Enacting Entangled Practice: Interagency Collaboration in Domestic and Family Violence Work

Introduction

Globally, it is estimated that one in three women have experienced physical and/or sexual violence at the hands of a male partner (World Health Organization, 2013). In some parts of the world, the estimates are much higher. In Australia, one woman a week is killed by a violent partner or ex-partner, with domestic homicides accounting for almost half of all murders, many of these being murders of children (Australian Institute of Criminology, 2015). These are shocking statistics, but perhaps more shocking is the fact that these figures have not improved significantly over the past two decades. The complexity and apparent intractability of domestic and family violence (DFV) have thus led to its framing as one of an increasing number of 21st-century “wicked” problems (see Healey, Humphreys, & Wilcox, 2013; Kearns & Coen, 2014; Stanley & Humphreys, 2014). Interagency working (variously conceptualized as coordination, integration, and, more generally, collaboration) has become a preeminent response to this problem.

However, while interagency collaboration has “attained policy status as good practice” (Murray & Powell, 2011, p. 128) in DFV work, insufficient attention has been paid to how pro-collaboration policy plays out in practice, and with what effects for victims/survivors. While there is evidence of positive outcomes of some types interagency work for victims/survivors (see Banks, Dutch, & Wang, 2008; Healey, Frere, Ross, & Humphreys, 2009; Hester & Westmarland, 2005), research also suggests that well-intentioned interagency efforts do not always result in good outcomes for victims/survivors (see Postmus & Ah Hahn, 2007; Valentine & Hilferty, 2011; Visher, Harrell, Newmark, & Yahner, 2008). These mixed and inconclusive results demonstrate that interagency working “is not inherently a good thing” (Statham & Department of Children and Youth Affairs, 2011, p. 4). The study on which this article is based examines this apparent contradiction, seeking to answer the following research questions:

This research explores how it is that interagency processes and tools can sometimes constrain professional judgment in complex casework (Hood, 2014) and result in less than optimal outcomes for victims/survivors. Adopting Hood’s (2014) suggestion to take “a closer look at complexity as a phenomenon in its own right, rather than a convenient metaphor of difficulty” (Hood, 2014, p. 28), the study problematizes interagency DFV practice.

First, I explain the conceptual framework and methodology, clarifying key analytical concepts and providing details of the overall study design. Then, I discuss the analytic process and the findings that emerged, illustrating this through examination of one of several interagency practices investigated in the research. The article concludes with a brief discussion of implications for policy and practice in interagency DFV working.

What a Practice-Based Approach Provides

The value of practice-based studies, Gherardi (2012) argues, is that they foreground connections and relationships, providing a theoretical lens through which to view practice as more than merely the procedural application of knowledge. In Gherardi’s (2008) terms, practice is an “in-between concept akin to ‘knowing while doing’, between ‘habit’ and ‘action’, between ‘production’ and ‘reproduction’” (Gherardi, 2008, p. 523), a collective activity that is developed by doing and relating, as well as by thinking. Moreover, Gherardi argues that using a practice-based approach to research illuminates how new practices may emerge at the boundaries between professions and organizations and how “new knowledge may be produced and sustained (or hampered) by different forms of organising” (Gherardi, 2009, p. 355). Such an approach to researching professional practice is particularly applicable to a study of interagency DFV working, because it accounts for, and makes visible, the many occasions when decisions are made by practitioners based on their collective and diverse practice wisdom and intimate knowledge of their local context.

What a Sociomaterial Perspective Highlights

Weaving through the “practice turn” (Schatzki, Knorr-Cetina, & Von Savigny, 2001) in social theory, discussed above, there has been a corresponding “sociomaterial turn” in studies of professional practice (Fenwick et al., 2011). Much of the earlier research on professional practice, including interprofessional and interagency practice, paid relatively little attention to materiality or to how the personal/social/material dimensions of practice relate to each other. Indeed, in many accounts while participation and communication are foregrounded, “artefacts and other material aspects of practice remain theoretically in the background” (Zukas & Kilminster, 2012, p. 203).

