Abstract
A small body of research has started to explore the factors and processes that contribute to desistance from intimate partner violence (IPV). However, this evidence has been undertaken in the absence of consistent definitions of the construct that we are trying to measure and understand. We analysed the narratives of 15 female victim–survivors of IPV who reported a significant reduction or cessation of abuse for at least 6 months while still in the relationship. Our analysis revealed that complete cessation was rare; more commonly, physical abuse ceased while other forms of abuse either continued, escalated or emerged. The study also identifies the phenomena of false desistance, in which the cessation of abuse is attributed to the victim–survivor's actions rather than the offender's genuine change, and substitutive desistance, in which one form of abuse is replaced by another. The article concludes by highlighting key questions for advancing IPV desistance research.
A small but growing body of research is emerging in relation to intimate partner violence (IPV) desistance; that is, the processes and factors that contribute to the reduction or cessation of abusive behaviours within relationships (Giordano et al., 2015; Halpern-Meekin and Turney, 2021; Walker, 2017). Research in this area has highlighted the importance of social bonds, mutual recognition and identity transitions in sustaining change. However, even within this growing relational literature, the perspectives of those most directly harmed – victim–survivors of crime – are rarely centred. In the context of IPV, where abuse is embedded in intimate, ongoing relationships, overlooking the perspectives of victim–survivors is not just an empirical gap; it is a theoretical one. Victim–survivors are often the first to notice whether change is occurring, feels genuine and makes them safer. They are also uniquely positioned to assess whether reductions in violence represent meaningful and legitimate change or simply reflect strategic adaptations on the part of the perpetrator.
Victim–survivors may also play a role in enabling or sustaining change, often through strategies that remain invisible to professionals and researchers. When these strategies come at a personal cost – including undermining well-being and autonomy – victim–survivors may be experiencing what Feld and Straus (1989) refer to as false desistance. This highlights the need to understand desistance in context – attuned to the relational dynamics, through which change is produced or maintained, the role of others in these processes, and the emotional, social and physical costs that may accompany this.
Clarifying how desistance is defined is a necessary starting point for research in this space. Although researchers have traditionally shaped these definitions, others – particularly professionals like probation officers – have mostly been tasked with identifying, rather than defining, desistance. More recently, a small body of research has begun to treat these professionals as research participants worthy of examination in their own right, exploring their perceptions of their own roles in desistance processes (see, for example, Arseneault et al., 2023; Ugelvik, 2022).
However, one group remains largely excluded from the conceptual development of desistance: victim–survivors of crime. In IPV research, victim–survivors are sometimes asked to corroborate whether a reduction or cessation in abuse has occurred, often as a means of validating offender self-reports (Walker, 2017). Yet, they are rarely invited into the more foundational project of defining what desistance is, what it looks like and what it requires. This is a critical distinction. To confirm the presence of change is not the same as shaping the meaning of change. For victim–survivors of IPV, desistance is not an abstract academic concept – they have (sometimes literal) skin in the game, when it comes to assessing whether existing definitions reflect real and meaningful change.
There are important precedents for collaborative definitional work between IPV victim–survivors and researchers. Goodman et al. (2015, 2016), for example, partnered with victim–survivors and practitioners to define constructs such as safety and survivor agency, leading to the development of new tools for evaluating domestic and family violence (DFV) services. Their work demonstrates the value of centring lived experience in developing key conceptual and measurement frameworks. A similar approach can be applied to desistance: reframing it not as something located solely in the actions or mindset of the offender, but as a relational and experiential process shaped by how change is interpreted, resisted or enabled by others, particularly those who have been harmed.
This study contributes to a small but growing body of work that takes a relational approach to desistance. Drawing on interviews with 15 women in Australia who reported a cessation or reduction in IPV, it explores how victim–survivors define and make sense of desistance, and what their insights reveal about the assumptions embedded in dominant frameworks.
Defining desistance: A review of the literature
Traditionally conceptualised as the complete cessation of offending, since the early 2000s desistance has been understood as a dynamic process, involving fluctuations between relapse and remission (Gadd and Farrall, 2004; Maruna, 2001; Walker et al., 2013). In this way, offenders may never completely stop, but de-escalate or reduce their involvement in crime over time (Shapland, 2022).
Although desistance has primarily been measured through behavioural cessation, many scholars argue that cognitive transformation is equally important (Giordano et al., 2015; Maruna, 2001; Paternoster and Bushway, 2009). Maruna and Farrall (2004) proposed a dual taxonomy: primary desistance, referring to observable reductions in offending, and secondary desistance, involving a shift in self-identity in which individuals no longer see criminality as part of who they are.
