Executive Summary

Domestic and family violence (DFV) policy in New South Wales and nationally has undergone substantial reform over recent decades. Policy advocacy focused on an intersectional feminist analysis frame has significantly improved recognition of coercive control, strengthened institutional responses, and enhanced protection for women and children facing high-risk physical violence. These gains are foundational and must be preserved.

However, current policy frameworks are primarily calibrated to detect and respond to forms of abuse that are direct, incident-based, and physically manifest. Where coercive control is expressed indirectly – through reputational, relational, or institutionally mediated mechanisms – harm may be less visible, more prolonged, and less consistently captured within existing systems.

These pathways may be associated with extended exposure to coercive dynamics and escalation outcomes that include psychological collapse and suicide. Current DFV mortality frameworks focus predominantly on intimate-partner homicide and do not systematically capture domestic-violence antecedents in suicide data, creating a structural gap in how harm is measured and understood (AIHW, 2025; House of Representatives SPLA Committee, 2025).

This article proposes the application of a complementary analytical lens – a causal behavioural systems framework – alongside existing pattern-based approaches. Under this model, domestic violence is understood as a dynamic interaction system in which outcomes emerge from:

• behavioural drivers associated with coercive control
• response profiles under sustained coercive pressure
• institutional interpretation and intervention

Within this framework, coercive control remains the central causal mechanism, but may be expressed through both direct and indirect forms. Outcomes are not determined solely by behaviour, but by how behaviour, response, and institutional systems interact over time.

A key implication is that early help-seeking functions as a critical inflection point between divergent pathways of disclosure and suppression. These pathways are not equivalent: suppression leads disproportionately to escalation and prolonged harm, particularly in indirect abuse contexts where anticipated system response can actively inhibit reporting. In parallel, perpetrators may adapt behaviour in response to institutional constraint, shifting modes of coercive control and reinforcing escalation dynamics over time.

The framework identifies three systemic risks:

• escalation risk driven by adversarial or retaliatory dynamics
• suppression risk where anticipated system response shapes non-disclosure or avoidance
• misidentification risk in ambiguous or indirect presentations

A critical implication is that early-stage help-seeking operates as a decision point between divergent pathways, with suppression leading disproportionately to escalation and prolonged harm, particularly in indirect abuse contexts.

Recognising these dynamics enables earlier identification of risk, more accurate classification, and additional opportunities for intervention. The policy implication is not to replace existing frameworks, but to extend them to better align system design with the full causal structure of harm.

Introduction

Frameworks and system design

DFV policy is not theory-neutral.

Every system implicitly defines both what counts as harm and what counts as reliable knowledge of harm. These assumptions shape intake, assessment, and intervention processes, and determine what a system can detect – and therefore what it can prevent.

In Australia, the adoption of an intersectional feminist analytical lens has delivered substantial improvements in recognising coercive control especially when involving physical or sexual violence. This has allowed front line responders to move beyond a mutual-conflict framing, and to prioritise protection for those at risk of severe harm – primarily women and children subjected to direct forms of coercive control.

These gains are significant. At the same time, reliance on a dominant analytical lens introduces a predictable limitation: systems optimised for one pattern of harm become less sensitive to others.

This limitation is not only one of pattern recognition, but of system structure. As developed in this article, institutional responses can shape behavioural pathways themselves, influencing whether individuals move toward disclosure or suppression, and how perpetrators adapt strategies under constraint.

Misidentification in cases involving cross-allegations is already recognised in Australian DFV research, with national guidance emphasising the need to identify the “person most in need of protection” through structured, pattern-based assessment rather than first-contact assumptions (ANROWS, 2020a; 2020b).

This article proposes a complementary analytical approach. Rather than focusing primarily on observable patterns, it applies a causal behavioural systems lens to examine how harm is generated through interaction between behaviour, response, and institutional processes over time.

