Protecting victims of intimate partner violence: Swedish prosecutors’ experiences of decision-making regarding restraining orders

ABSTRACT

Restraining orders can be used as a risk management strategy to reduce the likelihood of intimate partner violence (IPV) re-victimisation. The aim of this study was to examine how prosecutors work with cases of IPV, with a focus on their collaboration with police, use of violence risk assessment and implementation of restraining orders. A qualitative analysis was conducted based on semi-structured interviews with five prosecutors operating in two northern police districts in Sweden in 2016. Data were analysed using latent content analysis. Three overarching themes arose: The case, Organization of resources and Interpretation of the law. Each theme was discussed in the context of the prosecutors’ work with IPV. Prosecutors pointed to several inadequacies in the legislation and offered potential solutions that would ameliorate their work. Results also showed that prosecutors seldom used violence risk assessments conducted by police as a basis for issuing restraining orders. The primary reason for this was a lack of clear routines governing cooperation between police and prosecutors in the application process. The results from this study can be used when training criminal justice personnel in order to obtain a better understanding of the difficulties that prosecutors face when trying to protect victims of IPV.

KEYWORDS:

• Intimate partner violence
• prosecutor
• restraining order
• police
• criminal justice system
• qualitative study

Introduction

Intimate partner violence (IPV) is recognized as a prevalent global problem, and although underreported, those cases which are reported are managed primarily by the criminal justice system. The global prevalence of IPV perpetrated by men towards women is estimated to be between 15% and 71% (Garcia-Moreno, Jansen, Ellsberg, Heise, & Watts, 2006). A Swedish national survey conducted by the National Centre of Crime Prevention (National Council for Crime Prevention [NCCP], 2014) found the lifetime prevalence for victimization of IPV to be 24% for women and 15% for men, and that 7% of both men and women had been victimized within the past year.

Members of the criminal justice system, specifically police and prosecutors, are tasked with responding to acts of IPV and protecting victims from future violence. Rapid responses are important since research shows the highest risk for IPV recidivism is within the first three months following a report of IPV (Peterson & Strand, 2017; Richards, Jennings, Tomsich, & Gover, 2014; Stansfield & Williams, 2014; Svalin, Mellgren, Torstensson Levander, & Levander, 2014). The implementation of restraining orders can be an effective way to reduce IPV recidivism, specifically if it is used as part of a risk management plan (Lindström, 2005; Strand, 2012). A restraining order is a legal restriction that prohibits one person from contacting another person. In Sweden, victims of IPV apply for restraining orders through the Probation Authority, most often with the help of the police, and prosecutors decide whether applications are granted or not. Since prosecutors have the authority to grant or reject restraining order applications, it is important to determine how prosecutors in Sweden work with cases of IPV. Further, given the substantial ramifications that restraining order decisions can have on victims, it is important to gain a deeper understanding of the complex task that prosecutors in Sweden face.

During the last two decades, IPV has been a priority on the political agenda in Sweden, and many policies have been established to end men’s violence against women (NCCP, 2014; Lindström, 2005). The law (1988:688) governing restraining orders was introduced in 1988. The law was designed to protect and support all individuals exposed to harassment and stalking. Although initially, the law was interpreted and used as a way of combating male-to-female intimate partner violence, a change in the law in 2011 broadened its scope to better include all victims (i.e. not just female victims of IPV, but also other victims of stalking and harassment) (NCCP, 2015; Lindström, 2005).

Although legislation related to restraining orders differs from country to country, restraining orders are to a large extent similar globally, with some differences in terminology (alternative names include protective order and no contact order) (Strand, 2012). Police in Sweden are obligated to inform victims of IPV about restraining orders and how to apply for them. Restraining orders forbid the prohibited person (the perpetrator) from actively trying to engage in contact with the protected person (the victim) during a specified period of time set by the prosecutor, typically 3–12 months (NCCP, 2015; Lindström, 2005). Four types of restraining orders have been enacted in Sweden to extend and specify the restrictions being placed on the perpetrator (see Table 1).

Table 1. Types of restraining orders in Sweden (Law 1988:688).

