IPV. The social organisation in rural areas, characterized by strong peer support and kinships, resulting in increasing social control, is somewhat contradicting since it contributes to both helping and supporting abused women as well as maintaining the patriarchal norms that accept values that legitimize IPV (DeKeseredy, 2016).
One must have in mind that rural areas also differ across the globe, specifically comparing Global North with Global South, and therefore cannot be generalized. Although the research conducted by feminist criminologists (see DeKeseredy & Rennison, 2020) has strong empirical support, more research is needed in a variety of countries around the world to obtain more comprehensive knowledge of the occurrence and maintenance of attitudes supporting IPV in rural areas.
Situating the normalization process within the framework of isolation, patriarchal norms, and male peer support inherent in a rural and remote context provides an increased understanding that women subjected to IPV in such areas are even more vulnerable than those in urban areas.

The Law of Gross Violation of a Woman's Integrity
Sweden has a long history of government responses to address men's violence towards women, stretching as far back as the eleventh century when King Magnus III enacted a law protecting women against violence, rape, and abduction (Hassan Jansson, 2002). Modern efforts within Sweden to combat violence against women include legislation against marital rape enacted in 1965 (Swedish Government's proposition 2004/05:45), and the reassignment of assault to an offense of public prosecution in 1982, meaning that the government could also lay charges in addition to those laid by victims (Swedish Government's proposition 1981/82:43). These legislative enactments are not exclusive to Sweden. However, Sweden has enacted unique legislation explicitly aimed at combating men's repeated (defined as occurring on more than one occasion) violence toward women. The offense of Gross Violation of a Woman's Integrity (in Swedish 'grov kvinnofridskränkning') was enacted in the Swedish Penal Code in 1998 and revised twice; January 1, 2000, and July 1, 2013 (Law 2013:367). 1 The offense is overarching and is applied when a woman's self-confidence is severely damaged by her current or former partner's repeated and systematic abuse. The purpose of the law was to recognize the repeated and systematic nature of violence against women in intimate relationships and the impact of such abuse according to the theory of normalization, and in doing so, appropriately punish the perpetrator (Burman, 2010). In terms of the systematic nature of IPV, studies have reported high recidivism rates among partner violent men (e.g., Babcock, Green & Robie, 2004;Cattaneo & Goodman, 2005). In Sweden, which to a large extent consists of rural areas, where approximately 34% (3.3 million) of the Swedish population live in rural or remote areas, recidivism rates have been found to vary between 13% and 42% (Belfrage & Strand, 2012;Petersson & Strand, 2017;Svalin et al., 2018), however, not specified on rurality. Nevertheless, these high figures are problematic, considering that repeated and systematic IPV can escalate into severe and life-threatening violence regardless of geographical location (Lindström, 2004).
Specifically, Swedish studies of homicides committed by a current or former male partner showed that 46% of the femicides (homicide of a female) were preceded by separate incidents of threats or violence towards the victim (Belfrage & Rying, 2004;NCCP, 2007). Although the offense of Gross Violation of a Woman's Integrity is unique to Sweden, the issue of repeated IPV is not (Felson, Ackerman & Gallagher, 2005). In sum, repeated IPV occurs in all countries, and such violence has a high recidivism rate, where recidivism also includes femicide.
To support a charge of the offense of Gross Violation of a Woman's Integrity, the abusive acts perpetrated must each be a criminal act. Crimes subsumed within the overarching offense include, for example, physical assault, unlawful threat, deprivation of liberty, coercion, sexual assault or exploitation, harassment, and destruction of property. The overarching charge of the offense is laid instead of laying multiple individual charges. According to the law, there are no restrictions placed on the number of crimes that must have occurred for the offense to be laid. However, in practice, case law suggests that the charge of Gross Violation of a Woman's Integrity is more likely to be laid as the number and severity of the crimes committed increases. According to a recent evaluation (NCCP, 2019), a minimum of three 1 3 acts of physical violence is generally viewed as a pre-requisite by prosecutors.
Although Gross Violation of a Woman's Integrity is a crime of public prosecution, meaning that anyone can report it, the crime is most reported to the police by the victim herself (NCCP, 2019). Since the law's inception, police officers have been able to recommend the charge to the prosecutor based on the female victim's report of repeated and systematic violence by a current or former partner. The police investigation is reviewed by a prosecutor, who determines whether to lay a charge and for what offense (Swedish Prosecution Authority 2006;2012). Official statistics show that a slight majority of reported incidents of Gross Violation of a Woman's Integrity (M = 59%, SD = 4%, range 53%-68%) occurred in urban areas, however, proportionally, it is more common in rural areas (NCCP 22021 Storey & Strand, 2019).