Building upon this framework, Graham and McNeill (2017) introduced tertiary desistance, which extends beyond individual transformation to encompass social (re)integration. This dimension highlights the extent to which (ex)offenders establish a sense of belonging within a moral or political community, with an emphasis on overcoming the stigma associated with prior offending (Graham and McNeill, 2017; Weaver, 2013).
Despite these conceptual advancements, desistance research continues to rely predominantly on behavioural measures, with definitions incorporating attitudinal change or social reintegration remaining less prevalent (Rocque, 2021). Although this ‘gap’ between conceptual and operational definitions of desistance is often attributed to the methodological challenges of capturing subjective transformation, the prioritisation of behavioural indicators is also ontological in nature. Scholars note that while many individuals express a desire to change, sustained behaviour modification is the critical marker of desistance. As Shapland (2022: 273) contends, ‘[t]here clearly needs to be a reduction in, or cessation of, offending behaviour, not simply a wish to desist’.
We agree that behavioural change should remain a key indicator of desistance. However, the unique dyadic context within which IPV occurs necessitates close interrogation of whether the non-occurrence of abusive behaviours is enough to classify someone as an ‘ex-abuser’. In other words, reduction or cessation of IPV should be considered a necessary, but not sufficient, condition in this context. In one of the first articles written about IPV desistance 35 years ago, Feld and Straus (1989) cautioned both desistance and IPV researchers that samples of ex-abusers could include ‘false desisters’, defined as the absence of overt forms of abuse and violence being attributable to the emotional labour and regulatory activities of victim–survivors, rather than the agentic decision-making and behavioural change of the abuser.
Victim–survivors frequently utilise strategies designed to mitigate the risk of violence, often by modifying their own behaviours rather than directly challenging the abuser's conduct (Goodman et al., 2003). This includes acquiescence to demands, avoidance of known triggers, emotional regulation to prevent escalation and assuming primary responsibility for household or child-rearing duties to minimise potential conflicts (Boxall, 2023). Although these strategies may have the effect of bringing about an end to some forms of abusive behaviour, they can have a detrimental impact on victim–survivors and can be experienced as a form of abuse. As Feld and Straus (1989: 148) observe: If the wife is effectively intimidated, she may be able to avoid assaults … Under such conditions, the threat of violence is ever present, and the wife remains its victim. Such desistance fits the letter of the definition, without fitting its spirit. (emphasis added)
Given this distinction, retaining the terminology of ‘false desistance’ remains analytically valuable. Conceptualising these cases as a variant of ‘desistance by default’ risks overlooking the central issue: that desistance in these cases is not merely externally driven but is maintained and conditional upon mechanisms and strategies that are themselves abusive. In criminological discourse, desistance is generally framed as an interaction between agentic efforts towards change and external social structures that facilitate or inhibit desistance (Maruna, 2001). False desistance, however, represents a dynamic in which the mechanisms that sustain non-offending serve to maintain control of the victim–survivor.
Differentiating between genuine and false desistance, therefore, requires direct engagement with victim–survivors’ lived experiences. Although most IPV desistance studies incorporate victim–survivor data to corroborate offender self-reports (Walker, 2017), these accounts are typically limited to assessments of whether physical violence has ceased, rather than examining whether the cessation itself has been contingent upon victim–survivor regulatory labour. This omission risks conflating reductions in visible violence with substantive desistance, thereby obscuring the structural and relational mechanisms through which coercive control is maintained.
These complexities raise important questions about how we define desistance – particularly whether partial or gradual reductions in abuse can be meaningfully distinguished from full cessation (Shapland, 2022). Rocque (2021) argues that reductions should be included in operational definitions of desistance, because this is consistent with desistance being a process, rather than a binary event. He contrasts this with recidivism, which is treated as ‘an all or nothing type of outcome’. As he goes on to explain: Desistance, as a process, implies that a certain amount of failure may be expected on one's journey toward cessation of criminal conduct … recidivism does not necessarily equate to failure. (Rocque, 2021: 7; emphasis added)
Building on the preceding discussion, it is clear that determining whether a period of IPV cessation or de-escalation constitutes desistance requires engagement with victim–survivors. They are best placed to reflect on the overall trajectories of abuse within their relationship, and also ascribe mechanisms or causes to explain any observed changes. In particular, if victim–survivors are able to attribute the changes to factors beyond their own agentic actions and emotional labour, it may constitute desistance. Further work is also needed to understand whether reductions in abuse are experienced as meaningful by victim–survivors and align with their own definitions of desistance. To address this gap, the current study examined how female victim–survivors define and make sense of desistance.