Current pattern-based model (simplified)

Dimension	Current system orientation	Implication
Analytical lens	Pattern-based, calibrated to dominant trajectory (direct, physically violent harm patterns)	High accuracy in visible, high-risk cases
Gender framing	Strongly gendered, reflecting well-evidenced patterns of male perpetration and female victimisation in severe physical violence	Improves protection in dominant pathway but may reduce sensitivity to non-dominant presentations
Expression of harm	Primarily direct, incident-based, physically manifest	Indirect or cumulative harm less visible to system
Victim treatment	Largely undifferentiated response profiles	Limited ability to explain variation in exposure duration and escalation
Evidence model	Event-based, point-in-time assessment	Fragmented or longitudinal patterns harder to interpret
Outcome interpretation	Focus on severe physical harm endpoint (for example homicide)	Other escalation pathways, including psychological collapse, suicide or familicide-suicide, are under-captured

This model has delivered substantial gains in detecting and responding to the dominant pattern of harm. However, it leaves the causal drivers of variation in other pathways less well explained, particularly where harm is indirect, cumulative, or mediated through institutional processes over time. It is therefore highly effective within its primary domain, but less well suited to cases where harm does not align with these structural assumptions.

The contribution of this article is to integrate these dynamics into a unified causal behavioural systems framework, clarifying how interaction between behaviour, response, and institutional processes produces divergent trajectories.

Bounded success

Existing DFV frameworks have corrected major historical failures, particularly the mischaracterisation of domestic violence as mutual conflict rather than patterned coercive control. This has led to:

• improved credibility of victim reports
• stronger institutional responses
• recognition of non-physical abuse
• improved protection outcomes in high-risk cases

National data continue to demonstrate that women bear the greatest burden of severe physical harm, including intimate-partner homicide, which remains a central focus of current policy frameworks (AIHW, 2025).

These successes must be preserved.

The question addressed here is whether exclusive reliance on a single analytical lens limits system capacity to detect other harm pathways, and whether complementary approaches can improve early identification and prevention.

From patterns to causal structure

Patterns describe how harm presents. Prevention requires understanding how those patterns are produced.

A causal behavioural systems approach shifts analysis from descriptive classification to underlying structure. Domestic violence is understood as a dynamic system in which outcomes emerge through interaction between:

• control-seeking behaviour
• response under coercive pressure
• institutional interpretation and intervention

A critical feature of this system, illustrated in Figure 1, is that institutional engagement does not simply interrupt abusive dynamics but can reshape them. Perpetrators may adapt behaviour in response to perceived or actual institutional constraint, including shifting between direct and indirect modes, modifying narrative strategies, and deploying systems-based forms of coercion. These adaptations feed back into the system, altering subsequent interaction and detection pathways.

This shift allows earlier identification of risk, particularly in cases where observable patterns are incomplete or ambiguous.

Behavioural drivers and expression

Coercive control as the core mechanism

Across trajectories, coercive control is the central causal mechanism.

It may be expressed:

• directly, through physical violence or intimidation
• indirectly, through reputational, relational, or institutional means

A substantial body of psychological research demonstrates that aggression is expressed through different modes. On average, men are more likely to engage in direct aggression, while women show a greater propensity toward indirect or socially mediated forms such as reputational harm and third-party influence (Hess & Hagen, 2006; Davis et al., 2019).

These are population-level tendencies with significant overlap and do not determine individual behaviour. However, they have important implications for detection. Indirect forms of coercive control are structurally less visible, more dependent on institutional interpretation, and have been less consistently measured.

Australian data systems do not currently capture the prevalence, patterns, or institutional amplification of indirect coercive control or systems abuse in a consistent way (AIHW, 2024; BOCSAR, 2025).

As discussed later, indirect coercive control can also emerge as a perpetrator stratagem where intervention by institutional responses has effectively prevented ongoing patters of direct coercive control.