Official statistics show that the offense's conviction rate has been remarkably low. In 2017, the conviction rate for the offense of Gross Violation of a Woman's Integrity was 9% (NCCP, 2019). Additionally, one of three alleged IPV perpetrators charged for the offense of Gross Violation of a Woman's Integrity is instead convicted of other IPV related crimes (NCCP, 2019). Of those perpetrators sentenced for the offense of Gross Violation of a Woman's Integrity during the first years following the enacted law (i.e., 2000-2003), 85% were sentenced to imprisonment with an average prison sentence of 11 months, while the other 15% received probation in combination with treatment (Lindström, 2004).
Aside from official crime statistics, research related to how the criminal justice system handles the offense is very limited, with only a few published studies. Burman (2010) and Eliasson (1999) described the context of the Gross Violation of a Woman's Integrity law and the goal of gender equality in Sweden that motivated the law's enactment. After its implementation in 1998, a follow-up of the offense was done some years later. It revealed that only a minority of such offenses reported to police ever resulted in prosecution. Further, prosecutors often laid the charge of Gross Violation of a Woman's Integrity for offenders with previous violent convictions and more social problems than those without such a history. According to Elman (2001), in contrast to popular belief in the 1990s in Sweden, not only socially marginalized women were subjected to violence from their current or former partners. Victims without social problems would not be adequately listened to since prosecutors were less likely to prosecute if there were no previous convictions for the perpetrators. According to Elman, this offense has therefore not been effective. Conversely, Lindström (2004) showed that police recommendation to prosecute perpetrators under the law, the district attorney's prosecutions, and the length of prison sentences handed down by the court increased after the first years of the law's implementation.
Instead of prosecuting offense-by-offense, it became more common to prosecute with the overarching offense of Gross Violation of a Woman's Integrity. This, Lindström stated, was primarily due to education campaigns. After such educational activities, proficiency on the part of the police and prosecutors applying the law and the establishment of dedicated police victim protection units in Sweden increased. Despite these positive findings, the increase in sentences for the offense of Gross Violation of a Woman's Integrity may not reflect an overall increase in the number of men sentenced to prison for IPV in general. In the past, the same perpetrators may have been convicted under other laws such as assault. Thus, the law was unlikely to be capturing new offenders; it was merely re-classifying offenders that were already known to police and sentencing them to a longer prison sentence.
In the most recent and comprehensive evaluation of the law of Gross Violation of a Woman's Integrity to date, it was concluded that the law, since its inception, has not been applied according to the legislators' intentions, i.e., that to commit several IPV related crimes is a severe crime in itself and a conviction of such an offense include a prison sentence (NCCP, 2019). For example, the findings of this evaluation demonstrated that the application of the law has gradually been restricted, resulting in fewer reports, charges, and convictions. This can partially be explained by the fact that prosecutors have begun to instruct the police to register crime reported by victims separately, since it is more likely to get a conviction for separate charges, resulting in a decline of reported crimes of gross violation of a woman's integrity (NCCP, 2019). The evaluation also demonstrated that most convictions for Gross Violation of a Woman's Integrity consisted of three acts of physical assault, and that physical violence is regarded as a prerequisite by prosecutors for laying a charge for the offense. Thus, the legislators' intentions of the law as encompassing the victim's overall situation, including the psychological forms of violence, have not been reflected in practice (NCCP, 2019).

The Study Area
In the present study, rurality has been defined as per the Swedish Board of Agriculture (2021). Remote areas are defined as areas with less than 5 inhabitants per km 2 . Rural areas are defined as areas with at least 5 inhabitants per km 2 and towns with up to 25,000 inhabitants. Areas with greater numbers of inhabitants are considered to be urban. In Sweden approximately 34% (3.3 million) of the population live in rural or remote areas, which is almost twice as much as the US where 19% live in rural areas (The United States Census Bureau, 2021).