While the theoretical origins and historical roots of sociomaterial approaches vary, Fenwick et al. (2011) suggest that there are some broadly common themes and conceptual resources. First, they enable a relational account of practice, focusing on the connections among entities through which actions happen. Second, they decenter the individual human as the only source of agency. Third, they engage productively with diversity and multiplicity, recognizing complex systems, overlapping activities, and ambiguous practices. Fourth, they disrupt notions of context as fixed, understanding it as always moving, partial, and constitutive of action. These themes are highly relevant to a study of interagency collaboration in the networked, technology-mediated, and fluid world of 21st-century workplaces.

Moreover, attending to the sociomaterial invites consideration of the materially mediated and embodied dimension of practice, affirming that it is both inherently material and social (Orlikowski & Scott, 2010). Viewing practice through a sociomaterial prism thus enables a view of how material arrangements become entwined in and form part of everyday work (Fenwick, Nerland, & Jensen, 2012; Nicolini, Mengis, & Swan, 2012). Noting how practice involves non-animate actors, not just as background or contextual features, Zukas and Kilminster (2012) make the assertion that material artefacts are in fact “centrally implicated in practice” (Zukas & Kilminster, 2012, p. 200).

What ANT Offers

As noted by Fenwick et al. (2011), several theories have been advanced within the sociomaterial tradition. One of these is ANT. Gherardi (2000) identifies ANT as one of the strands of research that she calls “practice-based theorising” (Gherardi, 2000, p. 212) because it contributes to the ongoing conversation about the continuity between knowing and doing. It does this by illustrating “how knowing is enacted in sociomaterial networks of human and non-human actants” (Gherardi, 2012, p. 200).

Emerging from post-structuralism in the early work of Bruno Latour (1986), Michel Callon (1986), and John Law (1992), ANT views relationships between things, concepts, and people as potentially transient and in flux, in a constant process of making and remaking themselves and needing to be repeatedly “performed” to prevent dissolution. In the constantly changing and highly contested environment of interagency DFV work, ANT provides what Nicolini (2011) calls a “powerful theory/method package” (p. 1394) with which to unravel and analyze practice.

Key Concepts

The following key ANT concepts are central to the analysis and discussion that follows:

Study Design

The study adopted a field method described as praxiographic (Mol, 2002), that is, it was an ethnographic study of practice. As Mol (2002) notes, broadening the focus of investigation beyond the “ethno,” or human, participants leaves open “who or what the actor is” (p. 143). This draws attention to the more-than-human aspects of working together, consistent with the sociomaterial perspective outlined above.

The fieldwork took place over 6 months in an outer metropolitan suburb of Sydney, Australia, where a local integrated DFV response had been operating for many years. The main data generation methods were focused workplace observations (including shadowing practitioners) and semi-structured interviews. I also examined key documents and artefacts with a view to understanding them as nonhuman actors in the participating networks. These nonhuman actors included victim support and referral cards, police intake forms, risk factor checklists, court orders, case files, case plans, spreadsheets, electronic databases, interagency protocols, multiagency agreements, and memoranda of understanding.

Twenty-one field visits, each of half to a full day, generated 186 pages of field notes and a further 135 pages of transcription of 20 interviews and 17 meetings, in which 21 frontline practitioners and 15 operational managers, from a broad range of local services, participated. Agencies represented included large government bodies (police, child protection, corrective services, and health), as well as small community-based and nongovernment organizations (a DFV specialist casework service, a women’s refuge, a relationship counselling organization, and a women’s and children’s advocacy and support service).

Integrated Case Coordination

In the context of this study, the practice of integrated case coordination entailed local service providers meeting together monthly, as per their locally agreed interagency protocol. In this meeting, they discussed and took responsibility for managing different aspects of case plans of local high-risk DFV offenders.

The integrated case coordination meetings took place at the local police station and lasted 2 hr. Representatives from the key local government and nongovernment service providers that were signatories to the interagency protocol (police, corrective services, community services, family violence service, and women’s support service) attended. Most of the agency representatives were operational frontline managers. Child protection caseworkers and probation and parole officers who were working directly with offenders and/or victims presented their reports and provided updates on the cases under consideration. Usually, there were six or seven active cases on the list for discussion at each meeting.