Behavioural risk factors

Non-physical forms of intimate partner violence, including coercive control, are associated with behavioural risk factors such as emotional dysregulation, retaliatory escalation, and traits linked to personality pathology.

Research shows associations between coercive control and borderline traits, narcissistic traits, and broader personality disorder severity across both sexes (Spencer et al., 2019; Day et al., 2025; Oliver et al., 2024).

From a policy perspective, the key variable is behaviour rather than diagnosis.

It is also relevant to policy that research indicates that this key risk variable does not show substantial gender-based disparity. We should therefore consider the possibility that the gendered difference in coercive control behaviour may be a difference in mode (male = direct, female = indirect), rather than a gendered difference in frequency of offending. This requires further research to confirm, as indirect coercive control has been relatively less researched.

Response profiles under coercion

Outcomes are shaped not only by perpetrator behaviour, but by how individuals respond under sustained coercive pressure.

These response profiles are best understood as adaptive responses to coercive environments, not causes of abuse.

• Profile A – Stable / self-protective – Clearer boundaries, earlier help-seeking, shorter exposure
• Profile B – Accommodative – Conflict minimisation, delayed disclosure, prolonged exposure
• Profile C – Dysregulated – Emotional escalation, high-conflict interaction, rapid system engagement

These profiles influence both exposure duration and system interpretation.

Model matrices – interaction pathways over time

The following matrices operationalise the causal behavioural systems model. They illustrate how different pairings of behavioural expression and response profile progress through key system phases, and how institutional interaction shapes outcomes.

Matrix 1 – Behaviour × response across system phases

The matrix shows that system performance is highest where behavioural expression is direct and response is stabilised, and lowest where harm is indirect and response is accommodative.

Matrix 2 – Anticipation and adaptive strategy dynamics

Behaviour is not only reactive, but anticipatory. Individuals adjust both disclosure and coercive strategies based on expected system classification and response.

While Matrix 1 describes how trajectories unfold across system phases, Matrix 2 shows how those trajectories are shaped in advance by anticipated system response, and subsequently altered through behavioural adaptation.

Behavioural expression	Response profile	Expected system response	Pre-engagement behaviour	Post-intervention adaptation	System risk
Direct / violent	Profile A – Stable	Likely validation and protection	Early exit or rapid help-seeking	Reduced direct behaviour under constraint	Low residual risk if intervention is effective
	Profile B – Accommodative	Uncertain response; risk of escalation	Suppression or delayed help-seeking	Shift to indirect control strategies	Escalation risk prior to detection
	Profile C – Dysregulated	Adversarial interaction expected	Reactive engagement	Continued escalation; contested narratives	High-conflict entrenchment
Indirect / institutional	Profile A – Stable	Uncertain recognition of harm	Evidence accumulation; cautious disclosure	Strategic narrative contestation	Delayed classification risk
	Profile B – Accommodative	Risk of misidentification or disbelief	Suppression, withdrawal, or partial disclosure	Increased institutional leverage by controlling party	Misidentification and prolonged exposure
	Profile C – Dysregulated	High-conflict response expected	Reactive or premature disclosure	Narrative escalation and institutional overload	System amplification of conflict
Mixed / adaptive	All profiles	Variable	Strategically tailored behaviour	Dynamic switching of tactics	Complex classification risk

These matrices extend the earlier summary of the current pattern-based system by showing how behaviour, response, and institutional processes interact dynamically over time. They illustrate why systems calibrated primarily to direct, visible harm pathways may intervene later, or less accurately, in indirect or behaviourally complex cases.

These dynamics contrast directly with the pattern-based system outlined in the Introduction, demonstrating how underlying system assumptions shape both detection capability and behavioural adaptation.

System overview – causal behavioural interaction model

Figure 1 summarises the full system in a single view, showing how behaviour, response, and institutions interact dynamically over time.

Outcomes are shaped by behavioural response to anticipated system interaction, with suppression leading to escalation and prolonged harm, and disclosure enabling classification and early intervention. Behavioural adaptation and feedback loops recursively shape subsequent system entry.

