The propensity to control: Non-sexual violence as probative of sexual offending in the intimate partner context

Abstract

This article explores, with reference to four recent New Zealand appellate court decisions, the use of evidence of prior non-sexual offending by a defendant against the same complainant offered to prove sexual offence charges. Such ‘relationship propensity evidence’ can be particularly crucial for explaining the defendant–complainant dynamic in cases involving intimate partner violence. However, in some cases, courts have applied traditional common law ‘similar fact’ notions of linkage and coincidence to exclude evidence of non-sexual offending in relation to sexual charges. We argue that this is an unsatisfactory outcome that is largely resultant from the governing provision being designed to assess similar fact reasoning. Rather, we submit that in the context of intimate partner violence, seemingly discrete and unrelated forms of violence should be understood as potentially linked by the underpinning dynamic of coercive control. This shift in characterisation of varying forms of violence—from discrete and dissimilar to connected by a dynamic of coercive control—will result in a more flexible approach to the cross-admissibility of relationship propensity evidence in appropriate cases, including when it comes to offering evidence of physical violence as probative of sexual offending.

Introduction

Intimate partner violence can take a range of forms, from physical and sexual to emotional and financial. In some cases involving domestic abuse charges, it can be difficult for factfinders to understand the complex nature and dynamics of a defendant–complainant relationship, particularly when a number of separate incidents and charges are at play. As such, the cross-admissibility of seemingly disconnected types of offending—for example, non-sexual and sexual assaults—can take on particular importance when it comes to the prosecution of intimate partner violence.

Commonly referred to as ‘relationship propensity evidence’, evidence of separate and seemingly dissimilar offending against the same complainant by the same defendant can be crucial for explaining the dynamic of the defendant–complainant relationship, ¹ rather than presenting an artificial and sanitised picture. In this article, we explore, by way of reference to four recent New Zealand appellate decisions, the admissibility of relationship propensity evidence in the context of intimate partner offending.

The analysis illustrates how in some cases, courts apply traditional ‘similar fact’ notions of linkage and coincidence, resulting in the exclusion of evidence of non-sexual offending in relation to sexual offending. We argue that such an approach represents a problematic and artificial delinking of sexual and non-sexual violence, and reflects a flawed understanding of the complexity of intimate partner violence. Rather, we submit that in the context of intimate partner violence, seemingly discrete and unrelated forms of violence should be understood as potentially linked by the underpinning dynamic of coercive control. This shift in characterisation of varying forms of violence—from discrete and dissimilar to connected by a dynamic of coercive control—may impact the cross-admissibility of relationship propensity evidence, including when it comes to offering evidence of physical violence as probative of sexual offending.

We begin with an introduction to the admissibility rules governing propensity evidence in New Zealand. These are set out in sections 40 to 43 of the Evidence Act 2006 (hereafter ‘the Act’), which govern the admissibility of what was known at common law as ‘character evidence’. The application of the Act's propensity rules to relationship propensity evidence is somewhat muddled, as the propensity rules were tailored to govern propensity evidence offered to support an orthodox ‘similar fact’ argument—that because the defendant has behaved similarly, in relation to a different complainant, on a different occasion, they are more likely to be guilty of the current charges. Importantly, the courts have emphasised that in assessing the probative value of relationship propensity evidence, it is crucial to identify ‘broad similarity’ between the propensity and index offending. ²

In the next section, we introduce Evan Stark's influential theory of ‘coercive control’. The ‘coercive control’ framework conceptualises certain relationships as involving a pattern of coercive, controlling behaviour, rather than a series of situational, unrelated violent incidents. Where a relationship is characterised by coercive control, this has implications for the assessment of ‘similarity’. Viewed through a coercive control lens, discrete and seemingly dissimilar incidents of violence are recast as similarly relating to the offender's overarching objective of achieving dominance and control within the abusive relationship.

In the third section, we introduce two recent Court of Appeal judgments, S (CA145/2020) v R ³ and H (CA595/2021) v R, ⁴ in which evidence of a defendant's non-sexual physical violence towards a complainant was deemed inadmissible in relation to separate sexual violence charges. In both cases, this was despite indications that the relationship involved an overall pattern of coercive control. These cases illustrate how propensity rules, developed to deal with orthodox ‘similar fact’ prior offending, are easily misapplied in relation to evidence of relationship propensity evidence. In both cases, the non-sexual physical offending was deemed insufficiently similar to be probative of sexual offending. This resulted in an artificial delinking of discrete and different forms of the same broad coercive control dynamic.

In the fourth section, we compare these cases to two other recent decisions, Pahi v R ⁵ and Holmes v R. ⁶ These cases, in comparison to S v R and H v R, illustrate an approach to relationship propensity evidence that takes into account how, in certain abusive relationships, coercive control underpins discrete and seemingly dissimilar forms of violence. This shift in characterisation of varying forms of violence—from discrete and dissimilar to connected by a dynamic of coercive control—results in a more flexible and appropriate approach to the cross-admissibility of relationship propensity evidence.

We conclude with reflections on how, by recognising intimate partner relationships as possible sites of coercive control, courts can more accurately assess the probative value of separate incidences of offending offered to illustrate the dynamic of a relationship. We also discuss the need for further guidance from the Supreme Court for judges applying a propensity analysis to evidence offered for non-similar fact reasons.