A case was determined to be eligible if it attracted a score of more than 10 on a risk assessment tool jointly developed by the signatories to the protocol. The tool listed weighted evidence-based risk factors. A score of more than 20 signaled high priority; a score of between 10 and 20 was for “secondary consideration”; and a score of below 10 was not serious enough for integrated case coordination. The risk factor checklist was thus an actor that ordered practice and set in train a series of events that nudged various other actors into alignment in the assessment and management of risk. However, regardless of the score on the checklist, the police guidelines stipulated that if there had been three domestic violence (DV) incidents recorded in the previous 6 months, the system automatically flagged the case as “high risk.” This is indicative of the relative strength of the police service network in the integrated case coordination meetings.

Each case on the active list had a case plan, which outlined the various actions that the different agencies had agreed to take to improve victim safety and ensure offender accountability. In this way, the case plan functioned as a central nexus of practice, where various strands of discourse, policy, and protocol intersected. Each case had a lead agency with a designated representative initiating discussion and taking responsibility for maintaining the case plan. Group members were expected to bring to the meetings relevant information about the cases on the active list, a process made possible by the information-sharing provisions in Part 16 A of the NSW child protection legislation (Children and Young Persons [Care and Protection] Act NSW, 1998).

Cases remained on the active list until each agency’s goals (referred to by the group as “end outcomes”) had been met (e.g., the mother had engaged with a support service or early intervention program) and/or there were no longer “significant safety concerns” for the victim (e.g., the perpetrator had been incarcerated). Some cases were “suspended” for 6 months (rather than “closed”) and then reviewed if the agencies agreed that this was warranted (e.g., the perpetrator was due for release from jail and there were concerns that the violence would recommence). A master spreadsheet (developed and maintained by the family violence service) did the work of keeping track of the progress of each case, functioning as a ledger that quantified the cases that were “active,” “suspended,” and “closed,” over a specified period.

Following are two data excerpts, taken from the transcript of one meeting, that illustrate how differently the practice of integrated case coordination unfolded, with varying effects for the victims/survivors, depending on which agency took the lead and which actor-networks were activated. I present the vignettes as typical examples of the many cases discussed at the integrated case coordination meetings. The first excerpt, “Shan and Gavin,” shows integrated case coordination enacted in one way. The next excerpt, “Chloe and Jamie,” shows the same interagency practice enacted in a different way. Names have been changed to protect identities.

“Shan and Gavin”

In this excerpt, the members of the integrated case coordination group were discussing the case of Shan and Gavin. The Department of Corrective Services was the lead agency, and the Probation and Parole Officer who was working directly with Gavin was attending the meeting to present his report:

In this case vignette, the strength of the managers’ and frontline practitioners’ enrollments in their own home agency networks was evident, in their reporting on the outcomes of their work with victims and offenders. Maintaining a focus on managing the behavior of the offender, Gavin, enabled Police and Corrective Services to emerge as primary actors here, enrolling the other group members into their own networks and convincing them (at least for now) to take on the criminal justice system (CJS) framing of DFV as primarily a crime. Similarly, the Department of Community Services succeeded in enrolling others (including actors that were not signatories to the protocol), demonstrating the strength of its network in responding to DFV as principally a child protection issue. These enrollments enabled the statutory agencies to achieve the tasks as outlined on their case plans, thereby strengthening their own individual agency networks.

The enrollment of legislated information-sharing provisions enabled individual agencies to access information to which they would not otherwise have been privy. In this case, negotiating access to the offender’s mental health assessment report was important for the Department of Corrective Services to achieve the “end outcome” on their case plan. However, this could not happen before renewal of the “exchange of information request.” The exchange of information request thus functioned as a gateway through which the Department of Corrective Services was required to pass before this access was granted. However, having already enrolled alcohol and drug experts into their network, the Department of Corrective Services could undertake urinalysis testing to support them in completing their designated task of monitoring Gavin’s marijuana use. In this way, the case plan effectively translated “high-risk management” into monitoring the offender’s drug habits. The point is that this was a task that mattered to the Department of Corrective Services manager at least as much as keeping the victim safe.