This simplified representation highlights three defining features of the model:

• outcomes emerge from interaction, not single variables
• institutional response is part of the causal system, not external to it
• behaviour is adaptive and anticipatory, creating feedback loops that shape future trajectories

Figure 1 Domestic violence interaction system showing divergent disclosure and suppression pathways and associated feedback dynamics.

Interaction pathways and institutional dynamics

Abuse trajectories evolve across stages:

1. Primary behavioural interaction
2. Help-seeking inflection point
3. Institutional engagement
4. Post-intervention adaptation

Within this structure, help-seeking functions as a critical system inflection point rather than a linear stage. Individuals move into divergent pathways of disclosure or suppression, each associated with distinct institutional responses and markedly different downstream outcomes. As shown in Figure 1, these pathways are not equivalent: disclosure enables classification and potential early intervention, while suppression opens a pathway toward escalation, delayed system entry, and prolonged harm.

At the point of institutional engagement (following the decision regarding help-seeking), institutional systems become active participants in shaping outcomes.

Existing tools – including DVSAT and Safer Pathway – are highly effective in detecting high-risk, physically visible cases but are less well calibrated to identify indirect or narrative-based harm patterns (NSW DCJ, 2017; 2019; ANROWS, 2020a). Where behaviour is indirect, evidence fragmented, and narratives contested, classification becomes more uncertain.

Feedback loops and system effects

Three interacting feedback loops shape outcomes, operating across behavioural, institutional, and adaptive layers of the system:

1. Misidentification Loop
2. Retaliation Loop
3. Suppression Loop

These loops do not operate independently but interact across pathways, forming a recursive system in which behaviour, institutional response, and adaptation continuously reshape one another.

Misidentification loop

Misidentification is acknowledged as a persistent issue in DFV systems, particularly in cross-allegation contexts (ANROWS, 2020a; Engel, 2024).

Ambiguity in presentation, particularly in indirect or high-conflict cases, can lead to classification errors at the point of institutional engagement. These misidentifications are not static: once an initial classification is made, subsequent interpretation of behaviour, evidence collection, and institutional response tend to reinforce that classification over time.

As illustrated in Figure 1, the misidentification loop is primarily driven by institutional processes but interacts with both perpetrator adaptation and victim response. In indirect abuse contexts, where harm is mediated through narrative and relational mechanisms, perpetrators may actively shape institutional interpretation through narrative strategy and systems abuse, increasing the likelihood of adverse classification. Victims, in turn, may adapt behaviour in response to perceived or actual misclassification, including withdrawing from engagement or altering disclosure patterns, which can further reduce the system’s ability to accurately identify coercive dynamics.

This recursive process is particularly pronounced in cross-allegation contexts and in cases involving indirect coercive control, where evidentiary ambiguity is higher and system interpretation plays a greater role in determining outcomes. The result is a self-reinforcing loop in which initial misclassification shapes subsequent behaviour and system response, reducing the probability of corrective intervention over time, and increasing the likelihood of increased harm.

Retaliation loop

Intervention can act as a trigger for behavioural escalation, particularly where coercive control is challenged or constrained. This is true of both Direct and Indirect abuse pathways.

As shown in Figure 1, this process is closely linked to perpetrator adaptation. When direct forms of coercive control are disrupted through institutional action, perpetrators may respond by intensifying behaviour or shifting to alternative modes of control, including indirect, reputational, or systems-mediated strategies.

This adaptive escalation is not random but strategic. It reflects an attempt to restore control under new constraints, often through mechanisms that are less visible to the system or more difficult to classify.

In some cases, this may also involve increased narrative contestation or the use of institutional processes themselves as instruments of control. This is often the case in Indirect control, where reporting results in an escalation of narrative control, and increase weaponisation of institutions as tools of abuse.