Compared to an urban context, the criminal justice system's response to IPV in rural and remote areas has several challenges. For example, primary responders (e.g., police officers) to IPV in rural and remote areas face practical issues such as longer geographical distances, thus limiting accessibility and response time to urgent calls for help (e.g., Stassen & Ceccato, 2019). Furthermore, rural police are generally under-staffed. For example, Lindström (2015) noted that rural areas in Sweden had few police officers and that the number of police officers in such areas had been decreasing for quite some time. Moreover, first responders to crime in rural areas generally have fewer resources (e.g., Lindström, 2015). In addition, victims of IPV in rural and remote areas might wait a long time to report their victimization to the authorities and thereby having an increased severity of violence when reporting, most likely owing to their more limited access to, and the availability of, support and criminal justice services (Storey & Strand, 2019). This is especially concerning in the light of Ceccato and Dolmen's (2011) findings that the increase of IPV between 1996 and 2007 was markedly higher in remote and rural areas, compared to in urban areas. Thus, responding to, and dealing with, IPV in rural and remote areas presents several unique issues for the criminal justice system.
In this study, data were retrieved from two police districts, Jämtland and Västernorrland, situated in the northern part of Sweden. At the time of data collection 3% of the Swedish population lived in the rural area of Västernorrland, which had 11.3 persons per km 2 (29.3 persons per mile 2 ), and 1% of the population lived in the remote area of Jämtland, which had 2.6 persons per km 2 (6.7 persons per mile 2 ; Statistics Sweden, 2016).

Sample
For this study, we used data from an 8-year prospective longitudinal research project carried out in collaboration with 2 (out of 21 at the time) police districts in Sweden, 1 rural and 1 remote. 2 The project aimed to implement and evaluate the use of structured violence risk assessment instruments for IPV, stalking, and honor-based violence. The two police districts that participated in this project, did not, at the time of the study comply with the Swedish National Police Board's mandatory guidelines for using structured violence risk assessment instruments for IPV and elected to participate in the project. To this end, the Swedish version of the Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER; detailed below) was implemented through this study within the two districts.
A total of 726 male perpetrators were reported for IPV to the police in the two districts between 2011 and 2014 and were assessed using the B-SAFER assessment. Inclusion criteria for this study, however, were cases that consisted of a male perpetrator who had been reported to the police for committing repeated and systematic IPV against a current or former female partner, where a risk assessment with the B-SAFER was completed, and where at least one of the following had occurred; (1) police had recommended a filed report of Gross Violation of a Woman's Integrity, (2) prosecutors had charged the perpetrator with Gross Violation of a Woman's Integrity, or (3) there was a police report of IPV for a perpetrator with a previous conviction for Gross Violation of a Woman's Integrity. Perpetrators for whom it was impossible to obtain follow-up data on recidivism were excluded from the sample. As a result, three perpetrators were excluded because they lacked a Swedish social security number, meaning that they were not registered in the national crime registers, and recidivism data could not be retrieved. Of the 726 alleged perpetrators, 258 (36%) were eligible and included in the study. Of the 258 cases, 233 (90%) were included due to a police report of Gross Violation of a Woman's Integrity (scenario 1), 10 (4%) were included based on a charge of Gross Violation of a Woman's Integrity (scenario 2) and 15 (6%) were included due to a previous conviction for Gross Violation of a Woman's Integrity (scenario 3).

Materials
As a measure to reduce the likelihood of continued IPV, the Swedish Police use the B-SAFER (Kropp et al., 2010) to conduct IPV risk assessments. The B-SAFER has shown both valid and reliable results internationally (Au et al., 2008;Helmus & Bourgon, 2011;Kropp, 2008;Winkel, 2008) and for the Swedish police (Belfrage & Strand, 2008, 2012Storey et al., 2014). However, one study has indicated low predictive validity (AUC = 0.54) within one Swedish police district (Svalin et al., 2018).
The B-SAFER (Kropp et al., 2010) consists of ten risk factors, diveded into two parts i.e., intimate partner violence and psychosocial adjustment, and one part consisting of five victim vulnerability factors (see Table 2), scored on a threepoint nominal scale. For the purposes of research these ratings can be translated into numerical ratings (no/absent = 0, possibly/partially present = 1, and yes/present = 2). The risk factors are assessed in both the past (prior to the last 4 weeks) and currently (in the last 4 weeks). Based on the presence of risk and vulnerability factors, summary risk ratings are made on a three-point nominal scale (low risk = 0, moderate risk = 1, or high risk = 2). The police officers made two summary risk ratings in each case, one to indicate the risk for acute/imminent violence and one to indicate the risk for severe or lethal violence.

Procedure
The present study received ethical approval from the Swedish Ethical Review Board and the Swedish National Police. Data collection was carried out at police headquarters in each police district, which required access to the B-SAFER assessments completed by the police. Demographic information (e.g., age, sex, type of relationship and children) and the risk management strategies recommended by police were collected in each case.