Raising the issue of Shan’s pregnancy, and the fact that no caseworker had been allocated because no “risk of harm reports had been made” to the child protection agency “for quite a while,” the manager of the feminist-based family violence service refocused discussion on the victim/survivor. In ANT terms, the Department of Community Services had become a primary actor through the repeated creation of risk of harm reports, strengthening and reifying the child protection system’s network. In this situation, no risk of harm report meant no allocated caseworker for the victim/mother, no intervention deemed necessary. Therefore, the case plan allocated no tasks to the Department of Community Services in this case. However, this had the effect of creating a “double bind” for Shan because evidence suggests that pregnancy is a high-risk time for women in abusive relationships (Jasinski, 2004). Shan’s pregnancy had thus translated her (the primary victim) into a potential risk to her unborn child, but no early intervention support was forthcoming because there were no new reports of children at risk (yet).

The convergence of the Apprehended Violence Order expiry date and the birth of the victim’s baby would soon combine to ramp up the level of risk to both Shan and her children, including her unborn child (Coker, Sanderson, & Dong, 2004). The matters of the Apprehended Violence Order expiry and the birth of the baby thus intersected with the child protection discourse to constitute a “high-risk case,” but as nothing serious (enough) had happened (yet), no further action was to be taken. Gavin was still considered a “high-risk” offender and Shan was left unsupported–and the integrated case coordination group had six more cases to get through before 3 p.m.

“Chloe and Jamie”

In the following excerpt, the integrated case coordination group was discussing Chloe and Jamie. The family violence service had been working intensively with Chloe for some months and was the lead agency in this case:

In contrast to the previous vignette, the group’s enactment of integrated case coordination in relation to Chloe and Jamie involved a series of enrollments that generated more positive effects for the victim/survivor. Through this way of working together, the interagency network enrolled the police and corrective services representatives, and the police network enrolled the victim/survivor, Chloe. This reciprocal process resulting from multilateral negotiations enabled the system to be “tweaked” to become more responsive to Chloe’s needs.

The omission in the police narrative meant that no one had arranged to interview the children immediately following Jamie’s assault of Chloe. When the police officer undertook to “get someone onto it tomorrow,” he was challenged by the family violence service manager as to the appropriateness of this action at that time. She made the point that the children were already traumatized and that this police intervention, so long after the incident, would be likely to retraumatize them. In this way, the family violence service manager was able to challenge the authority of the police, drawing on her role as trauma expert to shape the actions to be taken, thereby strengthening the interagency network in which her service was a focal node.

Referring to the children’s mother as “very sort of. . . erratic,” the police officer invited another instructive response from the manager of the family violence service. The picture painted of Chloe as “your typical worn-down, beaten domestic violence victim [with] her own history of childhood abuse” whose mental health was poor and whose life was “a mess” underscored the long-lasting and far-reaching impacts of trauma. Explicitly making the links between Chloe’s own abusive childhood and the current abusive situation that she and her children were experiencing, the manager of the family violence service pulled the threads of the victim’s past through into the present in a way that could inform the future. The arrest warrant for Jamie was a crucial actor in this scenario, effecting the intended outcome in this case. The fact that the warrant was not related to DFV charges or to breach of an Apprehended Violence Order was less relevant than the fact that it would provide the police with a legitimate reason to take the offender into custody. Thus, the manager of the family violence service had succeeded in enrolling both the Police and Corrective Services into the interagency network by aligning their interests. This convergence had the effect of creating a single focus for action: victim protection and support.

Taking a trauma-informed approach enabled the family violence service manager to emerge as a primary actor here, having successfully enrolled the statutory authorities into accepting interrelated roles with the victim, thereby taking on (at least for now) the family violence service understanding of DFV as a trauma-related issue, rather than simply a CJS concern. This enabled them to see beyond their own agency priorities and engendered a better understanding of the impacts of offender behavior on victims, resulting in a more coordinated and nuanced interagency response to integrated case coordination.