Type of restraining order	Definition
Ordinary restraining order (1§)	Forbids the prohibited person from trying to engage in contact with the protective person
Restraining order in shared residence (1a§)	In addition to the ordinary restraining order, this order forbids the prohibited person from residing in a shared residence with the protected person
Extended restraining order (2§)	In addition to the ordinary restraining order, this order forbids the prohibited person form being in a specified geographical area surrounding the protected person’s work place or residence or other space that is commonly used by the protected person
Especially extended restraining order (2a§)	This order increases the geographical area described in the extended restraining order and assesses compliance using electronic monitoring

The law governing restraining orders has been criticized due to the low number of applications granted by prosecutors, and because the orders do not provide the level of victim protection originally intended by the legislators (NCCP, 2015). In response to these critiques, the law in Sweden has been revised. The most comprehensive revision of the law was conducted in 2011, as three major changes were implemented. First, electronic monitoring of perpetrators was implemented. However, owing to technical difficulties with the monitoring systems, this addition has rarely been used (NCCP, 2015). Second, when determining whether to grant an order, prosecutors may consider whether the perpetrator has committed a crime, specifically a violent crime, against someone other than the victim applying for the order. This information about prior criminality can be used as evidence of a risk of future violence. Previous legislation only considered whether the perpetrator had committed a crime against the applicant. Over time, a prior crime against the applicant became an unofficial requirement for granting a restraining order (Lindström, 2005). Thirdly, the prosecutor can impose a longer prohibition time than was previously available for restraining orders in a shared residence, extending the maximum period from 30 days to two months (see Table 1).

The NCCP (2015) was given the task of evaluating the effects of the changes to the law on restraining orders and did so by interviewing prosecutors and police. The results highlighted two recommendations. First, non-specialized police who work with restraining orders require special training in order to better understand what is required as part of an application (i.e. that there is a risk that the prohibited person will commit a crime towards the protected person). Secondly, the police must also work in close contact with the decision-makers (i.e. prosecutors), so that they can clearly identify for victims what must be included in their restraining order applications. It was also notable that most prosecutors reported that their practice had not been influenced much as a result of the revision of the law.

In Sweden, victims apply for restraining orders from the Swedish Prosecution Authority. Prosecutors then decide whether the order will be granted. To help guide their decision, prosecutors have a handbook that directs them on how to consider applications (The Swedish Prosecution Authority, 2016). Per the guidelines, all applications should be handled quickly, which in practice means that a decision should be made within one week. However, this is only a guideline, and timeframes can be extended under certain circumstances.

The handbook also requires that the decision be based on a proportionality assessment. This means that that the restraining order is granted based on the severity of the crime committed towards the victim, as well as the outcome of a violence risk assessment, where the risk for future violence, harassment and/or threats is considered. This means that an order can be granted where a severe crime has occurred, or where a crime was not severe but the violence risk assessment indicated a need for victim protection in the case. The NCCP has in several reports concluded that to fulfil this requirement and ensure uniform decisions, prosecutors should use the structured violence risk assessment tools that all police in Sweden complete as part of their investigations (NCCP, 2007b, 2015).

Police use of structured violence risk assessment tools to prevent future IPV has been a priority in Sweden over the past decade (The Swedish National Police Agency, 2010). How cases were handled by the police may have a significant impact on whether victims will be exposed to recurrent violence or not (Belfrage & Strand, 2012; Strand, 2012; Strand & Storey, in press). Specifically, structured risk assessment tools help to guide decision-making on case prioritization and the implementation of appropriate risk management strategies such as safety talks, alarm packages, contact with social services and shelter accommodation (Broidy, Albright, & Denman, 2015; NCCP, 2010; Storey, Kropp, Hart, Belfrage, & Strand, 2014; Belfrage & Strand, 2012). A commonly used tool amongst police officers in Sweden to assess risk for future IPV and identify appropriate management strategies is the Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER; Kropp, Hart, & Belfrage, 2008, 2010). The B-SAFER has shown good reliability and validity and has been shown to assist police officers when working to prevent future IPV (Au et al., 2008; Belfrage & Strand, 2008; de Reuter, de Jong, Reus, & Thijssen, 2008; Kropp, 2008; Storey et al., 2014; Winkel, 2008).

Prosecutors do not use the B-SAFER when assessing risk for further IPV, nor do they use the risk assessment conducted by the police. They do not, as a standard procedure, use any risk assessment tools. Instead, they use unstructured professional judgement when conducting risk assessments, meaning that they rely on their own professional judgement and experience when assessing risk for future IPV (The Swedish Prosecution Authority, 2016). Research has shown that violence risk assessments based on unstructured professional judgement are less reliable and valid than violence risk assessments that utilize structured methods of assessment; in fact, predictions using unstructured professional judgement are typically no better than chance (see Otto & Douglas, 2010). Prosecutors argue that using unstructured judgement, they maintain an objective and holistic perspective. In fact, the only structure they apply to their judgement is the use of a handbook which holds that one of the most important factors to consider is the presence of a prior conviction against the current victim (The Swedish Prosecution Authority, 2016). As such, evidence regarding prior violence becomes extremely important in their decision-making about granting a restraining order.