Bad character evidence, similar fact reasoning and the ‘background loophole’: Common law and the Evidence Act 2006

In early common law, it was common to offer evidence of a defendant's ‘standing and reputation’ (Lacey, 2011: 159). By the time of the 1810 decision of Cole, ⁷ an exclusionary rule had emerged, on the basis that a defendant's guilt ought not to be judged on the basis of his general character. However, despite this ‘bad character’ exclusionary rule, character evidence was frequently admitted to support ‘similar fact’ reasoning, provided it was sufficiently probative (Munday, 2018: 382). So-called ‘similar fact’ character evidence is evidence of similar offending on a separate occasion, ordinarily against a separate complainant; such evidence shows ‘a pattern of conduct extrinsic to that in relation to the charge’ (Downs, 2021: 175). In many cases, evidence of prior wrongful behaviour (usually prior convictions) of a kind sufficiently similar to the charges at hand could be offered as circumstantial evidence of guilt, provided it was ‘more than barely relevant to guilt’ (Munday, 2018: 383 (footnote omitted)). ⁸

As the common law approach to ‘similar fact’ bad character evidence developed, it became common to sidestep the restrictions on bad character evidence by categorising the evidence in question as ‘background’ (Munday, 2018: 383). Such material could be offered to assist understanding of the context of the offending—such as explaining the nature of the defendant–complainant relationship—without being subject to the ‘bad character’ exclusionary rule (a rule which was confusing and inconsistently applied). Instead of being treated as ‘bad character’ evidence, such so-called ‘background’ evidence was considered directly relevant to the case, as it formed ‘part of the narrative’. ⁹

So-called ‘background’ evidence would commonly take the form of evidence of a defendant's previous but separate misconduct towards the same complainant. Although such evidence often impugned the defendant's character, it was not subjected to the same degree of scrutiny as character evidence offered on a ‘similar fact’ basis (Munday, 2018: 383).

This approach to ‘background’ character evidence led to abuse of the category. The characterisation of evidence as ‘background’ rather than ‘character’ avoided the exclusionary rule and allowed prejudicial evidence to be admitted on the basis of simple relevance (Munday, 2018: 387 and 411). This eventually lead to legislative reform in New Zealand (Evidence Act 2006) and like jurisdictions. ¹⁰

Propensity evidence under the Evidence Act 2006

New Zealand evidence law was codified with the enactment of the Evidence Act 2006. The ‘propensity rules’, set out in sections 40–43, govern the admissibility of what was known at common law as character evidence. The Supreme Court has noted that the Act ‘substantially codified’ prior case law on propensity evidence, but that moving forward, the focus should be firmly on the terms of the Act. ¹¹

Propensity evidence is defined (in section 40(1)) as:

…evidence that tends to show a person's propensity to act in a particular way or to have a particular state of mind, being evidence of acts, omissions, events, or circumstances with which a person is alleged to have been involved; but

does not include evidence of an act or omission that is—

one of the elements of the offence for which the person is being tried; or

the cause of action in the civil proceeding in question.

The Act further provides that propensity evidence about a criminal defendant, if offered by the prosecution, must have a probative value, in relation to an issue in dispute in the proceeding, which outweighs the risk that the evidence may have an unfairly prejudicial effect on the defendant (section 43). ¹²

The same section goes on to non-exhaustively list factors that may be considered in assessing probative value (section 43(3)). As noted by William Young J, this list was ‘plainly prepared with orthodox similar fact evidence in mind’. ¹³ The factors listed are: the frequency; connection in time; similarity; number of people making similar allegations; possibility of collusion; and unusualness. While useful when assessing the probative value of propensity evidence offered to support a ‘similar fact’ argument, these factors offer little assistance in the context of ‘background’ character evidence offered for other purposes. ¹⁴

Section 43 also provides that when balancing probative value and prejudicial effect, the ‘nature of the issue in dispute’ must be taken into account. ¹⁵ It has been stressed that the issue should be identified as precisely as possible. ¹⁶ (We return to the concept of ‘nature of the issue in dispute’ below, discussing how a coercive control lens can impact characterisation of the issue in dispute and, in turn, the admissibility of relationship propensity evidence).

Closing the loophole: ‘Background evidence’ as propensity evidence

As noted above, the Act defines ‘propensity evidence’ very broadly. Despite this, courts in New Zealand initially continued to sidestep the propensity rule in relation to evidence of a defendant's previous behaviour offered for ‘background’ purposes, including relationship propensity evidence. ¹⁷ This sidestepping was arguably evident in the Court of Appeal's approach in Mahomed v R. ¹⁸

The evidence in issue was evidence of uncharged misconduct that had occurred prior to the victim's death—the victim's parents left her unattended in a hot van (the ‘van evidence’). ¹⁹ The victim, a very young baby, was allegedly murdered by her father some weeks later. The Court of Appeal, in a similar vein to the common law treatment of ‘background’ evidence, twice declined to apply the Act's propensity rules to the van evidence. The court found that the ‘van evidence’ was offered to explain the dynamic of the defendant–complainant relationship, not to support a ‘similar fact’ argument. Instead, the van evidence was admitted on the basis of its ‘direct relevance’ for showing a ‘train of neglect’. ²⁰

The appeal from this decision led to what is now widely regarded as New Zealand's leading propensity evidence precedent—the Supreme Court decision of Mahomed. ²¹ The Supreme Court held that evidence of a defendant's separate misconduct toward the complainant is propensity evidence and subject to the propensity rule, irrespective of whether it is offered for ‘similar fact’ purposes or not. That is, the type of evidence known at common law as ‘background’ evidence is also subject to the heightened scrutiny of the Act's propensity evidence rules. As William Young J explained (at [61]):

The definition of ‘propensity evidence’ is exhaustive and there is no exclusion in relation to evidence of conduct between the defendant and victim. This means that evidence which tends to show a propensity by the defendant to act towards or think about the victim in a particular way is necessarily within the definition of ‘propensity evidence’ irrespective of why the Crown wishes to lead the evidence.

His Honour acknowledged that the section 43(3) factors would likely not assist greatly in assessing the probative value of such ‘relationship propensity evidence’, but noted that the same relevance factors which formed the basis for admission of such evidence as ‘background’ at common law could be invoked when weighing probative value under the section 43 propensity analysis. ²²

The effect of the Mahomed decision was to clarify that ‘relationship propensity evidence’ about prior behaviour of the defendant towards the complainant is included in the scope of propensity evidence, regardless of whether it is offered to invoke similar fact reasoning or for some other purpose.