The suggestion by the family violence service manager of there being “maybe a third way” for the police to execute the warrant introduced the possibility of being able to do their job in a way that was less prescriptive than usual, less “by the book.” However, it involved considerable negotiation, in more than one direction, with several parties. In ANT terms, this technique was successful because, by the end of the conversation, she had enrolled the police into the interagency network, thereby strengthening it. She did this by suggesting that enrolling the victim into the police network would help the police to “get their man.” By persuading the police officers that it was possible to enlist the help of the victim in executing the warrant on the offender, the manager of the family violence service also strengthened her position in the integrated case coordination network; she was the lynchpin here, occupying the node where multiple networks intersected. She would negotiate with the victim, if the police agreed to negotiate with her. The transaction involved a trade-off of sorts, there being no deal without strings attached. These “strings” effectively enrolled Chloe into the police network, translating her from victim into police informant. The possibility of a “win–win” situation thus aligned multiple interests, as the actors negotiated a different way of working together, a way that acknowledged and incorporated the victim/survivor as a key actor.

Interagency Collaboration Through Multiple Connecting Actor-Networks

Analyzed using the core ANT concept of translation, interagency practice was shown to be enacted via processes of network formation, maintenance, and collapse. From this analysis, two types of networks emerged: single agency networks and interagency networks. Importantly, the same actors participated in both types of network, illustrating both the entanglement of the networks and the mutual constitution of actors and networks.

While most of the single agency networks participated in each of the practices examined in the study, what also emerged strongly was that each interagency DFV practice generated its own interagency network, which then sustained the practice. Importantly, however, without the single agency networks there could have been no interagency networks, as all new networks grow from others (Latour, 2005), either nesting within them (as in the police, corrective services, and court networks within the CJS) or existing within them and extending beyond. Significantly, the interagency networks reached beyond the existing networks, connecting them in new, often challenging, ways.

Underscoring the challenges of visual representation for ANT theorists, Latour (2011) remarked on ANT’s “impoverished visual vocabulary” (p. 800). Noting that while the concept of the network is useful to emphasize the (re)distribution of action, networks are “extremely poor metaphors, since they remain entirely made of nodes and edges to which are added some conveniently drawn potato-like circles” (Latour, 2011, p. 800). Nevertheless, when synthesizing and looking across the interagency practices examined in this study, forms of visual representation were developed, both for illustrative purposes and to notice and probe the aggregate data in new ways. This was crucial in the interpretive work relating to the relative stability of single agency versus interagency networks. Notwithstanding Latour’s (2011) caveats, Figure 1 offers a useful visualization of the multiple intersecting and entangled networks that participated in the enactment of interagency DFV work in the research site.

OPEN IN VIEWER

Figure 1.

Actor-networks in interagency DFV work.

While all networks are inherently open to instability, the analysis suggests that the networks of the single agencies were significantly more stable than the interagency networks that emerged. The networks nested within the CJS, and those of the statutory authorities particularly, have been shown to be especially durable because of the many successful translations that have occurred over time in the processes of their formation. This resonates with Harvie and Manzi’s (2011) argument that the criminal justice discourse has become one of the dominant discourses in interagency DFV work, leading to the centrality of the statutory sector and the marginalization of nongovernment women’s organizations (Hague & Malos, 1998).

A key actor to emerge as a facilitator of interagency network formation in this study was a coordinating service that specialized in DFV work. In three of the four interagency practices examined in this study, including integrated case coordination, the family violence service (FVS) played this coordinating role. The need for dedicated specialist services such as the FVS, to give a voice to victims/survivors, is a recurrent theme in the literature on interagency collaboration in DFV work (see Hague & Mullender, 2006; Murphy & Fanslow, 2012).

Two Modes of Working Together

The study found that the enactment of interagency DFV work involved oscillation between practitioners working together, while remaining to some extent apart, separate from each other and working together in a more engaged, integrated way. This movement between two modes of working together was evident across all the practices examined, including integrated case coordination, the practice under discussion in this article. The first mode involved multiple service providers working together but operating with their own multiple foci for action. I call this mode of working together “multiple practices” (plural) because it emerged from and maintained multiple individual/separate agency networks (e.g., police, court, corrective services, child protection, and women’s services).