Methods
This article presents an analysis of transcripts from interviews conducted with female victim–survivors of IPV living in Canberra (Australia), who participated in a larger-scale research project exploring victim–survivors’ roles in IPV desistance (Boxall, 2023). To be included in the broader study, respondents had to be female, 18 years or older, and self-identify as being subjected to IPV perpetrated by a current or former male-identifying intimate partner. For recruitment, IPV was defined as physical and/or sexual violence (including threats and attempts) and emotionally abusive, harassing and controlling behaviours (e.g. stalking, verbal insults and restricting movements).
Recruitment and interviews took place in 2019. Respondents were recruited from the community through social media [X (formerly Twitter) and Facebook], local television, radio and print magazines.
For the current study, respondents also had to report that the abuse within their relationship had stopped or decreased in frequency and/or severity for at least 6 months while they were still partnered with their abuser. This period of reduced abuse could have occurred either while they were still partnered or after they had separated. This left a final sample of 15 women who are the focus of the current analysis.
In this study, our definition of IPV was not limited to behaviours reported to, or identified by, the police. Similarly, desistance was not measured through involvement with the criminal legal system. Although administrative data is frequently used to measure reoffending in IPV research, many incidents are never reported and others fall outside formal legal definitions of abuse. Most participants in this study described multiple incidents of abusive behaviour – including physical and sexual violence – that were not reported to police. As such, this study adopts a relational approach to desistance, grounded in victim–survivors’ accounts of change within their relationships.
The recruitment criteria for the current study – experiences of abuse reduction or cessation for 6 months or longer – is broadly consistent with those used in other studies (Anderson and Kirkpatrick, 2016) with a focus on primary forms of desistance, although the period of de-escalation or cessation was shorter. However, this was only as a starting point for discussions with participants about what desistance looked like for them. Consistent with narrative interview principles, rather than following a strict process of question and answer, participants were invited to tell the story of their relationship in their own words, and to define desistance based on their lived experiences.
Participants were generally interviewed once (one was interviewed twice), with interviews ranging in length from 30 to 118 min (median = 86 min, information missing for two interviews). Interviews were transcribed verbatim and entered into NVivo and analysed through multiple rounds of coding.
Recognising that victim–survivors may be subjected to multiple forms of abuse that co-occur over the life of their relationships (Morgan and Boxall, 2022), we grouped abusive behaviours into broad categories to support analysis and interpretation. These were: physical abuse (e.g. slapping, hitting, punching, non-fatal strangulation); sexual abuse and coercion (e.g. being intimidated into having sex); emotional abuse (e.g. name calling, being told they are worthless); and coercive control. Coercive control was defined as a pattern of persistent non-physical abusive behaviours that had the effect of entrapping participants in the relationship and limited their day-to-day autonomy. Although legal definitions of coercive control are now emerging in several Australian jurisdictions (see, for example, Crimes Act 1900 (NSW) s 54D), this study pre-dated these developments and so adopted the definition used in international IPV scholarship (Stark, 2009).
Ethical considerations
Given the sensitivity of the topic, extensive safeguards were implemented across all stages of the research. A key consideration was that participants were recruited through the general community, rather than specialist DFV support services. This meant that researchers could not assess participants’ emotional state, or determine whether they remained in an abusive relationship, at the time of recruitment. To address this, strict recruitment protocols were followed. For example, participants were invited to advise how and when it was safe for them to be contacted and whether it was safe to leave a voicemail message on their phone. All received detailed project information, including its voluntary nature and their right to withdraw at any time. Interviews were conducted using trauma-informed approaches and support information was provided, following participation and during a follow-up conversation with the lead author (for further detail, see Boxall, 2023). The project was approved by the Australian National University's Human Research Ethics Committee (protocol number 2019/142).
Sample
The median age of participants was 43 years (range 28–72). The majority (87%) were employed at time of the interview and were working in a range of professions, the most common being health services (42%) and education/community services (25%). Two-thirds of participants (67%) had a university education. Two of the participants were born in a country where English is not the dominant language.
The median length of relationship between participants and their abuser was 10 years. Two-thirds (67%) had at least one child with their partner, with a median of two children. Half were married to their partner and all but one (93%) had lived with their partner at some point during the relationship.