Recidivism data were collected from the national crime register, including all convictions, and the police database, including all reported offenses. IPV recidivism was defined as either a new report or conviction for an IPV related crime. General recidivism was defined as either a new report or conviction for a non-related IPV crime. Reports of IPV in the police database include information concerning the type and date of the offenses and their status (e.g., under investigation, awaiting trial). Convictions in the national crime register include information regarding the type and date of the offense(s) committed and the type and date of a sentence received. The follow-up time for cases was calculated from the date of the first risk assessment and ranged from 35 days to 3.5 years (M = 19.41 months, SD = 9.39 months).

Analysis
Data were mainly categorical, and thus non-parametric analyses, such as χ 2 analyses, were performed. When expected values were below five, Fisher's Exact Probability Test was applied. The effect sizes for χ 2 analyses were presented using φ and Cramer's V. For parametric data independent t-tests, and ANOVAs were used. Effect sizes were calculated using η 2 , where 0.01 was considered a small effect, 0.06 a moderate effect, and 0.14 a large effect (Cohen, 1988).
Games-Howell was used as the post hoc test for ANOVA analyses. All statistical analyses were conducted using SPSS version 24.

Reporting and Outcomes
Most of the perpetrators were initially reported by the victim to the police for Gross Violation of a Woman's Integrity (n = 233, 90%). However, only 30% (n = 78) were charged with Gross Violation of a Woman's Integrity, and 11% (n = 29) convicted (see Table 1). Some of the perpetrators, who were reported for but not convicted of Gross Violation of a Woman's Integrity, were found guilty of other IPV related offenses such as assault and unlawful threats (n = 60, 24%). In total, 35% (n = 90) of the perpetrators were charged with any IPV-related offense, and 8% (n = 20) were still under investigation at the time of data collection. Just over half of perpetrators, 56% (n = 145) were not charged or convicted. Cases were victims cooperated with the police investigation (n = 189, 74%) were 7.3 times 95% CI [3.0-18.0] more likely to result in a charge of Gross Violation of a Woman's Integrity compared to cases where victims did not cooperate (44% vs. 10%), χ 2 (1, 223) = 23.34, p = 0.001, φ = 0.32. Similarly, cases in which the victim cooperated were 6.1 times 95% CI [2.7-13.6] more likely to result in conviction for any IPV related offense compared to cases where the victim did not cooperate (34% vs. 13%), χ 2 (1, 233) = 23.09, p = 0.001, φ = 0.32.
In Table 1 can be seen that of the 89 perpetrators convicted of an IPV crime, 40 (45%) were sentenced to prison, and the remaning 49 (55%) were sentenced to either forensic psychiatric care, secure youth care, were given fines, conditional sentences, or probation. Specifically, perpetrators convicted of Gross Violation of a Woman's Integrity were more likely to be sentenced to prison than those reported for Gross Violation of a Woman's Integrity but convicted of other IPV related offenses (72% vs. 46%), χ 2 (1, 88) = 13.84, p = 0.001, φ = 0.40, and were also less likely to be sentenced to probation (0% vs. 19%), χ 2 (1, 88) = 6.18, p = 0.014, φ = 0.26. No significant differences were found for other sentences such as fines, conditional sentences, forensic psychiatric care, or secure youth care.
There was also a significant difference in length of the prison sentence, measured in months, where perpetrators convicted of Gross Violation of a Woman's Integrity (M = 10.95, SD = 6.70) were sentenced to longer prison terms than those convicted of other IPV related offenses (M = 4.71, SD = 3.36), t(32) = 3.20, p = 0.003. The magnitude of this difference was large (mean difference = 6.2, 95% CI [2.3, 10.2]) (η 2 = 0.24).

Demographic Characteristics and Criminal Justice System Outcomes
Recidivists who committed general criminality were younger in age (M = 35.39, SD = 12.28) than both the non-recidivist group (M = 41.83, SD = 14.14) and those who recidivated towards the same victim (M = 42.49, SD = 13.36), F(2, 255) = 6.21, p = 0.002, η 2 = 0.05. Victims followed the same pattern, that is, the victims of perpetrators who later recidivated via general criminality were younger in age (M = 32.87, SD = 11.45) than victims of both the non-recidivism group (M = 38.71, SD = 13.46) and those who recidivated towards the same victim (M = 38.52, SD = 13.43), F(2, 255) = 5.14, p = 0.007, η 2 = 0.04. Relationship status was the same for all three groups for former or current cohabitation´s or girlfriend's, but not for wives, there were more married couples in the non-recidivists' group (n = 41, 33%) than in the general criminality recidivist group (n = 11, 16%) and the group who recidivated towards the same victim (n = 15, 25%), χ 2 (2, 258) = 6.94, p = 0.031, Cramer's V = 0.16. The three groups did not differ on any other demographical variables including if there were children living at home, or the length of the victim-perpetrator relationship.