The second mode also involved multiple service providers, but working this way, practitioners maintained a singular focus for action, the victims/survivors’ interests. This mode entailed a tighter, more entangled working-togetherness, which emerged from and maintained a singular interagency network. Borrowing from Mol’s (2002) conceptualization of the multiple phenomenon, I call this second mode practice multiple, to capture the notion of multiplicity enacted in practice with a shared single focus. In making this distinction, I am not proposing a dichotomy. Rather, I am suggesting that practice multiple and multiple practices coexist, not as alternatives or opposites, but as entangled modes of practice that together make up the textured fabric of interagency working.

The differences between the two modes are clarified below, with specific reference to the case vignettes presented earlier.

Multiple Practices

In the first vignette (Shan and Gavin), practitioners from different agencies worked together, but as if in parallel, not always connecting, even though they were in the same physical space. Interagency procedures unfolded strictly as per the participating agencies’ own priorities, and their “territory” was staunchly defended. Enacting interagency work this way, practitioners were generally acting from “spaces of prescription” (Fenwick, 2010, p. 126) with a rigid task orientation, adhering strictly to their “home” agency mandates and staying firmly within the boundaries of their professional roles.

The enactment of integrated case coordination as multiple practices thus effected a somewhat managerialist approach to the complexity (Harvie & Manzi, 2011) of interagency DFV work here. Task-oriented accountability resulted in a narrowing of focus to each of the individual statutory agencies’ own primary matters of concern. A consequence of this was that Shan did not receive the support or protection that she needed, when she needed it. Moreover, the domination of criminal justice concerns had the effect of focusing action on apprehending Gavin, at the expense of attending to Shan’s support needs.

The effect of this was that Shan became less and less visible, effectively being filtered out of the support system, relegated (at least for now) to continuing to live a life impoverished by fear and anxiety. Moreover, when enacted in multiple practices mode, integrated case coordination tended to privilege the child protection system, with abused mothers, like Shan, being frequently assessed as inadequate parents. Holding mothers responsible for child protection (see Beeman, Hagemeister, & Edleson, 1999; Hester, 2008, 2009; Thiara & Humphreys, 2017) while ignoring the impacts on fathering by abusive men not only fails to differentiate between the parents, but also frequently results in blaming the mother for the father’s abusive behavior.

Across all the interagency practices examined in this study, when multiple practices dominated, victims/survivors tended to be categorized based on individual agency eligibility criteria. Interagency success was viewed largely in terms of how well the system managed the increasing volume of “cases” and how quickly and efficiently these could be moved through the system. Productive orderliness was emphasized as the participating agencies fulfilled their mandatory responsibilities, as prescribed by their own agency networks, thereby demonstrating their accountability to each other. In this process, there was a tendency for management of offenders’ criminal behavior and compliance with child protection policy and legislation to take precedence over attention to the needs and safety of the adult victim/survivor (usually the mother). This tendency, noted frequently in the literature about mandatory reporting of children living in DV situations (see Humphreys, 2008), highlights another quality of multiple practices mode: the concentration and accumulation of power at a small number of network nodes. In this study, these nodes were frequently where the networks of the statutory authorities of the police, corrective services, and child protection intersected within the CJS.

Practice Multiple

In contrast, in the second vignette (Chloe and Jamie), integrated case coordination was enacted in practice multiple mode, characterized by a greater responsiveness to victims/survivors. This was achieved partly by the group members being open to doing business not-quite-as-usual and working together as diverse practitioners with a shared priority, as actors in an interagency network. In this mode, the practitioners were encouraged to widen their focus to take in the broader circumstances surrounding Chloe’s story. They were more inclined to identify workarounds together, acting from “spaces of negotiation” (Fenwick, 2010, p. 126), so that their interagency business focused on how best to ensure the system met the needs of Chloe and her children.