The sample used for the current study is unlike that in most other studies of its kind, because participants were recruited from the community, rather than DFV services or the criminal justice system. 1 This may explain the higher education and employment rates among participants, as those using DFV services often lack the resources to manage abuse independently (Voce and Boxall, 2018; Woffordt et al., 1994). Around half had contact with DFV services (n = 6) or police (n = 8), but six women reported no contact with either. These figures suggest that, although not recruited through services, a number of participants were still involved with formal support systems, and the relationship between employment and service engagement should therefore be interpreted with caution. It is also important to acknowledge that the recruitment site is also relatively unusual within Australia, because it has higher rates of employment, education, and health and well-being, compared with the rest of Australia (for further detail, see Boxall, 2023). As such, the sample could simply be consistent with the broader population of the recruitment site (Table 1).
Sample characteristics (n = 15)
Source: Ending Intimate Partner Violence Study (2019–2020) [Computer file]. Canberra: Australian National University. Principal Investigator: Hayley Boxall. Unpublished dataset. Available upon request, subject to ethical approval.
Values are percentages unless indicated otherwise.
Includes full-time, part-time and casual work.
Limited to women who were employed at time of interview. Excludes two participants for whom this information was missing.
As shown in Table 2, four participants (27%) reported that the violence and abuse had reduced or stopped entirely for a period of 6 months while they were still partnered with their abuser. Six (40%) reported that the period of reduction/cessation had lasted for 1–3 years and, for over half (n = 8), the reduction/cessation lasted for 5 years or more. As such, for most of the women in the sample, the period of reduction/cessation was sustained for a long period within their relationship (Table 2).
Characteristics of relationship, by participant
Source: Ending Intimate Partner Violence Study (2019–2020) [Computer file]. Canberra: Australian National University. Principal Investigator: Hayley Boxall. Unpublished dataset. Available upon request, subject to ethical approval.
Limitations
Several limitations of the current study need to be acknowledged. First, the sample was small, limiting the extent to which the findings can be extrapolated to the broader population of women (or men or non-binary people) subjected to IPV. Second, although recruitment involved diverse strategies – radio, magazines and social media – these were more likely to reach women with digital access and who engage with these forms of media. The recruitment materials were in English only, which may have excluded women from culturally and linguistically diverse (CALD) communities.
As such, the sample may not fully reflect the diversity of women's experiences, particularly across different cultural, socio-economic and linguistic backgrounds. This limited both the nature of the sample (who was included in the research) and also our ability to explore variations in desistance experiences of women. Notably, none of the women in the sample identified as Aboriginal and/or Torres Strait Islander, and only two participants were from CALD backgrounds. Although the broader study involved 40 interviews, it was exploratory in nature and not designed to support detailed intersectional analysis, including by language spoken at home or CALD status. 2 These limitations reflect a broader pattern in desistance research, which has often neglected how gender, sexuality, ethnicity and other axes of identity shape desistance pathways (Barr, 2019). As evidence grows that these factors influence desistance pathways (Barr, 2019; Giordano et al., 2002), more inclusive and purposive sampling approaches are urgently needed. Future research should be specifically designed to explore how these intersecting factors shape victim–survivors’ perspectives on desistance.
Finally, the study did not examine the sociocultural structures – such as class, religion or community norms – that may have influenced participants’ experiences. This was a pragmatic decision owing to time and scope constraints, rather than a dismissal of their relevance. Future studies should examine how these broader structures intersect with IPV desistance trajectories.
Findings
Analysis of the participants’ narratives demonstrated significant variation in the patterns of violence and abuse within the relationships and during the period of self-identified desistance. In essence, what emerged was evidence that the various forms of violence that participants were subjected to (characterised as physical violence, sexual violence, coercive control and emotional abuse) were often on different trajectories, during the identified period of desistance. Our examination of the abuse patterns and trajectories reported by participants led to the identification of four primary groups: 3 the cessation of all forms of abuse, overall reduction in abuse, false desistance and what we term ‘substitutive’ desistance. The characteristics of these groups are described in detail below (Table 3).
Trajectories of violence and abuse, by participant and type of violence and abuse reported
Note: X = cessation, ↑ = escalation, ↓ = de-escalation, ↔ stayed at the same level, ONSET = commenced.
Source: Ending Intimate Partner Violence Study (2019–2020) [Computer file]. Canberra: Australian National University. Principal Investigator: Hayley Boxall. Unpublished dataset. Available upon request, subject to ethical approval.