Restraining order decisions are public, meaning that the decision as well as the information used to make a decision are not confidential. Further, the potentially prohibited person has the right to be informed about the information that constitutes the basis for the decision. This is regulated through a principle called party insight, which means that prohibited persons have the right to know the basis for the decisions made about them (The Swedish Prosecution Authority, 2016). This requirement has been cited by prosecutors as a reason for not considering the violence risk assessments completed by the police, because those assessments contain sensitive information about the investigation and the risk management strategies implemented by the police to protect the victim. However, in a report by The Swedish National Police Agency (2010), it was stated that there were ways to work within the requirements of party insight without revealing this sensitive information. Specifically, it was suggested that by only obtaining a summary of the violence risk assessment or an overall recommendation from the police, sensitive information about the victim could be kept out of the restraining order application.

Prosecutor’s decisions on restraining orders are not absolute and can be overturned. From October 2011 to October 2013 in Sweden, 26,045 decisions were made by prosecutors on requests for restraining orders, and 32% of decisions were to grant the orders (The Swedish Prosecution Authority, 2015). Two per cent of decisions were appealed in court, where 17% were overturned. In cases where a restraining order was granted but then rejected by the court, it was typically motivated by the fact that the prohibited person had not been previously convicted of a crime against the protected person or anyone else (The Swedish Prosecution Authority, 2015). Thus, although a prior conviction is not necessary, in practice prior convictions have become one of, if not the most important, requirements for granting a restraining order (Lindström, 2005).

The aim of this study is to examine how prosecutors make decisions regarding restraining order applications made by victims of IPV. Specifically, we will investigate, how prosecutors experience their task of deciding upon restraining orders, and how they use information from violence risk assessments completed by police in their decision-making process.

Material and methods

A qualitative interview study was conducted to deeply investigate prosecutors’ experiences of their work with restraining orders. An inductive approach was chosen which is suitable when knowledge is fragmented (Elo & Kyngäs, 2007; Hsieh & Shannon, 2005). This type of design is also preferable when there is no existing theory in the area and the research literature is limited (Hsieh & Shannon, 2005).

Participants

This study was part of a larger research project concerning the risk assessment and risk management of IPV conducted in the rural and remote counties of Jämtland and Västernorrland in Sweden (see Peterson & Strand, 2017; Peterson, Strand, & Selenius, 2016; Strand & Storey, in press; Strand, Petersson, Fröberg, & Storey, 2016). Rural and remote counties were chosen for the larger research project since Sweden to a large extent consists of rural areas and previous research on IPV has shown differences in IPV perpetration and practices based on rurality (Edwards, 2015; Lanier & Maume, 2009; Peek-Asa et al., 2011). Nine prosecutors were approached, of whom four declined participation due to workload. The remaining five prosecutors, four women and one man, were interviewed. The prosecutors had been working in their current positions for between 4 and 20 years. They all had experience deciding upon restraining orders as part of their current position.

Procedure

Prosecutors were recruited via an email sent to the heads of the Swedish Prosecution Authority in Jämtland and Västernorrland. The names of all nine prosecutors who were, at the time, the only prosecutors working on IPV cases were obtained, and each prosecutor was thereafter contacted by phone to schedule an interview. Data were collected in April of 2016 through interviews held at the Prosecution Authorities office. Interviews were conducted one-to-one in a place of the prosecutors’ choice and lasted for a mean of 39 min (range: 27–49 min).

Before the start of the interview, prosecutors signed an informed consent form. All interviews were audio-recorded. Before transcribing the material, a transcription notation system was constructed to ensure uniform transcription. Transcription was done orthographically and began in connection to the first interview. Transcription was completed three days after the last interview. As the interviews were conducted in Swedish, quotes used in this paper were translated into English. Some restructuring of the wording of quotes was necessary to improve clarity, but care was taken to maintain the quotes’ original meaning and content.

Material

Data were collected through semi-structured interviews where topics and questions were organized in advance. This method allowed the interviewer to be flexible when probing and encouraging prosecutors to develop their thoughts and ideas. An interview guide was constructed with open-ended questions and follow-up questions. As recommended by Braun and Clarke (2013), the interview guide began with opening questions that were neutral to get the interviewees warmed up and introduce to the topic of interest, and later more specific questions targeting the aims of the study were used. Questions that were asked during the interview included ‘What do you think of the restraining order as a risk management strategy for IPV?’, ‘What difficulties do you encounter in your work with restraining orders?’ and ‘How do you work with violence risk assessments in the decision-making process of restraining orders?’ At the end of the interview, prosecutors were given the opportunity to make additional comments.