Few differences were identified between the three groups in terms of criminal justice system outcomes. No differences were found in terms of the perpetrator being charged or convicted of Gross Violation of a Woman's Integrity or other offenses across the three groups. The sentences for those convicted were also distributed equally across the three groups. The only difference in recidivism between the groups was in the presence of prior convictions. When divided into three groups: no prior convictions (n = 136, 53%), one or two prior convictions (n = 60, 23%), and three or more convictions (n = 61, 24%), those with more prior convictions were more likely to recidivate either by engaging in general criminality (n = 46, 65%) or IPV toward the same victim (n = 60, 50%) compared to the no recidivism group (n = 45, 36%), χ 2 (4, 257) = 25.32, p = 0.000, Cramer's V = 0.22.

Risk and Vulnerability Factors and Summary Ratings
The two recidivist groups [i.e., (1) IPV against the same victim and (2) general criminality] and the non-recidivist group were compared in terms of the presence of both current and past B-SAFER risk factors (see Table 2). Differences were found for perpetrators on B-SAFER items 4 (current: Violation of court orders), 6 (current and past: General criminality), 8 (current and past: Employment problems) and 10 (past: Mental health problems). Only one significant difference was found for the victim vulnerability factors, Item 15 (current: Personal problems). All risk and vulnerability factors were more often present in the two recidivist groups than the non-recidivist group.
The three groups differed significantly on the number of perpetrator risk factors that were present, F(2, 257) = 3.31, p = 0.038, η 2 = 0.01. The non-recidivism group (M = 7.10, SD = 2.77) had significantly fewer risk factors than those who recidivated against the same victim (M = 8.10, SD = 2.84). There were no differences between perpetrators who recidivated into general criminality (M = 8.00, SD = 3.33) compared to the other two groups. Summary risk ratings were dichotomized, where low risk included summary risk ratings of low risk, and elevated risk included summary risk ratings of moderate and high risk. Significant differences were found for acute/imminent risk, where the two recidivist groups were more likely to be rated as having an elevated risk than the non-recidivist group (see Table 2). Table 2 Risk factors [risk factors are dichotomized as present (i.e., B-SAFER ratings of present or partially present) and absent] present in the current and past situation, as assessed with the B-SAFER among 258 cases of gross violation of a woman's integrity, divided into three groups based on recidivism towards the same victim, recidivism in general criminality and no recidivism Note. B-SAFER Brief Spousal Assault Form for the Evaluation of Risk (Kropp et al., 2010), C Current situation, P Past situation. Significant risk factors and risk levels are shown with bold letters. a The distribution of missing ratings for each risk and vulnerability factor did not differ significantly between the three groups b Cells have less expected count than 5

Discussion
The aim of this paper was to study the Swedish criminal justice system's response to the offense of Gross Violation of a Woman's Integrity, which is specific for Swedish legislation, in a rural and remote context. More specifically, among reported cases of IPV, we explored and examined the rates of police reports, charges, convictions, sentences, and victims' cooperation in such offenses. We also compared recidivists with non-recidivists in terms of the perpetrator and victim demographic characteristics, criminal justice system outcomes, perpetrator risk factors, victim vulnerability factors, and summary risk ratings. Gross Violation of a Woman's Integrity is regarded as a severe offense in Sweden and is characterized by repeated and systematic abuse of a current or former intimate female partner. The law prohibiting the offense is internationally unique, but the offense is consistent with the concept of repeated victimization, which is globally prevalent (Hilton et al., 2010;Klein & Tobin, 2008;Lin et al., 2009).
This study can be seen as a case study of the specific Swedish crime incorporated in two rural police districts in Sweden. The outcome of the study can thereby not be generalized for rural areas but are to be seen as a pioneer study on how severe IPV is being handled in Swedish rural and remote contexts.