Subsequent case law has highlighted the importance of relationship propensity evidence for providing background to intimate partner violent offending, while affirming it is still subject to the propensity rule: ²³

[E]vidence of other misconduct by a defendant against the same victim is not orthodox similar fact evidence. It does not rely upon standard ‘unlikely coincidence’ reasoning. Rather, the rationale for its admission rests on its supporting a conclusion that the relationship between the protagonists is conflictual and that the defendant deals with that conflict by expressing hostility and violence toward the complainant.

Despite the acknowledgement in Mahomed that such evidence is not typically offered to support an orthodox ‘similar fact’ argument, the Court of Appeal subsequently emphasised, in this context, that ‘[t]he crucial element is that the propensity evidence involves broadly similar offending against the same person when they are in disagreement’. ²⁴ As a result, the concept of ‘broad similarity’ has become a primary consideration when assessing the probative value of relationship propensity evidence.

Similarity in the context of domestic abuse: Discrete acts linked by a dynamic of coercive control

The enquiry into ‘broad similarity’ raises questions about what similarity means in the context of a relationship characterised by abuse, and what degree of similarity might be required to overcome the prejudicial effect of separate abusive incidents. In this section, we introduce Stark's theory of ‘coercive control’ as a useful framework for conceptualising apparently discrete and dissimilar (according to law's taxonomy) forms of violence and abuse within the context of an abusive relationship. The ‘coercive control’ framework conceptualises domestic violence as a pattern of coercive, controlling behaviour, rather than a series of separate events (as criminal law tends to do). This allows for individual incidents—events which may appear to be dissimilar in terms of form and immediate impacts, events which law classes as dissimilar based on actus reus (physical versus sexual violence)—to be understood as similarly pointing to an overarching relationship dynamic of dominance and control.

Coercive control

The criminal justice response to domestic abuse has historically been focused on acts of physical violence. In the 1980s and 1990s, feminist critiques highlighted that by focusing on individual acts of physical violence, the criminal justice system ‘obscured the nuances and complexity of domestic abuse’ (Burman and Brooks-Hay, 2018: 68–69). It is now well accepted that abuse in the context of intimate relationships is not limited to purely physical acts or threats of physical harm. Rather, there are various forms of abuse—including emotional, verbal, economic and sexual—which can be yielded in interrelated ways, based on an overarching pursuit of power and control.

This focus on power and control, rather than physicality, is at the heart of Evan Stark's (2007, 2009) formulation of ‘coercive control’, a framework that has been key in shedding light on the complexity of domestic abuse. Stark describes ‘coercive control’ as a pattern of behaviour used by perpetrators to control and dominate their intimate partner. Applying this lens to domestic abuse allows for fuller understanding of the various dimensions and methods of abuse, including non-physical elements which have often been underemphasised. In turn, this approach renders visible the broader ‘patterns and processes’ of abuse at play (Burman and Brooks-Hay, 2018: 69).

Stark's model is in part a critique of the previously prevailing paradigm of domestic violence, one which defined such abuse as incident-specific, focusing narrowly on physical and psychological assault, and assessing severity based on obvious injuries or trauma. This restrictive understanding of intimate partner violence resulted in disaggregation, trivialisation and normalisation of the oppression experienced by abused women (Stark, 2009: 1510).

The coercive control framework, by comparison, is one that recognises different types of violence as united by a common ultimate aim—to control and dominate one's intimate partner. In other words, coercive control is a ‘cumulative form of subjugation’, one that employs ‘a range of tactics’, not necessarily including nor limited to physical violence, to isolate and entrap women in relationships of control, fear and disempowerment (Burman and Brooks-Hay, 2018: 68; Otter and Bosanko, 2022). Ultimately, coercive control allows domestic abuse to be understood not as a series of discrete incidents, but as a pattern of abuse designed to subjugate and control (Candela, 2016).

Stark's model of coercive control has been highly influential. Growing recognition of coercive control as a possible underlying dynamic of abusive intimate relationships has led to targeted criminalisation in numerous jurisdictions in recent years, including Scotland, ²⁵ England and Wales, ²⁶ Ireland ²⁷ and New South Wales. ²⁸ There is no separate coercive control offence in New Zealand, although the Family Violence Act 2018 does state that violence may entail a pattern of behaviour, and may be coercive or controlling (section 9).

Coercive control and dominance feminism

Stark's theory of coercive control is situated in dominance feminism more generally. Dominance feminism, alongside ‘liberal feminism’, is one of the two most well-known and accepted feminist theories of equal rights (Becker, 1999: 32). Dominance feminism is particularly associated with legal academics, most notably the work of Catharine A MacKinnon (see, for example, 1987), and emphasises the need to eradicate female subordination (Baker, 2001). In the context of domestic violence, dominance feminism ‘brings into relief men's power over women’ by ‘suggest[ing] that men use violence, just as they use other forms of power, to perpetuate difference and thereby assert control over women’ (Baker, 2001: 1473 (footnote omitted)). As Malik and Lindahl (1998: 410) noted (over a decade before Stark set out his theory of coercive control), where domestic violence is present, ‘relationship power is a construct that is at once plainly involved’.

As a framework for understanding intimate partner violence, dominance theory allows for the recognition of different types of violence as stemming from societally-enabled norms of male dominance/female subordination which in some cases are replicated within a particular household/relationship. Stark argues that in relationships of coercive control, the male partner is able to exploit the institutional dominance granted to him by society to control his partner, thereby living up to misguided social expectations of male superiority/female inferiority. In other words, intimate partner violence can be motivated by an urge to regain control as the dominant partner in light of increased gender equality more broadly (Stark, 2007: 171). Dominance theory, then, goes hand in hand with an understanding of domestic violence as gender-based; it situates domestic abuse ‘within the context of women's (and girl's [sic]) status in society, taking into account norms, social structures and gender roles, which greatly influence their vulnerability to violence’ (McFeely et al., 2013: 2 (citations omitted); see, further, Malik and Lindahl, 1998: 416–417).