Cessation of all forms of abuse
As demonstrated in Table 3, it was very rare for participants to experience the cessation of all forms of abuse within a relationship that was ongoing at time of the interview. Only two respondents (Yvette and Alison) reported that the abuse had stopped completely for a period of 6 months or longer, while they were still partnered. Yvette said that, at the time of the interview, her partner (who she was still in a relationship with) had not been physically or sexually violent towards her, or controlling, during the preceding 6-month period. Although the period of non-violence had been relatively brief, Yvette strongly believed that the abuse would not start again. She based this assessment on what she perceived as her partner's genuine remorse for his previous abuse towards her. This pattern of abuse had culminated in a significant episode of violence, which he was very ashamed of, as well as significant changes in the dynamics in their relationship. Specifically, after this significant episode of violence, she threatened to divorce him and received a protection order, which placed strict conditions on his communication with her. 4
Emboldened by the protection order, Yvette also implemented new rules within the relationship, whereby her partner was not allowed to live at home with her full-time, until he had demonstrated his ability to be non-violent. Yvette attributed her partner's behaviour change in part to his realisation that she was not willing to accept his behaviours anymore and that he could lose the relationship if he was abusive. As she put it: I think he knows that he's hanging only, his fingernails on the side of the, the side of the roof you know. And it's not gonna take a whole lot for me to just go bzzzt, you're done. I don't need you anymore. I'm doing this on my own and I don't need you. (Yvette, 54 years old) I didn't [believe the behaviour change was real] for the first 4–6 months, I kind of just went along thinking, ‘oh it'll start up again’ … .[But], it was nice and calm. I felt really good. Things were better than they had ever been in our relationship … things changed and the whole dynamic in the relationship changed and it was a normal, non-abusive relationship. Like, and as I said, I didn't believe it to start with, I was like, ‘this is just too good to be true’, like, this man can't behave like this and continue to behave like this. (Alison, 40 years old) Per chance, he came upon my folder of everything. The documents going into everything he'd said, everything he'd done, his behaviours, his … explosiveness, all of the notes that I'd taken from [DFV service] both in [CITY 1] and [CITY 2]. Like, everything. He'd found it all and confronted me over it. And started bawling, said ‘I’m gonna change’, you know, ‘this isn't the man I want to be. I see now what you've been saying, I understand the … this isn't OK and I wanna change and I wanna get help’. (Alison, 40 years old)
Overall reduction in abuse
The most common definition of desistance described by participants (n = 6) was an overall reduction in the abuse they were being subjected to. For six participants, the overall reduction in abuse was described as being attributable to the complete cessation of physical violence within the relationship, and the reduction (n = 2) or stabilisation (n = 4) in other forms of abuse. As such, it involved a reduction in the versatility of abusive behaviours perpetrated against them. For example, Grace described that her husband was previously sexually violent towards her on a regular basis, which she attributed to his desire to impregnate her, and sense of entitlement to her body, emboldened by a belief that rape was legally permissable within marriage. When she resisted his sexual advances, he would overpower her with physical violence. However, at time of the interview, Grace said that her husband had not been physically violent towards her for almost 10 years and that the sexual violence had reduced significantly.
Grace attributed her husband's behaviour change to their engagement in relationship counselling together, and their counsellor reiterating to him that marriage could not be used as a defence against charges of rape. This realisation then led to constructive conversations between Grace and her husband, facilitated by their counsellor, about consent within the relationship: Initially, when we spoke to the counsellor, he still said ‘oh there's no such thing’ [as raping your wife]. And I thought, well, I can raise it in such a way as asking the counsellor ‘what sort of boundaries there would be? When would it look like something…?’ And so, in a way the counsellor seemed like a safe space. And she raised the concept of there is such a thing as rape within marriage, which to him was totally foreign…. And so I think then, in a way, I felt quite empowered by the process of knowing, ‘ok you know, I’m [not] coming up with some outrageous thing, this is completely normal’. (Grace, 45 years old)
For one participant, desistance was described as the de-escalation in the frequency or severity of all forms of violence and abuse in the relationship, but not cessation. Antonette reflected that her husband's physical violence towards her de-escalated for a 6-month period prior to the end of their relationship. Although he was still violent towards her, she believed that it was less frequent and that he was not attempting to significantly injure her, as he had done previously. She attributed this to her decision to leave the relationship, which she believed was a ‘relief’ to him, because he no longer had to feel shame related to his extramarital affairs and abusive behaviours.