Perpetrators and Criminal Justice System Outcomes
The results showed that most alleged perpetrators in our sample had low rates of charges and convictions. Although all the perpetrators were initially suspected of repeated and systematic abuse, only 11% were convicted of Gross Violation of a Woman's Integrity. This figure is in line with previous Swedish research and evaluations of the law (i.e., 9-11%: Elman, 2001;Lindström, 2004;NCCP, 2019). The remainder of the sample in this study were convicted of other types of IPV related offenses (24%) or not convicted at all (56%). In sum, there seem to be no differences in conviction rates at the time for the study period, with earlier studies conducted in both rural and urban areas. The same problems seem to remain for rural and remote areas in Sweden as for nationwide results, indicating that the problem with low conviction rates for this specific IPV offense might be its complexity in finding substantial evidence. Although Swedish results clearly show that the law doesn't work as it was intended to, not much has been done to improve it.
Perpetrators convicted of Gross Violation of a Woman's Integrity were more likely to receive longer prison sentences than those convicted of other IPV-related crimes such as assault or threats. If only one offense can be proven in court, the prosecutor will lay a charge for that offense, and the perpetrator will thus be convicted of a single offense. Problematically, this single conviction does not reflect the victim's exposure to the reported systematic and repeated violence within the relationship. Also, some crimes may have been time-barred. i.e., it was too long ago they were conducted so they cannot be reported as a crime anymore, and thereby not possible to prosecute. Severe crimes of IPV such as the offense Gross Violation of a Woman's Integrity are difficult to investigate and bring to trial since substantial evidence is required for each violent, both physical and psychological, act and in many cases it is "word against word" (NCCP, 2019). However, when the police are more observant of the repeated and systematic violence described by the abused women, the police will more likely identify several IPV offenses that can be included in the overarching crime of Gross Violation of a Woman's Integrity. Such observant police work by the Swedish police has been reflected in the increase in police reports of Gross Violation of a Woman's Integrity and charges of the offense (NCCP, 2014). Research also shows that victims from rural and remote areas report more severe violent crimes than in urban areas (AbiNader, 2020;Edwards, 2015;Logan et al., 2003;Peek-Asa et al., 2011;Rennison et al., 2012;Strand & Storey, 2019), which indicates that police in rural areas identify repeated violence cases to a high degree. However, it is still complicated to prove all single incidents in court, and they face the same problems as police in urban areas.
Our results indicate that the first step of the criminal justice process, the police, has functioned well in terms of recognizing the severity of repeated and systematic IPV because they help the victims who report the crimes that it is the offense Gross Violation of a Woman's Integrity. It is not up to the victim to know what crimes under which section of the law they have been subjected to but to describe the crime they want to report to the police and the police will then in the report name the crimes the victims have been subjected to. However, this is only the first barrier to conviction for Gross Violation of a Woman's Integrity and there is a substantial failure in these cases progressing to a charge of Gross Violation of a Woman's Integrity. One reason that cases do not progress to court could be that victims choose not to cooperate with the police investigation, subsequently making it difficult for prosecutors to press charges due to the limited or complete lack of evidence where there is no victim testimony (NCCP, 2016). Our results showed that prosecutors were 7.3 times more likely to press charges if the victim cooperated with the police investigation. However, one must keep in mind that in cases of repeated and longterm IPV, victims may become normalized to the violence and do not want their partner to get sentenced, which means that victims get used to the violence and tend to minimize the seriousness of violent acts, and specifically afterward if an abusive partner apologizes and promises not to do it again they are less likely to continue to cooperate with the police investigation and more likely to withdraw their reporting of the crime (Pornari et al., 2013).
Consequently, the criminal justice system in Sweden has significant difficulties in handling repeated and systematic IPV cases and applying the legislation of Gross Violation of a Woman's Integrity, regardless of the rural or urban context (NCCP, 2019). Our results suggest that efforts should be made to support victims in cooperating with the criminal investigation since that in itself significantly increased the possibility of charges and convictions.
Despite difficulties in pressing charges and obtaining convictions, our results showed that those who were reported and convicted of Gross Violation of a Woman's Integrity were more likely to be sentenced to prison and received longer prison sentences than those reported for other types of IPV offenses. This indicates that the law's aim of sentencing repeated and systematic IPV perpetrators to longer prison sentences are being met. Our results support other Swedish studies (Elman, 2001;Lindström, 2004;NCCP, 2019) that the main problem was that not enough perpetrators were convicted. The low conviction rate for IPV related offenses in the present study were consistent with international research (c.f., Garner & Maxwell, 2009).

Recidivism
Overall, rates of recidivism by IPV perpetrators have been found to be high (Belfrage & Strand, 2012;Hilton et al., 2010;Klein & Tobin, 2008;Lin et al., 2009;Loinaz, 2014aLoinaz, , 2014bSvalin et al., 2018), and our results were in line with this as 24% of perpetrators recidivated towards the same victim. Recidivist cases had more risk and vulnerability factors and received higher summary risk ratings for repeated violence than non-recidivists, thus supporting the predictive validity of the B-SAFER in the two rural and remote police districts. Although recidivism among IPV perpetrators in general is known to be high, clearly, not enough is being done to prevent further IPV in the studied areas.