At the same time, victims may respond to escalation with reactive or defensive behaviour, particularly under prolonged exposure or in high-conflict conditions. Where this behaviour is interpreted without context, it may contribute to further misclassification, linking the retaliation loop directly to the misidentification loop.

The overall effect is a feedback dynamic in which intervention, while necessary, can reshape rather than immediately resolve coercive dynamics. Without accurate interpretation and timely intervention, this loop may drive cases from initial detection into higher-conflict or more complex trajectories, increasing the risk of escalation and harm.

Suppression loop

Individuals may delay or avoid disclosure based on anticipated system response. Qualitative research indicates that anticipated disbelief or misclassification can significantly shape help-seeking behaviour (Fitz-Gibbon et al., 2025; Peterson, 2023).

Suppression operates through distinct mechanisms across direct and indirect abuse pathways.

In contexts of direct abuse, particularly where harm is physically acute, suppression tends to take the form of deferred disclosure. While victims may delay reporting due to denial, dependence, or hope that the abuse will cease, the immediacy and severity of harm create a sustained incentive to seek intervention, resulting in eventual system entry in many cases.

By contrast, in indirect or institutionally mediated abuse contexts, suppression is more structurally embedded. Here, anticipated system response plays a decisive role: where victims expect disbelief, narrative reversal, or misidentification, the perceived cost of disclosure may exceed the expected benefit of intervention. Suppression in this context is therefore not merely temporal delay but an active inhibition of reporting behaviour. These dynamics may be particularly pronounced among male victim-survivors, who are more likely to encounter indirect forms of coercive control and elevated risks of misidentification within gendered institutional frameworks.

The result is a divergence in pathways: direct abuse environments tend toward delayed disclosure, while indirect abuse environments are more likely to produce sustained non-disclosure, prolonged exposure, and downstream escalation.

Pathway	Incentive to report	Expected cost	Outcome
Direct	High	Moderate	Deferred disclosure
Indirect	Moderate	High	Suppression

Feedback Loops are Actor-Differentiated

As illustrated in Figure 1, these feedback loops are not independent processes but operate within an actor-differentiated system. Misidentification effects arise primarily through institutional interpretation, retaliation through perpetrator response to constraint, and suppression through victim response to anticipated system outcomes. Together, these loops recursively shape both behaviour and system entry over time.

Loop	Primary driver	Key mechanism
Misidentification	Institution	Classification reinforcement
Retaliation	Perpetrator	Adaptive escalation
Suppression	Victim / system interaction	Non-disclosure

Mortality and escalation pathways

Current DFV frameworks prioritise homicide as the primary fatal outcome. However, Australia lacks a coherent system for linking domestic violence trajectories to suicide.

This gap has been recognised in parliamentary inquiry and research literature, with no systematic coding of DFV antecedents in suicide data (House of Representatives SPLA Committee, 2025; Fitzpatrick et al., 2022).

The result is a structural asymmetry:

• physically visible harm is recorded and measured
• indirect, cumulative, and psychological harm is under-represented

This does not diminish the severity of physical violence. It highlights a limitation in how harm pathways are currently captured.

Policy implications

Early behavioural triage

The analysis identifies a structural gap in current intake and referral pathways. Entry into coordinated DFV response systems is predominantly mediated through police or crisis-driven triggers. This creates a mismatch for cases involving indirect coercive control, where harm may be significant but does not initially present as a discrete, reportable incident.

A pre-escalation behavioural triage pathway is proposed as a complementary entry point within the Safer Pathway framework. Its purpose is to enable early, structured assessment of coercive dynamics before escalation into crisis, adversarial processes, or misclassification.

The proposed triage pathway directly targets the system inflection point identified in this analysis. By intervening prior to formal classification, it alters the decision environment at the point of help-seeking, reducing suppression driven by anticipated system response and disrupting escalation pathways before they become entrenched.