Analysis

A latent content analysis, guided by Graneheim and Lundman (2004), was used. Content analysis is employed to identify patterns and themes in the data, and the latent aspect of the method refers to the analysis of the underlying meaning of the spoken words. The author SF conducted the initial analysis, findings were then discussed with author SS, and the final step of the analysis, constructing themes, were conducted by both SF and SS.

The first step in the analysis was familiarization with the data. This was completed by compiling all data into one unit (i.e. all written material from the interviews was put into one document). The document was then read three times to obtain a sense of the data (Braun & Clarke, 2013). During the final reading, initial ideas about the data were written in the margins of the material (Elo & Kyngäs, 2007). Next, the text was divided into meaning units (i.e. specific parts of the text that pertain to the aims of the study). Thereafter, meaning units were condensed to reduce and shorten the text within each unit without losing the core content of what had been said. The condensed meaning units were then marked with a code that reflected the content and the core of the meaning unit.

The next step in the abstraction process was to organize the codes into categories and potential subcategories by comparing the differences and similarities amongst the codes. This was done by first obtaining an overview of the coded material, and then preliminary categories and subcategories were created. A coding scheme was developed to define the categories. During this process, a mind map was used to help visualize how the codes could be grouped. The abstraction process was conducted through a back and forth analytical process (Braun & Clarke, 2013). After the initial grouping of the codes, the material was reviewed and the coding scheme was revised to fit the data. This resulted in two categories being merged into one, two categories being renamed and one subcategory being rearranged. When this was complete, the material was reviewed again to ensure that the coding was accurate. No further changes were made to the coding scheme or in the coding of the data.

In the final step of the analysis, the underlying meaning of the data was sought. Three overarching themes were formulated that were not mutually exclusive, meaning that a meaning unit, a code or a category could fit into more than one theme (Graneheim & Lundman, 2004). This stage of the analysis also consisted of defining the themes by identifying the core elements of each theme. This was done by going back and forth between the meaning units and the themes and accompanying the themes with a narrative. An example of a meaning unit is ‘You shouldn’t consider surrounding factors because the prohibited person must know what your basis for a decision is, they should be notified about the application and its contents’, which was condensed into ‘The basis for deciding whether a restraining order should be granted must only contain information that can be shared with the prohibited person’. The meaning unit was then coded as ‘Notification of basis for decision’. The code was part of the theme ‘The case’. For each theme, an explanatory text was written.

Results

The latent content analysis revealed three overlapping themes: The case, Organization of resources and Interpretation of the law. The findings presented are accompanied by narrative quotes from the prosecutors that illustrate the themes.

Theme 1: the case

The prosecutors described their need to balance victim protection and the criminal investigation within a case. This balance was described through their need to make compromises when balancing these often competing goals. For instance, if a restraining order is granted for the victim, the perpetrator knows that the victim has reported him to the police and filed charges. In some cases, prosecutor and police need time to investigate the crime and collect evidence before interrogating the perpetrator. Thus, the notice of the restraining order gives the perpetrator time to potentially tamper with evidence or create a false alibi. Prosecutors agreed that it is important to quickly secure evidence in order to proceed to trial. They believed that this sharing of information could jeopardize the investigation through the loss of evidence or evidence tampering:

You can ruin a criminal investigation by applying for a restraining order, because then the person under investigation finds out that they are suspected of a crime (pause) so that’s a huge drawback. (Prosecutor 1)

All prosecutors stressed their objective position and their primary function as the leader of the investigation. They noted that this role can come in conflict with their concurrent role of working to prevent harm to the victim through issuing restraining orders. Handling both restraining order applications and criminal investigations was further described as difficult due to the communication requirements between the prosecution and the defence when issuing a restraining order. As illustrated by the quote above, conflict arises when the communication requirement forces the prosecution to notify the prohibited person about a restraining order application before the criminal investigation is initiated. As a result, the prohibited person then learns that there will be a criminal investigation. The prosecutors also expressed concern about the need to inform the prohibited person about the basis for granting an order, since this could reveal sensitive information prior to trial about the case, such as what the victim has told authorities and where the victim currently resides. Many of the prosecutors stated that they, in consultation with the police and the victim, therefore sometimes decide to postpone the restraining order application:

Sometimes there may actually be a reason (to postpone a restraining order application) you know, you might report for example rape or something (pause) then you can ruin the whole investigation when he finds out that (sigh), now he is notified about the restraining order application and also by the way, that he is suspected of rape, yeah then the investigation is ruined. (Prosecutor 1)

The prosecutors discussed the consequences of postponing a restraining order application, since postponement would leave the victim unprotected during the critical time when the criminal investigation is initiated. Ultimately, the prosecutors concluded that their primary role is as the leader of the criminal investigation:

In some of the cases I have actually turned to the police and asked does she really want this? I mean how important is it to have a restraining order in comparison to having this crime solved? You know. And then sometimes it has happened that they have postponed the restraining order application until certain things were secured. (Prosecutor 1)

Theme 2: organization of resources

All prosecutors indicated the importance of the role that police play in their work, particularly in the criminal investigation but also in the work they do to protect victims. Prosecutors often requested the insight of the police into the cases. Police were described as an important source of information because they had typically met with the perpetrator and/or victim. Prosecutors found that generally they and the police viewed the cases in the same way and agreed on how to proceed. The prosecutors appreciated times where they and the police could discuss investigative methods and identify what evidence was necessary to gather. Although prosecutors agreed that they listen to input from the police they all pointed out that they as prosecutors must make the final decision:

It doesn’t happen often, but it surely happens that we have different perceptions, the (police) investigator and I. I appreciate if you can discuss the case, but in the end when it comes down to it, I’m the one who makes the decision. (Prosecutor 5)

Prosecutors had a clear view of their tasks and how their cooperation with police should function in criminal investigations as well as in granting restraining order applications, which are two separate processes. However, their view of police roles shows that there is ambiguity regarding the distribution of the resources within the police. In actuality, police are involved in both the criminal investigation and victim protection, and typically these roles will be taken on by two different police officers, although those officers may collaborate to some degree. The prosecutors reported that they believe there to be a lack of routines within the police regarding the division of responsibilities for the investigation as well as when applications for restraining orders are made. Prosecutors were unsure as to who within the police dealt with restraining order applications and who dealt with the investigation, and whether the same police officer was responsible for both tasks. The prosecutors suggested that the investigation and restraining order processes should be separate tasks:

The application for a restraining order, that’s a completely separate process from the rest of the primary investigation that’s a whole different department within the police. (Prosecutor 1)

The criminal investigation and the restraining order process were therefore described as two separate procedures, although it was stressed that they are somewhat dependent on each other. The prosecutors indicated that it was important to synchronize the two procedures to ensure that there was sufficient information to make a decision about granting a restraining order. Prosecutors reported that it was common for them to have to ask for additional information, particularly a victim interview conducted by the police, prior to deciding on a restraining order application. However, knowing that such an interview would take a lot of police time, prosecutors stated that if they knew that they were not going to grant a restraining order in a case, they would not request an interview as they felt it was important not to burden the police with additional work. Further, the short administrative timeframe that prosecutors have by law to determine whether a restraining order will be granted (i.e. one week) was identified as problematic because it limited their opportunity to obtain desired information from the police.

Although most prosecutors requested additional information, most reported that a history of violent offences that they could identify from a criminal record, was the most important factor when considering whether to grant a restraining order:

It’s not a simple question, we try to see these cases for what they are and you must go in and check the records, but what I can say is that in general, you dismiss the application if the person has no previous violent convictions. (Prosecutor 3)

Although prosecutors noted the weight given to a violent criminal history, there were three other circumstances under which they said an order could be given without such a history. The first is a pattern of harassment, evidenced by previous criminal reports or restraining order applications made by individuals other than the victim that were dismissed or not granted. Second, one prosecutor suggested that it is important to consider case-specific factors (such as severe threats) because there is a first time for every perpetrator and prosecutors must avoid too narrow a perspective. Third, all prosecutors agreed that where no criminal history exists, a restraining order can always be granted in cases where the violence risk assessment indicates high risk. Nevertheless, prosecutors argued that conducting violence risk assessments was difficult. For instance, two prosecutors explained that it is difficult to demonstrate a potential risk for future violence if there is a lack of information about the case. Most prosecutors described feeling insecure about how to conduct a violence risk assessment. The prosecutors described using unstructured professional judgement to reach decisions about the level of risk:

I can always make a decision, but whether the basis for the decision (the violence risk assessment) reflects reality to 90% or 50% or 100% we don’t always know, but we hope that we are as accurate as possible. Although, we can always make a decision. (Prosecutor 3)