Coercive control vs situational violence

The coercive control framework leads to a nuanced approach to the question of ‘gender symmetry’—whether domestic abuse is equally experienced and perpetrated by men and women (Dobash and Dobash, 2004). Intimate partner abuse is not, of course, carried out exclusively by heterosexual men. Indeed, there is evidence that both men and women are at risk of experiencing violence within intimate partnerships, and some evidence suggests that the risk is somewhat symmetrical (McFeely et al., 2013).

Michael P Johnson (2008), in response to the gender symmetry debate, devised a typology of domestic violence: (1) intimate terrorism; (2) violent resistance; and (3) situational couple violence. ‘Intimate terrorism’, described as using a range of tools to control, corresponds broadly with Stark's conception of ‘coercive control’ (Myhill and Hohl, 2019: 4494, n 1), and ‘is overwhelmingly perpetrated by men against women in heterosexual relationships’ (McFeely et al., 2013: 3). On the other hand, men who report abuse within a relationship are more likely to have experienced ‘violent resistance’ (overwhelmingly used by women to resist intimate terrorism) or ‘situational couple violence’ (verbal or physical violence in response to a discrete disagreement or conflict) (McFeely et al., 2013: 3).

The distinction between coercive control or intimate terrorism on the one hand, and situational violence on the other, is important in the context of relationship propensity evidence. Incidents best characterised as ‘situational violence’ may be more appropriately admitted on a ‘similar fact’ basis, whereas incidents that point to a pattern of intimate terrorism or coercive control are more likely to be admissible on a ‘relationship dynamic’ basis, a point we return to below.

Discrete acts as similar expressions of power and control

Despite the widespread influence of the coercive control model, researchers have noted that there is a still a deep-seated tendency to characterise domestic violence as situational and physical (Myhill and Hohl, 2019). Indeed, in relationships involving ‘situational violence’, to return to Johnson's typology, this may be an appropriate lens. But in many abusive relationships, a coercive control lens will be needed. Indeed, Myhill and Hohl have argued that coercive control should be the ‘golden thread’ running through the criminal justice response to domestic violence, allowing police (and by extension, judges and juries) to move beyond an ‘incident by incident’ approach and instead understand discrete events as evidence of longer-term patterns of behaviour.

As the case studies below will indicate, it will frequently be the case that where the prosecution seeks to offer relationship propensity evidence, the specific acts involved are categorised, in law, as ‘dissimilar’. This is because the law categorises violent acts based on actus reus, with varying offences based on varying specific acts or impacts (sexual violence; financial abuse; psychological abuse; physical violence). But in some relationships, there will also be indicators that these discrete acts are part of a pattern of ‘intimate terrorism’ or ‘coercive control’, in that they represent diverse methods of achieving a single aim: to control. Where such indicators are present, the research on coercive control as a pattern suggests that it is artificial to delink those diverse methods based on insufficient similarity. By applying a ‘coercive control’ lens to relationships of domestic violence, the overarching dynamic of dominance, control and terrorism should factor into the assessment of whether different ‘types’ of violence are ‘broadly similar’.

Delinking non-sexual and sexual violence propensities within a relationship: Two case studies

We turn now to consider two recent Court of Appeal cases, S v R and H v R. In each case, relationship propensity evidence of previous non-sexual violence between the defendant and complainant was ruled inadmissible to support charges of sexual violence, conclusions we submit were incorrect and reflect a problematic conceptualisation of violence in intimate relationships. While it is not suggested that these two cases are a representative sample, they do expose specific instances where the nature of the issue in dispute has been mischaracterised, resulting in an unsatisfactory delinking of sexual and non-sexual violence.

As seen through the analysis in these two cases, the Court of Appeal considered that the state of mind required to inflict physical violence was different to the state of mind required to inflict sexual violence. While this may be correct in some instances, we submit that a different approach may frequently be needed in the context of intimate partner violence, in light of the possibility of an underlying dynamic of coercive control.

Before turning to the cases, a note on the charges at issue—both concerned various sexual violation charges. Sexual violation is defined in section 128 of the New Zealand Crimes Act 1961, and the issue of consent is central to both actus reus and mens rea. To prove actus reus, the prosecution must show there was not consent to the connection; to prove mens rea, they must show that the defendant did not believe on reasonable grounds that the complainant consented. Consent is not defined in the Act, but has developed in case law to mean ‘full, voluntary, free and informed consent’ (High, 2022: 310).

S (CA145/2020) v R [2020] NZCA 362

The defendant, S, had been convicted of 16 charges of violent offending against his wife, M, including a representative charge for rape. When the appeal against the representative rape charge was allowed, ²⁹ the question arose as to whether the remaining 15 violence convictions could be led as evidence at the new rape trial. The propensity evidence was ruled admissible by the District Court. ³⁰ On appeal, the Court of Appeal ruled the evidence inadmissible, due to the absence of a logical connection between the non-sexual and sexual violent offending.

Facts

S and M were married in 2008 and had a son together. From about three years into their marriage, S was ‘always demanding sex’ and would get angry if denied. S would force M to have sex with him when she said no, in ways she did not want, and she described sex as a punishment. M also described being scared at night, when S would watch pornography and require M to learn from and mimic the pornographic moves. In her brief of evidence, M noted that she thought she had to do what S wanted, that S had told her it was her wifely duty to have sex, and that it was Chinese tradition for a wife to be subordinate to her husband: ‘China men are very superior at home, he's the boss, he can get what he wants’. ³¹ After their son was born, M would sometimes say yes to sex to stop his threats to leave her.

S also inflicted non-sexual physical violence against M and their son. Importantly, the instances of non-sexual violence were not closely connected in time with the alleged sexual offending. The defendant's convictions for violence related to stomping on M's stomach and legs, throwing her to the ground, punching her, kicking her and pointing a gun while threatening to kill M and their son.

Rulings on propensity evidence

In the District Court, Judge EM Thomas found that the proposed conviction evidence went to ‘concepts of power and dominance and control within the relationship’. ³² The evidence was assessed as having a high probative value because it shed light on a ‘marked power imbalance that is material to consent’; as such, excluding the violent offending would be artificial. ³³ The risk of unfair prejudice would be outweighed by the probative value of the evidence, and it was deemed admissible.