False desistance
Although they reported an overall change in their partners use of violence and abuse, three participants believed that the cessation or significant reduction in abuse they experienced only occurred because of the micro-regulation of their own behaviours and their relationship, and their prioritisation of their partner's needs over their own. Although Elaine reported that the physical violence stopped for a 12-month period, she attributed this entirely to becoming her abuser's ‘little minion’. She reflected that she did everything that he asked of her and completely sublimated her own needs to prioritise his. This left her feeling ‘numb’ and like a ‘robot’. As she explained: I just thought, ‘Right I’m not doing anything wrong. I’m not doing anything at all. I'm doing everything exactly the way he tells me. Everything is going to be fine’. Like, it'll all be fine. I just kept telling myself that. And the violence actually stopped for a long time. Because I was just his little minion. I did everything I was meant to. I didn’t talk to anyone. I did every single thing he told me to do … I think if I had said the wrong thing or not done what he'd asked, it would have just continued. But the fact that I just … knuckled down and just did exactly what I thought was going to make him happy, I think that's what made it stop for that time. (Elaine, 30 years old) …his hands are fucked now … he's [got] arthritis in the hands. So, if he hits me, it’ll hurt him a hell of a lot longer than it hurts me, and that's when it got to the point where it's like, ‘yes I can leave now’. (Chelsea, 33 years old) …it was almost as soon as I really kind of confronted him, it was … his look just changed, that's the best I can describe it, it was almost like something had come over him. And then he…he does this particular expression with his face and then … yeah, put his fist up. Kind of got right up in my face and put his fist up … And then he just left. (Genevieve, 45 years old) [L]ife … there's always things … that he either didn’t want to hear or that a bill was due or that I couldn't do something. You know. So, there was always something there but, you know, that's life just happens to everybody. (Genevieve, 45 years old)
The end of Elaine's relationship was similarly precipitated by an incident of violence following a 12-month period of non-abuse. Reflecting on this incident, Elaine said that this time had helped her realise that, even when she did ‘everything right’, unless her partner was also motivated to stop perpetrating abuse and took actions to achieve this, he would never change. As demonstrated in the below quote, Elaine saw her partner's desistance as something that she was at least in part responsible for achieving: There's no reason for this [the abuse]. I’ve done everything I'm supposed to do …. I'm doing what you’re asking me to do. And then it's still happening. So … yeah, that was a big thing in my head as well. Like why is this still happening, if I’m doing everything right? (Elaine, 30 years old; emphasis added)
Substitutive desistance
The final cohort of participants were not easily classifiable into any of the three categories outlined above. Four participants said that physical forms of violence had stopped within the relationship, whereas other forms of abuse had escalated (n = 3; Mary, Tasmin and Elsie) or even started for the first time (n = 1; Bella). For these participants, there appeared to be no overall reduction in the abuse they were being subjected to. Instead, it appears that one form of abuse had been replaced with another. For this reason, we tentatively refer to this group as experiencing ‘substitutive’ desistance, which we define as the substitution of some – usually more overt forms – of IPV with non-physical abuse or new forms of violence.
Mary reported that her partner was not physically violent towards her for a period of 5 years, which she attributed, in part, to their son growing old enough to act as an effective deterrent; if her partner abused her, Mary's son would come to her defence. Notably, this period without physical violence was immediately preceded by an incident in which Mary's partner said their son was ‘getting too big for his boots’ and needed to be ‘put in his place’. Mary sternly told her partner that, if he did, she would report him to the police. Mary reflected that the threat appeared to shock him, because she had never threatened him previously when he had been violent towards her; instead, she usually acquiesced to most of his demands and tried to manage his moods. As such, he appeared to take her threat seriously and, after this point, never assaulted her again.
However, although the physical violence stopped, Mary said that her partner's controlling behaviours escalated significantly over time. He stopped her from using the family car, meaning she had to walk or use public transport for everything. Mary's partner also started locking her and their son out of the house when he believed they had broken the ‘rules’ of the house, which included challenging his authority or talking back. It appears that, in this situation the escalation of controlling behaviour was in part attributable to Mary challenging her partner's use of physical violence. Seemingly, Mary's partner substituted more subtle, non-physical abusive behaviours, to regain a sense of control over his family.
Bella reported that her partner's use of physical violence stopped entirely for 12 months prior to the end of the relationship, after she reported his threats of self-harm to support services at the university they were both attending as students, which subsequently led to her disclosing his abuse. As a result of the report, her partner was hospitalised, diagnosed with depression and anxiety, and made to attend counselling. He experienced this as humiliating, his feelings of shame apparently exacerbated by his Asian heritage. Bella, who was from a similar background, explained that their communities had very stigmatising attitudes towards mental illness and saw it as something shameful and indicative of a ‘weak character’.