Core function

The triage pathway would provide a standardised, behaviourally informed intake and assessment layer that sits upstream of formal classification and enforcement processes. It would:

• identify patterns of coercive control across both direct and indirect modes
• assess response profile indicators associated with exposure duration and escalation risk
• detect early signals of misidentification, suppression, or adversarial dynamics
• triage cases into appropriate downstream pathways, including police, crisis services, or multi-agency coordination.

It also reduces the scope for strategic perpetrator adaptation by enabling earlier detection of indirect coercive patterns before institutional processes can be leveraged as instruments of control.

Operational model

The pathway would operate as a parallel entry channel, accessible through multiple referral points:

• self-referral via a publicly accessible intake service
• referral by GPs, psychologists, and counsellors
• referral from community services and helplines

Accessibility, safety, and early reporting

A central design requirement is that the pathway is perceived as safe for early reporting, particularly for cohorts who face elevated misidentification risk in existing systems, including a significant subset of male victim-survivors. In these cases, help-seeking is often delayed or avoided due to the perceived risk that initial engagement will trigger:

• adverse classification or ‘perpetrator’ role assignment at first contact
• escalation into adversarial processes without adequate behavioural context
• amplification of indirect coercive strategies, including narrative manipulation or systems abuse

This is not a passive effect. It is an anticipatory behavioural response to expected system classification, and therefore a predictable feature of system design rather than an anomaly.

The result can be a prolonged “pressure-cooker” exposure dynamic, characterised by sustained coercive pressure without external interruption, validation or opportunity for relief. Over time, this increases the likelihood of:

• psychological deterioration and collapse
• retaliatory or reactive escalation
• misidentification scenarios in which the primary victim is misidentified as the aggressor
• role-reversal scenarios where the initial victim engages in reactive escalation from indirect coercive control to direct violence
• broader escalation risk affecting both parties and, where present, children

This dynamic increases the probability of reactive or situational violence, which may then be interpreted in the context of escalation to physical violence, or in the absence of context of prior abuse leading to misidentification.

The alternative entry pathway is designed to directly interrupt this trajectory by providing a lower-consequence, non-confrontational reporting option that:

• avoids immediate adversarial framing or enforcement escalation
• does not require immediate role assignment at the point of initial disclosure
• reduces incentives for strategic narrative manipulation in early stages of indirect coercive control by avoiding immediate institutional amplification of first-presenting narratives
• enables confidential, staged disclosure and evidence capture

By lowering the perceived risk of engagement, the pathway increases the likelihood of earlier reporting, particularly in indirect or ambiguous cases where current systems are least effective. In effect, the pathway shifts the system from a high-consequence, first-contact classification model to a lower-consequence, evidence-building and evidence‑preservation model, thereby improving both early engagement and downstream accuracy.

Assessment would be conducted using a structured toolset incorporating:

• behavioural pattern analysis rather than single-incident reporting
• prompts for indirect coercive control and systems abuse signals
• assessment of disclosure consistency and evidence trajectory over time
• screening for acute psychological risk, including psychological collapse, suicidality and familicide-suicide

Outputs would include:

• provisional behavioural profile
• risk grading across escalation pathways
• recommended referral pathway and urgency level

Integration with Safer Pathway

The triage function would be integrated, not separate, from existing Safer Pathway coordination mechanisms. Specifically, it would:

• provide higher-quality referrals into Safety Action Meetings (SAM)
• improve early-stage classification by supplying longitudinal behavioural context
• enable cases without immediate physical risk to enter coordinated response earlier

Decision thresholds and escalation

Clear thresholds would guide movement from triage to formal intervention:

• indicators of immediate danger trigger direct referral to police and crisis response
• evidence of escalating coercive control triggers SAM referral
• ambiguous or contested cases remain under monitored assessment with evidence accumulation

This prevents premature role assignment while ensuring serious risk is escalated without delay.