In the Court of Appeal, the defendant submitted there was ‘no logical connection’ between the violent offending and the alleged sexual violation. ³⁴ The defendant emphasised that M did not say she acquiesced to sex due to fear of violence, but because she thought the nature of the sexual activity was ‘normal’. To admit the violent offending would add an illegitimate element to the question of consent, one that would overpower the actual evidence given by M as to why she consented.

The Court of Appeal sided with the defendant, ruling the propensity evidence inadmissible. ³⁵ The analysis focused on whether there was sufficient ‘linkage’ between the non-sexual and sexual violence. On linkage, the court referred to one of Mahomed's most oft-cited passages: ³⁶

…the propensity [must] have some specificity about it. That specificity, in order to be probative, must be able to be linked in some way with the conduct or mental state alleged to constitute the offence for which the person is being tried.

The court concluded there was not a ‘cogent connection’ between S's tendency to act violently towards M, and the alleged sexual offending. ³⁷ Crucially, the court considered that evidence of physical violence was not ‘capable of demonstrating active hostility towards M, leading to sexual violation of her’. ³⁸

In reaching this conclusion, the court relied in part on the failure of the complainant to herself draw a link between the non-sexual and sexual violence. In her brief, M attributed her acquiescence to sexual contact to certain cultural expectations pertaining to the position of a husband over his wife, as opposed to a fear of physical violence. As M had not identified a link, the court held it should not be open to the jury to infer such a link. ³⁹ Any such inference would be based on the assumption that ‘physical violence begets sexual violence’, which is tantamount to reasoning guilt based on prior convictions alone. ⁴⁰

The court did accept that the propensity to relationship violence might bear on the issues of consent or reasonable belief in consent. ⁴¹ The court was unwilling to admit the propensity evidence on this basis, given that M's brief did not allow for a link to be drawn between the physical and sexual offending. Notably, the court's discussion here focused on M's understanding of her own reasons for acquiescence, but did not further engage with the question of whether despite M's failure to link her ‘consent’ with the physical violence, the context of violence could nonetheless go to whether her husband held an honest reasonable belief in consent.

H (CA595/2021) v R [2022] NZCA 4

The defendant, H, was on trial for charges arising out of two separate instances of alleged offending, one involving both physical and sexual offending, and the other only non-sexual offending. The District Court granted the Crown's application for the two sets of charges to be joined and heard together. ⁴² On appeal, the Court of Appeal found the two sets of charges were not cross-admissible propensity evidence and should be heard separately. ⁴³

Facts

The complainant, W, obtained a protection order against H in 2019. In April 2020, H and W were at the complainant's house. She was feeding their baby when H entered the room and asked for sex, which she declined. An argument followed, during which W attempted to push and kick H away. H took the baby from W, put the baby in a cot, told W to turn over and proceeded to have forceful non-consensual sex with W. Following intercourse, W picked up the phone to call police, but H grabbed and damaged it. H was charged with sexual violation by rape, breach of a protection order and wilful damage.

The second set of charges occurred approximately one year later. H and W had been at the Family Court; afterwards, W agreed to go H's home. They sat in a car outside his home for about an hour. When W said she needed to leave, H yelled at her to go inside and physically assaulted her. H was charged with breaching a protection order, three charges of assault on a female, and strangulation.

Rulings on propensity evidence

The trial judge held the charges were cross-admissible and could be heard together, as each set of offending provided context and narrative for the other set, and were each highly probative as to the complainant's credibility. ⁴⁴ The judge also determined that the complainant's evidence would be incomplete if the charges were not cross-admissible, and the jury would be deprived of the full picture of the relationship—it would be ‘artificial to present the two alleged incidents separately’. ⁴⁵

The decision was reversed in the Court of Appeal, on the basis that the two sets of offending were not ‘broadly similar’. ⁴⁶ The court reasoned that there was no evidence that the sexual violence was accompanied by or associated with non-sexual violence. ⁴⁷ As in S v R, this had the effect of delinking the physical and sexual offending as discrete and unrelated acts.

Discussion

Both case studies adopt the post-Mahomed approach of subjecting relationship propensity evidence to the propensity rules, even though it is offered for a purpose other than to support a ‘similar fact’ argument. This is undoubtedly correct. However, the principle of ‘broad similarity’ is, we submit, liable to be applied too strictly, as the case studies illustrate. In both cases, the Court of Appeal failed to perceive a sufficient connection between the physical and sexual offending. This effective delinking of sexual and non-sexual violence treats them as discrete, separate, unrelated acts, overlooking their inherent connectedness as expressions of coercive control and dominance.

Both cases suggest an understanding of physical and sexual violence as not ‘broadly similar’, in that they are motivated by different ‘types of hostility’. The Court of Appeal in S v R expressly distinguished the type of hostility that leads to sexual violation from the hostility that underpins physical violence. ⁴⁸ In some contexts, this rings true. For example, a man who punches a random stranger at a bar may not be more likely than another to also go home and sexually assault his partner. However, in the context of prolonged violence and control inflicted on an intimate partner, the coercive control framework suggests that physical and sexual violence can in fact be very closely linked, in that they separately contribute to an overarching pattern of coercively controlling behaviour (and regardless of whether a complainant herself is cognisant of that linkage). This understanding comes through very clearly in the reasoning of Judge EM Thomas in the District Court decision in R v [S], with his express reference to power, dominance and control in the admissibility analysis. ⁴⁹ His reasoning recognises that there may be a specific type of hostility at play within an abusive relationship, regardless of the form by which such hostility is expressed. This is a hostility characterised by the desire to control and dominate. If distinct expressions of violence are characterised in this way, as linked by an underlying propensity to control, then it is but a small step to recognise those expressions as ‘broadly similar’. Another way of putting this is to say that distinct expressions of violence demonstrate a propensity to have a particular state of mind: an intention to coercively control.