Most prosecutors were aware that the police had conducted a violence risk assessment in the case. Most prosecutors stated that the contents of that assessment would probably be of use to them, and some of the prosecutors also reported having used that information. The primary reason that prosecutors gave for not using the police assessment was the communication requirement, which would require the prosecutor to share the risk assessment with the perpetrator. Thus, there is a risk of revealing delicate and sensitive information about the protected person when using the violence risk assessment conducted by the police:

That is a difficult question because you don’t always want to reveal that it (a violence risk assessment) has been conducted, so it’s a question of confidentiality. I would say it’s (conducting a violence risk assessment) not always on the table so to speak, because if that would be brought into the case then there is a potential communication requirement regarding that information, and then you can ruin the investigation. (Prosecutor 3)

A second reason given by one prosecutor for not using the risk assessment completed by the police was that, as prosecutors, they were supposed to conduct their own assessment independent of the police. Third, prosecutors also described having limited access to the risk assessments because police did not always share the information.

Theme 3: interpretation of the law

Prosecutors’ perceptions of their work with IPV cases were described through their ability to work within the law on restraining orders (Swedish law 1988:688). Three of the prosecutors interviewed had worked on IPV cases before the revision of the law in 2011. The only difference that they noted post 2011 was a tightened timeframe, whereby they had to decide on restraining order applications within one week. Although the one-week timeframe is only a guideline, most prosecutors agreed that a decision should be made within that time:

We are very strict about this whole one-week thing, the guideline that there should be an investigation within a week was not really the deal before, so that has been considerably tightened. (Prosecutor 2)

The short administrative time provided to decide about granting a restraining order was described as an obstacle to the decision-making process since it did not always provide adequate time to gather all of the necessary information. One of the prosecutors suggested that it should be possible to expand the timeframe for the decision-making process if the investigation of the crime needs to be prioritized. However, the prosecutor also stressed that in order to expand the one-week timeframe, there should be extenuating circumstances, and that common investigation procedure is not a sufficient reason to expand the administrative time.

Prosecutors provided no opinions on the wording of the law on restraining orders when directly asked. However, several difficulties with the content of the law were raised during the interviews. One difficulty related to the prosecutors’ ability to make use of all four types of restraining orders set out in the legislation. In particular, they noted that the restraining order for shared residence and the two versions of the extended restraining orders were rarely used due to the high requirements for granting such orders set out in the law. In cases where these types of restraining orders were granted, it was usually the case that the perpetrator had already been incarcerated due to the severity of the case. One prosecutor described this barrier when trying to utilize the electronic monitoring function of the especially extended restraining order:

And then there is the electronic supervision, I mean that is never used, and you have an instrument that you never use because you have such high requirements so at the time it can be used, the person is already detained. I can think that it’s a failure of the legislator that you don’t use what you have. (Prosecutor 1)

Despite the fact that none of the prosecutors argued for a change to the law, they emphasized the need for changes to their code of practice. For example, they suggested that it would improve practice if some of the requirements like the timeframe between application and decision-making were changed (The Swedish Prosecution Authority, 2016). Although they described being tightly bound to the law, the prosecutors still experienced a certain degree of flexibility in how the law could be used when a convincing legal argument is put forward:

It might be that you will need to present better arguments for how it (the restraining order) is actually perceived for the victim (pause), I mean how we argue it’s just about finding the right things to point out that the court will find reasonable, you know. (Prosecutor 2)

Prosecutors identified restraining orders as a risk management strategy that does not guarantee victim safety. They viewed restraining orders as simply a decision on a piece of paper where the punishment for violation is low or non-existent. Most prosecutors voiced their unhappiness and frustration at the lack of consequences for violations. Despite these perceived limitations, prosecutors highlighted their obligation to do their best with the tools available to them. For example, one prosecutor described the importance of using the existing coercive measures, such as arrest and prosecution, against restraining order violators. Even where the court disagrees, one prosecutor suggested that at least they could feel as if they had done their job in pursuing the case as far as possible:

Now I know from experience, that it is quite difficult to get someone detained for violating a restraining order. Either there should be awfully many breaches in the case or it almost requires that you are guilty to another crime in connection to the violation of the restraining order in order to be punished for it. (Prosecutor 2)