However, although the physical violence stopped, Bella's partner started sexually abusing her, as ‘punishment’ for reporting his behaviours and his subsequent humiliation: To compensate for the troubles that I caused, I, I guess I became more intimate with him than I wanted. Because he kind of say things [like] ‘you did this to me, I deserve, you should be more intimate’ … In our relationship I kind of make it clear that there's no sex until marriage. But I was guilted into doing more than I wanted to do. (Bella, 28 years old)
Elsie reported that, during the early stages of their relationship, her partner would physically threaten her by punching walls next to her head. After she threatened to leave him if he did not stop, this aspect of the abusive behaviour ceased and never re-occurred. However, she reflected that over time, his controlling behaviours continued to escalate and, by the end of the relationship, she said she was ‘completely controlled’ by him. Reflecting on the relationship with the benefit of hindsight, Elsie questioned whether her partner's threatening behaviours were simply replaced by covert controlling behaviours that were less easily identifiable as abusive, but had the same desire outcome: entrapment. As she explained: So, he did stop that hitting. But he didn't stop the coercion or the control or that sort of thing obviously. He just figured out better ways to do it I suppose. (Elsie, 31 years old)
Discussion
The analysis identified both consistencies and inconsistencies between the definitions of desistance used by participants and those described in the broader desistance literature, as well as key considerations for future research.
A reduction in some forms of abuse was the most common definition of desistance used by participants
The most common definition of desistance used by participants (n = 7) was the overall reduction of abuse in the relationship. For most participants in this group, the overall reduction was attributable to the contraction in the versatility of the abuse they were being subjected to; specifically, the complete cessation of physical violence, while other abusive behaviours stabilised or de-escalated.
An important question that this research raises is how comfortable we should be with victim–survivors still being subjected to ‘some’ form of abuse and violence, albeit less serious and/or frequent. This feels instinctively unacceptable – no individual should be subjected to any form of violence or abuse. However, it is important that we acknowledge that the participants experienced de-escalation as an important change in the patterns of abuse and they were able to articulate the benefits of this for themselves and their families. In particular, some participants reflected that the abuse had reduced to a level that they believed was ‘manageable’, until such a time as it stopped entirely, or they felt safe enough to leave the relationship. For example, Grace felt that she needed to stay in the relationship until her children had become adults after which point she could leave. Meanwhile, Chelsea reported feeling trapped in the relationship until the physical threat posed by her partner had resolved. Further, Mary reflected that she had ‘managed’ her partner's abusive behaviours as best she could throughout their relationship, because she did not know where she could have lived with their son if she had left him.
The experiences of participants such as Grace, Mary and Chelsea are supported by the broader research, which has similarly demonstrated the range of structural barriers for women leaving abusive relationships, including custodial arrangements for shared children; victim–survivors may not leave abusive relationships because they are concerned about the safety of their children, if the abusive parent was given access to the children (Sani and Pereira, 2020). This fear was realised for many women who participated in this study, including Alison, whose abusive partner was initially awarded primary custody of their children, after she had fled the relationship and was living in a refuge. She reflected: People say to me ‘if you … knew what you know now, before you left, would you have left?’ And I said, unequivocally no. Because my children got court ordered back into a house with no protection with an abusive man. And me being in the house meant that I could protect them to some extent. I wouldn’t have left because I wanted to protect my children and I only left to protect my children. Honestly. So if I knew what was going to eventuate, I probably would never have left. (Alison, 40 years old)
Although this article has focused on the relational dynamics that shape how desistance is defined and sustained, the analysis also points to the importance of cultural context. As the accounts of Grace and Bella illustrate, cultural norms around family, privacy, mental illness, and shame influenced how abuse was interpreted and how change was assessed. While these dynamics were not a central focus of the study, they demonstrate the need to consider how culture, race, and migration background shape not only help-seeking behaviours, but also the meanings attached to desistance.
There was evidence of false and substitutive desistance, which were not experienced as ‘true’ forms of desistance
Although most participants expressed the belief that the violence had stopped or reduced in severity or frequency, for three, this change was dependent on their implementation of strategies that were themselves experienced as abusive. However, several other participants also reported a belief that the non-occurrence of abuse was at least in part attributable to strategies that they were using, and that these strategies had a negative (albeit minor) impact on them. For example, although Yvette reported that her husband had not been violent towards her for 6 months, she was constantly planning ahead, to mitigate any potential triggers for the abuse. Although Yvette did not see these behaviours as themselves abusive, she reported that they were certainly stressful.