In the context of intimate partner violence, seemingly dissimilar offending should, then, properly be treated as alike if discrete events are underpinned by a common dynamic of dominance, control and coercion. In other words, assuming there is a common underlying dynamic that can be said to characterise the relationship, all acts that perpetuate that dominance should be understood as ‘broadly similar’, even if the specific actions are not similar in a ‘similar fact’ sense. Categorising violence by type, rather than by reference to the underlying relationship dynamic, slips into an orthodox ‘similar fact’ analysis, which is to miss the point of offering relationship propensity evidence.

Importantly, the delinking of physical from sexual violence is likely to particularly impact sexual violence charges, and particularly the assessment of whether a defendant held an honest and reasonable belief in consent. For example, in S v R, M was unable herself to identify a linkage between her acquiescence and her husband's prior violence, which might have had a bearing on whether the ‘no consent’ element could be proven. But the physical violence would surely be relevant to her husband's state of mind when he demanded submission from her to sexual contact, regardless of whether the complainant in her own mind could identify a linkage. By excluding the narrative of prior physical violence, the jury would be given a sanitised and incomplete picture of the relationship dynamic, a dynamic which bears strongly on the assessment of mens rea.

Linking non-sexual and sexual violence propensities: Two case studies

In this part, we consider how viewing non-sexual and sexual violence, occurring in the context of an intimate partner relationship, through a broad coercive control lens can impact the relationship propensity evidence admissibility analysis. To illustrate this, we explore two recent Court of Appeal cases where evidence of non-sexual violence was admissible in relation to charges of sexual violence.

Pahi v R [2021] NZCA 348

The defendant, Mr Pahi, appealed his convictions for sexual violation on the grounds, inter alia, that the District Court erred in admitting propensity evidence of previous non-sexual violent offending towards the same complainant. ⁵⁰ The Court of Appeal dismissed the appeal.

Facts

Mr Pahi and the complainant, who had four children together, had been in an ‘on and off’ relationship for close to 10 years. During an ‘off’ period, on 17 May 2016, while visiting the children at the complainant's house, he refused to leave when asked. In the ensuing argument, Mr Pahi physically assaulted the complainant, including by way of strangulation. There were injuries to her neck, arms and legs, and her blood was found on her wall. Mr Pahi pleaded guilty to a charge of injuring with intent to injure arising out of this assault.

Immediately after the assault, Mr Pahi continued to verbally abuse the complainant. He then drove her to pick up her son from basketball. On their return, he again entered the house and started yelling at her; this was one to two hours after the earlier physical assault. He then began undressing her. She told him to stop, but he said he ‘need[ed] it’, and later, ‘just fucking let me do this and then…I’ll go home’. He proceeded to perform oral sex on her, and had sexual intercourse with her, which formed the basis of the sexual violation charges. At trial, Mr Pahi said the sexual activity was consensual ‘make-up sex’. The jury was unable to reach a verdict, but he was convicted on both sexual violation charges at his retrial.

At the retrial, the Crown was permitted to lead propensity evidence in the form of three previous convictions for violent non-sexual offending against the same complainant. The previous convictions related to violent non-sexual offending against the same complainant on two prior occasions (July 2013; July 2014). ⁵¹ On the first, Mr Pahi, who was at the complainant's house uninvited, refused to leave when asked and ripped off her necklace and squeezed her neck when she said she no longer wanted to be in a relationship with him. On the second, Mr Pahi threatened the complainant with a baseball bat when she denied having taken cash from his wallet, then punched her, dragged her by the hair and stomped on her ribs.

Ruling on propensity evidence

The trial judge ruled that the convictions were admissible, because the history of prior violence was ‘highly relevant’ to the questions of consent and reasonable belief in consent. ⁵² On appeal, the defence submitted that previous convictions for violence, ‘without more’, could not be propensity evidence on subsequent charges of sexual violation, being unconnected and therefore ‘irrelevant and extremely prejudicial’. ⁵³

The Court of Appeal rejected this argument, noting that in the context of a violent family relationship, such propensity evidence is more significant for what it says ‘about the dynamics of the relationship’ than for its similarity with the index offending. ⁵⁴ The court reasoned that the history of Mr Pahi's ‘controlling behaviour’, including his use of violence, was highly probative both as to consent and as to reasonable belief in consent. ⁵⁵ In particular, it went to the complainant's credibility in asserting that she was afraid of a further ‘hiding’ if she resisted his sexual advances any more actively than she did. ⁵⁶

This latter point suggests that the evidence was admitted in part on the complainant's identification of a link between her fear of violence and her acquiescence to sexual activity, a point of distinction from S v R (in which, as discussed, it seems the complainant's failure to make that connection herself was fatal). It is unclear whether the propensity evidence in Pahi would have been admitted in the absence of that self-identified linkage. However, the court's reference to a pattern of ‘controlling behaviour’ is certainly congruent with the coercive control framework.

Holmes v R [2022] NZCA 340

In Holmes, the Court of Appeal held that relationship propensity evidence of non-sexual violence by the defendant against a former partner was admissible in relation to charges of sexual violence against the defendant's current partner. The court reasoned that the propensity evidence showed a propensity to behave in a dominant and controlling way in intimate relationships, which was relevant to both consent and reasonable belief in consent.

Facts

Mr Holmes was convicted of multiple charges against his partner, the complainant, including for both non-sexual and sexual violence. Mr Holmes appealed his convictions for the sexual offences, one of the grounds being that evidence of his offending against a former partner should not have been admitted as propensity evidence.

Mr Holmes and the complainant, referred to as ‘Rose’, entered a relationship in mid-2015, around the time Mr Holmes was ending his relationship with ‘Sally’. Rose had four children, and she moved into Mr Holmes’ house with her two youngest children at the end of 2015. Mr Holmes started to emotionally control Rose. He demanded sex almost daily, and belittled her if she refused. His control and abuse of Rose included verbal abuse, controlling behaviour (including demands on how she dressed and ate, isolation from family and friends and monitoring of her phone) and verbal threats directed towards both Rose and her children. Except for one occasion where Mr Holmes bit Rose's ear during an incident of sexual violence, there were no allegations of physical abuse.