As the group members learned that it was possible to work flexibly and still operate within their agency guidelines, they could (re)center Chloe and her needs, integrating her presence into practice, despite her material absence. Taking trauma-informed action enhanced their practice in a way that opened the possibility of questioning the impacts of the system on victims/survivors and taking steps to remediate negative impacts. This enabled the group to include in their case planning what might have appeared to be peripheral to the main game of managing offender behavior. Engaging in practice multiple, they were reminded that there was always and already a bigger picture and a longer story than ever appeared on a case plan. Keeping the impacts of Jamie’s violence always in view had the effect of holding him responsible for his behavior and at the same time expanded the focus to encompass the system’s accountability to Chloe and her children (Hague, Mullender, & Aris, 2003). Enactment of integrated case coordination as practice multiple thus effected a rebalancing of the focus for action, ensuring that the practice remained principally about prioritizing victim safety and well-being.

Across all the interagency practices observed, practice multiple became the mode of working when practitioners connected in a way that held them in alignment with a shared priority, as actors in an interagency network. When this happened, practitioners’ role boundaries were stretched and rearranged in the collective effort to improve outcomes for victims/survivors. Enacted as practice multiple, interagency work challenged the supremacy of the CJS from time to time, the power of the statutory authorities being destabilized occasionally, in the pursuit of victim/survivor advocacy. Reorienting discussion back to the victim/survivor, to focus on her risks and needs enabled the practitioners to hold in view the impacts of the offender’s behavior and reduce the possibility of retraumatizing the victim/survivor. Importantly, it also made it possible for them to uphold the imperative to ensure children’s welfare while supporting the adult victim/survivor/mother, performing what Fleck-Henderson (2000) refers to as “seeing double.” In this way, accountability to victims/survivors was foregrounded as a matter of importance in interagency work.

Interagency tensions were confronted, and held open, rather than shut down or avoided, in practice multiple mode, enabling the group to move toward shared responsibility and joint problem solving (Murphy & Fanslow, 2012). The success of their work together was judged in terms of evidence of improved outcomes for victims/survivors, rather than merely discharging their own responsibilities.

The Textured Möbius of Integrated Interagency DFV Practice

As discussed, both multiple practices and practice multiple coexisted in interagency DFV work in this study, producing service provision that emerged from “a texture of connections” (Gherardi & Perrotta, 2014, p. 147). However, this texture varied depending on which networks were enacting a practice at any given time. The texture produced by multiple practices was often smooth, unfolding as per prescribed procedures, but in terms of interagency work, it was loosely woven, frequently with large holes, through which victims/survivors could slip. This was because multiple practices were not necessarily enacted in connection, but often in parallel with each other, failing to intersect appropriately.

However, the texture produced by practice multiple was more tightly woven. This texture was created by service providers working together in a way that their practice intersected, often surfacing tensions and differences. The resulting texture, while knotty in places, was taut enough to pull the threads of multiple practices together (even if only temporarily) into a web of safety for victims/survivors. It was precisely the entwined nature of multiple practices and practice multiple that gave the interagency DFV work in this study this complex texture.

Taking up Latour’s (2005) point that networks are poor metaphors for visual depiction, the single-/double-sided, three-dimensional, nonorientable Möbius band may offer a richer metaphor, one that combines and rounds out the “nodes . . . edges . . . and potato-like circles” (Latour, 2005, p. 800) that typify diagrams of networks. To elaborate on this analogy, I suggest that integrated interagency DFV service provision has a texture that has two-sides-in-one, because practice multiple and multiple practices coexist, not as alternatives or opposites of each other, but as entangled modes of working together. It is three-dimensional because what results is greater depth in the delivery of more fully developed holistic service provision. It is nonorientable because movement or action in any direction, starting from any place, will eventually end up at the starting point, in this case, the victim/survivor.

Foregrounding Complexity, Change, and Negotiation

First, policy needs to reflect the complexity of interagency DFV work to support the development of responsive practice (Hood, 2014), rather than aim for simplicity through standardization of responses. Policy making would therefore proceed from an understanding that nothing exists in stable isolation and connections are always in a state of flux and have material effects, even if these are not immediately visible. The implication of this is that, in the development of integrated approaches to DFV, the effects of introducing new elements, or of making changes to components of systems and frameworks, would need consideration in terms of both outcomes and process.