Staffing and capability

A critical requirement of the pathway is that intake and early assessment are led by practitioners with demonstrable expertise in psychology and behavioural assessment. This reflects the core premise of the model – that accurate early interpretation of coercive dynamics requires understanding of behavioural drivers, trauma responses, and pattern formation over time, rather than reliance on incident-based classification alone.

The pathway would therefore require a multi-disciplinary capability, combining:

• clinical and behavioural psychology expertise for pattern recognition, trauma-response assessment, and identification of coercive control dynamics across direct and indirect modes
• capability in assessing response profiles (stable, accommodative, dysregulated) and their implications for exposure and escalation risk
• DFV-specific expertise to align behavioural assessment with statutory frameworks and risk protocols
• legal and system expertise to ensure appropriate downstream referral and compliance with evidentiary standards
• trained intake practitioners capable of managing complex, contested disclosures through structured, non-adversarial engagement

Psychologically informed intake is particularly critical in cases involving:

• indirect coercive control or systems abuse
• inconsistent or staged disclosure patterns
• high-conflict or dysregulated interaction dynamics
• elevated risk of misidentification or suppression

Embedding psychological expertise at the point of entry ensures that early assessment reflects causal structure rather than surface presentation, improving both immediate triage decisions and the quality of downstream institutional response.

Data and evidence model

A central feature is longitudinal evidence capture. The pathway would:

• record behavioural indicators over time rather than single-point disclosures
• preserve narrative, relational, and institutional evidence
• create structured data inputs to support downstream investigation and risk assessment

This addresses a key limitation of current event-based models.

System benefits

The triage pathway directly addresses the three systemic risks identified earlier:

• reduces misidentification risk by delaying premature classification and enabling parallel evaluation of competing claims
• reduces suppression by providing a lower-consequence, non-police entry point for disclosure
• reduces escalation by identifying high-risk trajectories earlier and routing them to appropriate intervention

Safeguards

To ensure alignment with existing protections and avoid unintended harm, the pathway would:

• operate under clear non-mediation principles where coercion is suspected
• maintain strict thresholds for referral to enforcement pathways
• include oversight and audit mechanisms, including tracking of classification outcomes and corrections

This approach extends system capability without displacing existing strengths. It provides an earlier, behaviourally grounded entry point that improves detection, classification, and prevention across the full range of DFV harm pathways.

Conclusion

DFV policy in Australia has made substantial progress in recognising and responding to high-risk patterns of violence. These gains must be preserved.

However, systems optimised for dominant patterns remain structurally limited in their ability to detect less visible pathways. Where coercive control is expressed indirectly, and where response involves suppression or accommodation, harm may evolve in ways current systems are not well designed to detect.

The analysis shows that these limitations arise through identifiable system dynamics. Help‑seeking operates as a decision point between divergent pathways, with suppression producing structurally different outcomes from disclosure. At the same time, perpetrator adaptation and interacting feedback loops mean that institutional responses can reshape, rather than simply resolve, coercive dynamics. These processes are not incidental but central to how harm evolves within the system.

A causal behavioural systems lens extends existing frameworks by showing that outcomes emerge from interaction between behaviour, response, and institutional interpretation over time.

At its core, this analysis shows that system design not only determines what is detected, but also shapes behavioural pathways themselves, influencing whether harm is disclosed, suppressed, or escalated over time.

Current evidence systems are strongest in tracking acute physical harm but remain underdeveloped in capturing indirect coercive control, misidentification dynamics, and long-term outcomes such as suicide (AIHW, 2024; House of Representatives SPLA Committee, 2025).

Effective prevention requires system design aligned to causal structure. This includes:

• capturing behavioural evidence longitudinally
• recognising feedback loops and adaptation
• providing pathways that encourage early disclosure
• improving detection across multiple harm trajectories

Plural analytical lenses do not weaken policy coherence. They strengthen system capability.

A system that can recognise multiple pathways, interpret behaviour accurately, and intervene earlier is better positioned to reduce harm across the full spectrum of domestic violence.