There were multiple incidents of sexual violation. The ‘bathroom incident’ involved forced sexual intercourse, during which the defendant aggressively bit the complainant's ear, digitally penetrated her and raped her while ignoring Rose's protests. The ‘lounge incident’ involved the defendant digitally penetrating the complainant despite her protests; when Rose continued to protest, Mr Holmes threatened to leave her. Eventually, Rose became so worn down by the abuse and felt powerless, such that she decided not to resist the sexual intercourse. She feared that if she did continue to resist, she would have to endure ‘hours longer of his emotional and mental torment’. ⁵⁷ The ‘hospital incident’ occurred when Mr Holmes forced Rose to have sexual intercourse with him days after she had given birth to their baby. In addition to these incidents, Mr Holmes was convicted of a representative charge of rape that covered the 15-month period that Rose lived in Mr Holmes’ house.

At trial, Judge McDonald admitted propensity evidence from Mr Holmes’ previous relationship with ‘Sally’. Sally reported that Mr Holmes was often physically violent towards her, and would verbally abuse her in a similar nature to his verbal abuse of Rose. Mr Holmes was charged on a number of counts for physically assaulting Sally; when his application to sever the Sally charges from the Rose charges was dismissed on the basis that the evidence was cross-propensity evidence, he pleaded guilty to the Sally charges prior to the Rose trial.

Ruling on propensity evidence

In the District Court, Judge McDonald ruled that the Sally summary of facts should be admissible at the Rose trial. ⁵⁸ He accepted the Crown's submission that the evidence showed a tendency to ‘use techniques to totally dominate and control both women’, ⁵⁹ and was relevant to whether Rose had given true consent to sexual activity.

On appeal, counsel for Mr Holmes submitted that there were ‘profound differences’ between Sally and Rose's allegations. ⁶⁰ In particular, while Sally experienced physical violence, this was not a dominant feature of his relationship with Rose.

The Court of Appeal began by identifying the issues in dispute, which were (a) whether the sexual acts in issue took place; (b) if they did, whether Rose consented; and (c) if she did not consent, whether Mr Holmes knew she did not or lacked a reasonable belief in consent. In relation to Sally's evidence, the central issue was whether her description of Mr Holmes’ verbal abuse and degrading and controlling behaviour had probative value in relation to the second and third of these issues. ⁶¹

Importantly, the Court of Appeal did go through the factors in section 43(3) to assess probative value, despite those factors having more salience in relation to ‘similar fact’ propensity evidence. But when it came to analysis of ‘similarity’, the court took a broad approach, as is appropriate for relationship propensity evidence (distinguished from ‘similar fact’ propensity evidence). While acknowledging that Sally had not experienced sexual offending, the court noted that both Sally and Rose had experienced violence, because ‘sexual offending is violent offending’ and ‘[f]orced sex is a serious form of violence’. ⁶² The court further noted, in terms of similarity, that Mr Holmes had strikingly ‘tormented and dehumanised’ both women, using ‘similar tactics and language to belittle them and to diminish their senses of self-esteem and autonomy’. ⁶³ As such, while their physical experiences differed, their relationships shared a ‘remarkable feature’, that being ‘the way he undermined the dignity and autonomy of both women by using very similar techniques’. ⁶⁴ Relatedly, the court characterised the respective offending as unusual with respect to the ‘level and degree’ of domination, and the methods used to ‘wear down their sense of dignity and autonomy’. ⁶⁵

Discussion

Pahi and Holmes provide a counterpoint to S v R and H v R, in that the court in each of the former two cases took a broader and more flexible approach to similarity in the context of relationship propensity evidence. Rather than characterising physical and sexual violence as dissimilar, the courts implicitly recognised that in some cases, they represent the same underlying relationship dynamic of coercion and control.

In both Pahi and Holmes, the court also based this linkage in part on evidence from the complainant herself that suggested a pattern of control and domination. So, in Pahi, the complainant expressly linked her submission to unwanted sexual contact with her fear of being further physically assaulted by the defendant. Likewise, in Holmes, the complainant testified that she felt unable to resist the defendant's demands for sex ‘because of the way he controlled and dominated her’ ⁶⁶ and because of her fear of ‘further emotional and psychological abuse’. ⁶⁷ By comparison, in S v R, the complainant's inability or unwillingness to link her submission with the broader relationship dynamic of coercion was seemingly fatal to the admission of non-sexual propensity evidence. This latter approach represents a restrictive approach to discrete forms of partner abuse, failing to recognise that they might be linked by a propensity to dominate, regardless of whether a complainant herself identifies such a linkage.

Revisiting relationship propensity evidence through a coercive control framework

In this section we discuss the need for judges to apply a coercive lens to ‘dissimilar’ violent offending in appropriate cases, resulting in a more permissive approach to the probative value analysis. We also consider whether the Evidence Act 2006 provisions on propensity evidence about a defendant, which were developed with similar fact propensity evidence in mind, might provide insufficient guidance for judges considering relationship propensity evidence not offered to support similar fact reasoning.

While case law provides some guidance here, the studies presented herein point to inconsistency at the appellate level as to the treatment of relationship propensity evidence. Further, even in cases which have arguably adopted a coercive control lens, such as Pahi and Holmes, the courts at times seem to still require much of complainants in terms of self-identification of linkage, an issue which raises concerns relating to the essentialisation and agency of domestic violence complainants.

Finally, we discuss how case law has arguably developed in a confusing and haphazard manner here, in part because of the application of section 43 ‘similar fact’ logic to propensity evidence offered for other purposes, such as to illustrate the dynamic of a relationship. We highlight that the factors in section 43 are non-mandatory considerations, and as the Supreme Court has noted, not of particular assistance in relation to relationship propensity evidence. ⁶⁸ As such, the analysis of relationship propensity evidence admissibility should be focused on a broad and permissive assessment of probative value, on a case-by-case basis. It would be helpful for Supreme Court jurisprudence on this point, given the Mahomed decision involved a split on relationship propensity evidence, and given the confusion and inconsistency we have highlighted herein.