Prosecutors noted that it is sometimes better to arrest the perpetrator because then the restraining order is not immediately needed because the perpetrator will be detained. This was particularly relevant for the restraining order types of shared residence and especially extended restraining orders given the high burden of proof and risk to the victim required for the granting of such orders. Several of the prosecutors believed that the only way to prevent future violence was to detain the perpetrator. Thus, arresting the perpetrator was described as a more appropriate way to handle a critical situation because if the perpetrator was detained, there was no immediate risk to the victim. However, the prosecutors noted that eventually, the perpetrator would be released, and there would be a need for a restraining order:

Sometimes there is a fine line between whether you should issue a restraining order, or arrest someone. (Prosecutor 2)

Discussion

The aim of this study was to examine how prosecutors made decisions regarding restraining order applications made by victims of IPV. Specifically, we investigated how prosecutors experienced their task of deciding upon restraining orders, and how they used information from violence risk assessments completed by police in their decision-making process. The results reveal several important findings related to prosecutors’ experience working with restraining order in cases of IPV. Three overarching themes were identified: The case, Organisation of resources and Interpretation of the law. Within those themes, prosecutors relayed the dilemmas and compromises that had to be made within their work.

The first theme, The case, included three major findings. The first was the need to balance the criminal investigation and prosecution of the perpetrator with victim protection, as both could not always be achieved simultaneously. The second finding was that prosecutors needed to maintain objectivity as the lead investigators of the cases. This objectivity could be challenged when prosecutors had to investigate a case, while also trying to protect the victim from the accused. For instance, protecting the victim presumes a belief that the perpetrator is dangerous and either has committed or will commit a crime. Third, information sharing can jeopardize the outcome of an investigation if not handled correctly. Overall, prosecutors described that their work with IPV most often requires that they must put aside one need (e.g. victim protection) in favour of another (i.e. obtaining a guilty verdict), and that this required them to prioritize one outcome over the other while still maintaining a holistic perspective of the case.

The second theme identified was Organization of resources, where prosecutors identified police as important contributors to their work. This was especially true within the criminal investigation, where police were often asked for their insights into the case. Prosecutors noted, however, that they were always responsible for the final decision with respect to both the investigation and the restraining order applications. Prosecutors criticized what they thought to be a lack of routines within the police for identifying who was responsible for the restraining order application procedure and who was responsible for the investigation. When considering granting a restraining order application, prosecutors reported that a history of violent offences was the most important factor to consider. Although violence risk assessments might help with such decisions and are routinely completed by police, prosecutors did not use these assessments. Prosecutors stated that they did not want to use the violence risk assessments because the assessments would reduce their objectivity, and the assessments would have to be turned over to the perpetrator as part of the party insight requirement.

The third theme identified was Interpretation of the law, which includes findings related to how the prosecutors interpreted the law and performed their work accordingly. The short administrative time provided to make a decision about granting a restraining order was described as an obstacle, since more time is sometimes needed to gather sufficient information upon which to make a decision. Prosecutors firmly believed that a decision should be made within a week. However, it was argued by some prosecutors in the report by NCCP (2015) that they would like to include leeway to extend the timeframe to gather more information about a case. Although making quick decisions about restraining orders was important and crucial in some cases, ensuring that decisions rely on sufficient information was equally important according to the results of this study. Prosecutors also highlighted that they were unable to make use of all four types of restraining orders outlined in the legislation due to high legislative requirements. Although tightly bound to the law, prosecutors reported experiencing a degree of flexibility in how the law could be used. The prosecutors further highlighted the importance of using existing coercive measures such as arrest and prosecution for the breach of restraining orders.

The primary conflict that prosecutors encountered when working with cases of IPV was the need to remain objective in their investigation while also collaborating with police to protect the victim. The findings in this study show that the prosecutors are unsure of the organization and routines within the police who are responsible for initiating restraining order applications and working with victims. As a result, prosecutors found it difficult to know with whom to share information regarding violence risk assessment and victim protection. Studies examining violence risk assessment within the Swedish police show positive results regarding the use of structured risk assessment tools, including their ability to predict and prevent violence and to tailor risk management strategies to the risks posed in the case (Belfrage & Strand, 2008, 2012; Belfrage et al., 2012; Storey et al., 2014).

Prosecutors’ work on restraining orders would improve if they could find a better way to communicate about risk for violence with the police. The results showed that prosecutors do not use the risk assessments conducted by the police, but instead use their own unstructured judgement. This is problematic because restraining orders are important risk management strategies that are recommended by police for specific victims based on structured risk assessments. If prosecutors continue to disregard the violence risk assessments completed by police and fail to utilize the recommended risk management strategies, police may not be able to adequately protect victims.