Relatedly, some participants reported that, despite the cessation or reduction in abuse, they did not trust that their partner would not be violent towards them. Several referred to the ever-present threat of the violence starting again, or even escalating, and this was a significant motivator for their use of strategies and activities to maintain their partner's desistance. As Grace reflected, every few weeks or so, she would have sex with her husband, because she believed that doing so would help her to avoid more coercive or physically violent encounters.
There was also evidence of what we termed ‘substitutive’ desistance, where the cessation of physical violence coincided with the escalation or onset of other forms of abusive behaviours. Often, the former contributed to the latter; when a perpetrator's physical violence led to some form of sanction (e.g. a police report, a threat to leave the relationship), they would use covert forms of abuse, as a means of punishing the participant and re-establishing control over them.
This provides further empirical evidence of a phenomenon that has been described frequently within the broader IPV literature: that perpetrators may respond to sanctions by reducing their use of overt forms of violence, to minimise detection and criminalisation, and replace them with more covert forms of abuse. This finding has important implications for researchers seeking to explore IPV desistance pathways specifically. Desistance studies that use narrow definitions of IPV run the risk of inaccurately classifying individuals who do not report physical violence as ‘desisters’, when the abuse has simply changed in form and function. As such, researchers should use wholistic and broad understandings of IPV, when undertaking research on desistance.
Patterns of physical forms of IPV were easier for participants to describe than non-physical forms of abuse (including controlling behaviours)
Although most clearly demonstrated through the narratives of participants who experienced substitutive desistance, across all four groups, participants were more likely to report desistance from physical violence than from non-physical abuse. Consequently, many perceived themselves as having experienced desistance, raising critical questions about why physical violence was foregrounded in their narratives of IPV.
A key observation made by participants, though not part of the analysis here, was that they often did not recognise controlling and emotionally abusive behaviours as IPV when they started. Recognition of these forms of abuse often emerged only after prolonged exposure or upon exiting the relationship. Such abuse is inherently more covert, frequently co-occurring with tactics such as gaslighting or minimisation. This finding suggests that victim–survivors may more readily identify overt forms of abuse in their definitions of desistance, while overlooking or underestimating the significance of less visible ones, such as coercive control.
The difficulty in recognising non-physical abuse is well-documented in the broader literature (Coumarelos et al., 2023). Limited community awareness has driven legislative efforts to criminalise coercive control in Australia and internationally, with the aim of addressing prevailing attitudes that diminish the gravity of these behaviours. The push for criminalisation is also informed by a perceived gap between criminal justice frameworks and the lived experiences of victim–survivors (Stark and Hester, 2019). Traditional legal frameworks treat IPV as discrete incidents, such as physical assaults, which are more easily particularised and captured within criminal law. This is in contrast to the more complex and pervasive patterns of IPV, including coercive control, which are harder to define and legislate (Stark and Hester, 2019).
This literature provides a useful analytical lens for understanding why participants prioritised physical violence in their narratives; its immediacy and tangibility render it more recognisable and measurable as abuse. However, this emphasis poses significant challenges for IPV desistance research, which necessitates a nuanced approach to capturing whether abusive behaviours are escalating or reducing over time. Many desistance studies rely on quantitative ‘count’ data from criminal justice agencies, which privilege officially reported offences within predefined temporal frames. Such methodologies risk neglecting more insidious forms of IPV that may not be formally recorded. Future research should consider alternative data sources and methodological approaches that better encapsulate the complexities of non-physical abuse and its implications for desistance trajectories.
Conclusion
Historically, desistance and IPV research have rarely intersected, partly because of concerns about the applicability of desistance frameworks to IPV. This study makes a valuable contribution to addressing this gap, by examining traditional definitions of desistance through the experiences of 15 female victim-survivors. Our findings highlight key limitations in prevailing definitions, particularly the assumption that desistance entails cessation across all forms of offending. Instead, victim–survivors reported variable pathways, with some behaviours ceasing while others persisted or escalated, challenging conventional understandings of desistance.
The study underscores the necessity of incorporating victim–survivor perspectives alongside offender data in desistance research. Beyond assessing whether abuse has occurred, it is crucial to examine the strategies victim–survivors employ to mitigate harm and their role in the behaviour change process.
Overall, our findings highlight the complexity of IPV desistance and the importance of aligning research methodologies with the lived realities of victim–survivors. A more nuanced approach is required in both research and intervention design to capture the fluidity of desistance trajectories and improve responses to IPV.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