Previously, Lindström (2004) reported that social problems and extensive criminality characterized men convicted of Gross Violation of a Woman's Integrity. To our knowledge, there is a lack of research about the corresponding characteristics of such perpetrators who recidivate. By exploring the risk and victim vulnerability factors within the B-SAFER and the prevalence of prior convictions, we found in our study that recidivists were characterized by a more antisocial lifestyle than non-recidivists. More specifically, the results showed that the distribution of risk factors between same-victim-recidivists and non-recidivists differed, where recidivists to a higher degree violated court orders, had more general criminality and demonstrated more problems with employment and mental health. The number of risk factors in the psychosocial adjustment part of the B-SAFER was increased in the same-victim recidivist group, which could explain how personal problems made perpetrators less prone to take a pro-social role in society. Victims in both the same-victim and the general recidivist group were more vulnerable due to personal problems than in the nonrecidivist group. Some of the victims had substance abuse or mental health problems, which need to be addressed to reduce the risk of further victimization. The results therefore point to several perpetrator and victim characteristics (i.e., risk and vulnerability factors) that should be the focus of intervention in order to decrease the risk of continued IPV in the studied rural and remote areas.

Limitations
This study's strength was the longitudinal design, which made it possible to follow the cases prospectively. Although the study has not been conducted in randomly chosen areas since it was a convenience sample. Further, the design did not enable comparisons with other rural or urban areas. The two rural police districts can though be said to be typical for rural districts in Sweden, but it is more difficult to claim that they would be representative of rural districts internationally since rurality in itself is very difficult to define (see DeKeseredy et al., 2009). Nevertheless, the detailed definitions of rurality here can be used as comparisons for future studies.
Relying on official data to identify recidivism most likely underestimated the actual recidivism rate. Rennison, Dragiewicz and DeKeseredy (2013) found in their study on violence against women in North America that both IPV and sexual violence are underreported to the police. Even though we have been able to collect recidivism data according to reports or new incidents within the on-going investigations, there will have been missed instances of recidivism not reported to the police due potentially to the normalization of violence and the added difficulties of living in rural areas (NCCP, 2014). This would be in line with results from studies in North America that show less reporting of IPV in rural areas compared to urban and suburban areas (Rennison, Dragiewicz, & DeKeseredy, 2013).
Further, since 26% of victims chose not to cooperate with the police in the investigation, it might also be that victims did not report IPV recidivism. Also, in cases where the initial police report did not lead to the perpetrator being prosecuted or convicted, it is possible that the victim will have lost faith in the criminal justice system and chosen not to report future IPV. Although recidivism was likely underestimated there is no reason to suspect that there was a bias in the recidivism data that undermining certain groups of victims or perpetrators of more severe violence, and which would 1 3 therefore undermine the results as compared to other studies. More research needs to be conducted with other designs that investigate issues around victim cooperation in police investigations, specifically in rural areas, to more fully understand why some women chose not to report further abuse.
Another limitation is the varying follow-up times, meaning perpetrators had a differing time-at risk for offending. In line with different time-at risk, some perpetrators also were sentenced to prison and thereby could be considered to be at no or low risk, although some did re recidivate during their incarceration. Therefore, we have consistently used the term 'at risk' to define the period between when the risk assessment was conducted, and the perpetrator was followed-up. However, most recidivism is shown to occur within the first-year post-intervention (Klein & Tobin, 2008;Lin et al., 2009;Loinaz, 2014aLoinaz, , 2014bRichards et al., 2014;Svalin et al., 2018), and consequently, we may assume that most of the perpetrators in the present study had a long enough period of follow-up while not incarcerated (M = 19.4 months, SD = 9.4) to be detected for recidivism.