Contrary to what would be recommended by past research (Belfrage et al., 2012), our result show that the most important factor considered by prosecutors when granting restraining orders was prior violent criminal convictions, while other risk factors were not taken into consideration to the same extent. Similarly, previous research (see for example Jordan, Pritchard, Wilcox, & Duckett-Pritchard, 2008) has found that the lack of a prior violent act was one of the major reasons for not granting a restraining order. Although prior acts of violence, often measured as convictions for IPV, are important to consider when assessing the risk for future IPV (Kropp et al., 2008, 2010), it is crucial to bear in mind that there is a first time for every offender, and that not every criminal action results in conviction. Thus, the lack of prior convictions for IPV or other violent criminal acts should not lead to an immediate dismissal of a restraining order application. Using a structured violence risk assessment instrument, like the B-SAFER, can assist evaluators in maintaining a more holistic perspective since the instrument guides the evaluator in considering many risk factors related to the perpetrator and the victim. It is also important to recall that prior convictions of IPV are not necessarily sufficient evidence to conclude that there is a high risk for future IPV; each case is unique and risk determinations must consider multiple characteristics of both the perpetrator and the victim.

Our findings suggest that prosecutors perceive restraining orders to be less effective than legislators might have intended. Prosecutors described frustration over the lack of criminal sanctions when restraining orders are violated. Lindström (2005) similarly found in his study that not much was done for female victims by the criminal justice system in Sweden when they reported violations of restraining orders. Connelly and Cavanagh (2007) suggest that the failure to respond to violations may imply that ‘domestic abuse is tacitly condoned and certainly not criminalised in our society’ (p. 283). The authors argue further that perhaps the most significant problem is that victims of IPV will have no protection and will continue to be harassed and exposed to violence. A possible solution to this lack of action by the criminal justice system could be the use of structured violence risk assessment. Using structured violence risk assessments as grounds for decision-making on restraining orders would help to highlight the risk present in the case. Further, the risk assessment in a case could be updated to indicate any increase in risk following the violation of a restraining order, or to highlight the severity of the violation itself. Thus, with the use of structured assessments completed and updated by police and prosecutors, restraining orders might become a more effective management strategy for IPV.

Limitations and areas for future research

Although qualitative research does not necessarily aim to generalize findings beyond the examined sample, and thereby has no specific rule regarding the sample size required for a study, our study had a limited number of participants. This was primarily because there are few prosecutors working on cases of IPV in the area surveyed. Instead of increasing the sample size, we thought it was more important to consider a smaller group of individuals who worked within the same police districts, and with the same police officers, instead of adding another district where the conditions may have been different. Despite the small sample, we did reach some level of saturation. Future research faced with such small sample sizes might consider conducting multiple interviews over a period of time, or adding another police district if possible. An alternative design, such as focus groups, might also provide additional information where available samples are small.

Future research should identify what information best predicts the need and utility of a restraining order in cases of IPV. Identifying such information would ameliorate the efficiency and transparency of the decision-making process benefitting both the victim and the offender. This suggestion is supported by the results of Strand (2012), who found that restraining orders were effective at preventing future IPV amongst male IPV perpetrators assessed as low or moderate risk for recidivism using the B-SAFER, but not for high risk perpetrators, who are usually the ones given a restraining order. Identifying what information is needed could be achieved using a mixed design, employing both qualitative and quantitative methods.

Conclusion

Prosecutors face constant dilemmas when balancing the investigative and protective aspects of their work. They seldom use the violence risk assessments conducted by police as a basis for issuing restraining orders, and there seems to be several reasons for this. The primary reason is concerns regarding sharing sensitive information with the perpetrator, although it is also clear that prosecutors are uncertain regarding the routines of the police and how they should go about cooperating with the police.

To ensure consistent decision-making and wide information sharing between police and prosecutors in Sweden, structured risk assessments should be used as part of the restraining order decision-making process. Due to laws around party insight, which require information sharing with the offender, it has been argued that prosecutors cannot use violence risk assessments (The Swedish Prosecution Authority, 2016). We would suggest that the benefits of using risk assessments outweigh the potential costs. Further, as suggested by The Swedish National Police Agency (2010), summary violence risk assessments could be shared to facilitate risk communication in order to ensure victim safety and to safeguard the criminal investigation.

Funding

This research project was funded by the Swedish Crime Victim Compensation and Support Authority (Brottsoffermyndigheten).

Acknowledgements

The authors would like to thank the district attorneys in the counties of Västernorrland and Jämtland for their valuable help in this research.