Systemic reform processes would proceed as negotiated “settlements” (Latour, 2005) among the multiple agencies that have a part to play in responding to DFV. For example, it is important that different agencies’ policies intersect in productive ways, and that multi-agency initiatives are genuinely multi-agency and not dominated by the concerns of a handful of stakeholders (Reeves, 2013). In the case of Gavin and Shan, the expiration of the victim protection order coinciding with the imminent birth of the baby increased Shan’s level of risk. However, no action was suggested to address this, despite the FVS manager raising it as a matter of serious concern. Integrated governance structures, therefore, need to establish that the priorities of nongovernment organizations have equal standing to those of their government partners (Radermacher, Karunarathna, Grace, & Feldman, 2011).

Monitoring, Modification, and Maintenance

Second, we would expect that the implementation of integrated approaches would rarely happen in practice as planned or intended. The translations that inevitably occur are not always predictable, and so systems need to be agile and flexible enough to respond to such unanticipated consequences. This throws into relief the need for constant maintenance and modification of systems and procedures to adjust to rapidly changing environments. A sociomaterial perspective on the enmeshment of policy and practice eschews a “set-and-forget” approach. For example, in the case of Chloe and Jamie, the discussion of the execution of the arrest warrant served to reshape police practice to be more responsive to victim interests in this instance.

Methodologies for the evaluation of coordinated and integrated responses may also need reconsidering, tracing instead the effects of various associations, and “joining the dots” of these effects, rather than trying to measure the effectiveness of individual components of particular models. This would help to address the need for evaluation approaches that examine how interactional, organizational, and systemic determinants of collaboration intersect (San, Martin-Rodriquez, Beaulieu, D’Amour, & Ferrada-Videla, 2005), sometimes producing unforeseen consequences.

Rethinking the Design of Practitioner Tools

Third, nonhuman participants in interagency work would be taken seriously, not as mere tools, passive things in the background, but as active players in the enactment of practice (Zukas & Kilminster, 2012), of equal standing with practitioners and in their entanglement with “people factors” (Atkinson & High and Complex Needs Unit, 2007, p. 145). The implication of this is that the design of practitioner resources such as referral protocols and risk assessment forms would take account of how practitioners engage with these tools, how technology might “push back,” and what impact this might have. Gillingham’s (2011) caution about decision-making tools unintentionally reducing complex practices to box-ticking activities serves as a salutary reminder in this context. This is certainly not to suggest that technology may not be enrolled as an ally in interagency data sharing and communication, but rather to sound a note of caution about possible unintended consequences. For example, current efforts to simplify and streamline interagency DFV work through the introduction of common tools and processes, while intended to promote shared understandings, sometimes only succeed in reducing the complexity of practice on paper, while inhibiting the exercise of professional judgment, as was illustrated by the handling of the Gavin and Shan case.

Strengthening Interagency Networks

Fourth, the relative fragility of interagency networks needs to be recognized. In light of this study’s findings, it may be inferred that the emergence of practice multiple, and the corresponding stability of interagency networks, is facilitated by specialist victim-focused and trauma-informed DFV services assuming coordinating roles (Bagshaw, Chung, Couch, Lilburn, & Wadham, 2000; Hague & Bridge, 2008). An important implication for policy then is that expecting integrated responses to deliver optimal outcomes for victims/survivors in the absence of feminist-informed services is unrealistic. As the case of Chloe and Jamie illustrates, specialist DFV services help to stabilize interagency networks, fostering the development of practice multiple, which is vital to achieving the purpose of anchoring interagency practice to both adult and child victims/survivors (Humphreys et al., 2001). Recognizing the distinction between practice multiple and multiple practices and proceeding from an understanding of interagency DFV collaboration as enacting these entangled modes of working, the non-negotiable bottom-line policy and practice maxim would be to maintain a focus on victim impact in all interventions.