Conclusions and Implications
The overall conclusion of our study demonstrated several difficulties for the criminal justice system in the two rural areas when managing cases of repeated and systematic IPV according to the offense of Gross Violation of a Woman's Integrity, including charging, convicting, sentencing, and preventing recidivism among IPV perpetrators. The results are in line with other research and official statistics in Sweden of the offense of Gross Violation of a Woman's Integrity (Lindström, 2004;NCCP, 2019). Those convicted of the offense of Gross Violation of a Woman's Integrity were more likely than perpetrators convicted of other IPV related offenses to receive a prison sentence and be sentenced to longer times in prison, which is in line with the intention of the law. However, only 11% of perpetrators were finally convicted of Gross Violation of a Woman's Integrity, which is problematic. One reason for the low conviction rate could be that one out of four victims did not cooperate in the police investigation, which decreased the likelihood of the perpetrator being charged and convicted by over six times. The findings also suggest that individual factors such as general criminality, employment problems, substance abuse, and victim's personal problems contribute to the risk of recidivism, which is in line with other research on risk factors of IPV and recidivism (e.g., Belfrage & Strand, 2012;Lin et al., 2009;Loinaz, 2014aLoinaz, , 2014bSvalin et al., 2018). Consequently, there is a need for more focus on risk management strategies for perpetrators and victims of repeated and systematic IPV to reduce re-victimization. However, considering the similarities with previous research and evaluations of severe IPV, the difficulties in handling cases of repeated and systematic violence are not unique for a Swedish rural and remote context.
Our results demonstrate that the police were able to identify repeated and systematic IPV in the researched rural areas. However, considering the low rates of conviction and the high rates of IPV recidivism among the perpetrators reported of Gross Violation of a Woman's Integrity in this study, it is apparent that the subsequent parts of the criminal justice system failed victims. To reduce recidivism, the criminal justice system needs to have functional working policies that focus on victim protection, such as offering standardized victim safety programs, to better help victims of repeated and systematic violence cope with their situation. DeKeseredy, Donnermeyer and Schwartz (2009) suggest that a gendered Second-Generation Crime Prevention Through Environmental Design (CPTED) framework could be used to prevent IPV for women in rural communities. The suggested framework of a gendered CPTED includes parts like community culture, connectivity and pro-feminist masculinity, community threshold and social cohesion. By working actively to change culture and norms towards a more sharing and open community, reporting the IPV to the community would increase. Further, the pro-feminist approach will help to reduce the rural patriarchal norms and strengthen victims' pro-social networks (DeKeseredy, 2016). The results of our study support this suggestion since it most likely would increase the victims' participation in the judicial process. Thus, our results can be used to support the change and enhancement of policies on how to use resources for the protection of women victimized of severe, repeated, and systematic IPV, in the studied rural and remote areas.
The results related to victim cooperation suggest that police victim protection units that are specifically trained to work with IPV should conduct the investigations to gather the evidence needed for the prosecutor to press charges. Those units are specifically trained to respond to IPV victims, making them more likely to obtain victim cooperation rather than untrained police officers who may not have the same knowledge about IPV. However, there are few police officers in rural areas, and it might not be possible to have victim protection unit teams specialized in IPV in such geographical settings. At least, however, it is essential to have police officers specialized and responsible for IPV offenses and victims' protection in rural and remote areas. In our study the small police units had specific police officers assigned to these police tasks according to their organizational structure, although they were few and also assigned with other police tasks, which makes the organization vulnerable, and they did not have enough time allocated to the IPV police tasks. How the police work to prevent intimate partner violence is important and would benefit from being conducted in a more structured way than currently (NCCP, 2016), specifically in rural areas. One way to improve the police work is to learn from the structurally work police do when assessing risk for IPV, which have shown good results, and it is time to include this structure also when working with risk management to reduce recidivism (McEwan, Bateson & Strand, 2017), specifically in rural police districts such as the ones studied.
When women are repeatedly victimized, there is a need for protective actions that can be used over time to protect them immediately, during and after the judicial process. Protective actions such as protective living, alarm packages, and restraining orders are some of the measures that can be used. For these to work effectively, the cooperation between the victim, police officers, and other authorities needs to be established at the start. Our study shows that perpetrators were 6.1 times more likely to be convicted when the victim cooperated in the police investigation. The results provide support for victims' safety programs that provide imminent and ongoing support to victims, which is in line with the work done by Rennison, DeKeseredy and Dragiewicz (2012). For example, the victim may need help with relocation to move from the perpetrator, and therefore a safety plan is essential. For some victims, it can be crucial to get an individualized program. Then victims can get help with their specific needs related to the criminal investigation process, where victim advocates can be present and support the victim in cooperation with the police investigation, which increases the possibility to press charges to the perpetrator by six times. Victim advocates can provide support to victims to navigate the judicial process and provide testimony. Even though protective actions and victim advocates already exist, to some extent, they are neither mandatory nor comprehensive in Sweden. A victims' safety program for cases of repeated and systematic IPV, where recidivism was found to be high, might help to reduce repeated victimization in rural areas.