Defining the issue: A propensity to control and dominate

To return to the case studies: as discussed, in S v R the Court of Appeal ruled the propensity evidence of non-sexual violence against the complainant was inadmissible in relation to the charges of sexual violence against the same complainant. The court considered there was no ‘cogent connection’ between the defendant's tendency to inflict non-sexual violence and the state of mind required for sexual violation. ⁶⁹ Further, the court considered the probative value of the physical violence related to whether the defendant had an attitude of hostility leading to physical violence towards the complainant. ⁷⁰ While the evidence of non-sexual violence indicated such an attitude of hostility, this was viewed by the court as too remote from the question of hostility leading to sexual violence. ⁷¹

There is friction here, as the issue in dispute must be identified as precisely as possible, which lends itself to a narrow framing of probative value, such as a ‘proclivity to exercise power and control…in a physically violent way’. ⁷² However, this is arguably the wrong approach in relationships of coercive control, in which a wide range of behaviours may be closely and cogently connected by the underlying dynamic of dominance. As such, framing the issue in dispute as whether there was hostility between the protagonists of the type that results in physical violence creates an artificial delinking of the sexual violence and the non-sexual violence. This result arose because the court considered that the state of mind required to inflict physical violence was different to that needed to inflict sexual violence. But as we have argued, in the context of domestic violence, different episodes and methods of violence are frequently linked by the overarching theme of coercive control. In relationships of coercive control, the state of mind is to control, and the various forms of violence are expressions of that intention. In other words, in a relationship that is characterised by dominance and coercion, all acts that perpetuate that dominance and control are ‘broadly similar’, even if they do not necessarily appear that way in their manifestation (or by reference to traditional common law notions of similarity as applied to character evidence).

Instead of framing the propensity evidence as relevant to the issue of hostility leading to physical violence, as was the approach in S v R and H v R, courts should consider whether non-sexual violence points towards a relationship of dominance and control on the part of the defendant. That is, is the relationship one of ‘intimate terrorism’, as opposed to ‘situational violence’ (to return to Johnson's typology)? Framing the issue in this way allows the relationship to be viewed in a more appropriate light, as the factfinder is exposed to the true nature of the relationship as one characterised by various forms of violence. This in turn is important for the assessment of the elements of sexual violence: was there consent, and did the defendant reasonably believe there was? In other words, only by including a full picture of the dynamic of the relationship, one characterised by a propensity to dominate and control, including through physical violence, can the jury have true insight into whether the complainant was consenting to sexual activity, and, in light of how the defendant treated the complainant, whether he could reasonably have believed she was consenting.

Linking sexual and non-sexual violence: The onus on complainants

Notably, in S v R, the court appeared to suggest that if the complainant had herself identified a linkage between her acquiescence to sex and a fear of violence, that might have changed the analysis of probative value and admissibility. Indeed, in Pahi, which had similar facts to S v R, when the complainant made such a linkage, the physical and sexual offending was seen as sufficiently, cogently connected, despite being different in nature. This, we submit, is unsatisfactory; courts should be willing to recognise the similarity and cogency of physical and sexual offending as indicative of a relationship characterised by coercion, regardless of whether complainants self-identify a linkage, for the reasons canvassed above. Further, requiring a complainant to link sexual and non-sexual violence places a heavy burden on the complainant to articulate her motivations for acquiescence.

This discussion of self-identified linkage can be situated in the literature on domestic violence and agency. There is a notable tension, in the context of intimate partner violence, between taking women's subjective experiences seriously and understanding that women's subjectivity is shaped by the patriarchal world around them (Baker, 2001: 1471). When a complainant denies that there is linkage between her submission to sexual violence, and her fear of physical violence in the relationship more broadly, this tension arises. Does respect for agency require that we take this denial as conclusive proof the two are not linked? Or should judges recognise that there may, despite such denial, be strong evidence that she is trapped in a cycle of subordination and control which has impacted her response to sexual advances? As Baker (2001: 1465 (footnote omitted)) has noted:

Battered women are neither free agents as the law traditionally construed that term, nor victims incapable of acting on their own behalf. The reality lies somewhere in between.

This concern for agency relates also to the question of essentialisation. As Baker (2001: 1464) also notes, battered women are not a homogenous group, and their strategies, reactions and experiences in abusive relationships vary significantly. Returning to Johnson's tripartite framing of domestic violence, it is important to note that not all relationships of violence are properly characterised as involving ‘intimate terrorism’ or coercive control, and there are different degrees of control and terrorism at play in different relationships. In cases where there is clear and convincing evidence of coercive control as an underlying linking dynamic, we have argued that it is appropriate for judges to take this into account in assessing the cogency of connection between propensity and index offending. In some cases, such evidence will be entirely lacking, and a ‘similar fact’ analysis may be more appropriate. But it is inappropriate, in our view, to look solely to whether a complainant has self-identified a dynamic of coercive control as linking physical and sexual violence, as the court seemed to do in S v R; a complainant's own subjective perception should not necessarily be the be-all and end-all of the analysis. Even if a complainant is not herself identifying a linkage between the physical violence and her acquiescence, that is not to say that the physical violence can have no bearing on the ‘no consent’ element. This argument is supported by research suggesting that in relationships of prolonged abuse and control, a complainant may not necessarily consider a sexual experience as rape, viewing a non-consensual sexual act as less problematic than the physical violence also present in their relationship (Lilley et al., 2023: 7; New Zealand Ministry of Justice, 2015: 129). Emerging research indicates this is particularly so when there is an absence of physical force or violence during the non-consensual sexual act (Lilley et al., 2023: 7; Kirkman et al., 2021).

Moreover, relying on a complainant to identify the link between physical violence and their own consent/non-consent ignores that such violence might also be relevant to the other key element of sexual violence: the defendant's reasonable belief in consent. For example, on the facts of S v R, it seems clear that a jury informed of the broader context of regular non-sexual violence might reasonably take that into account in assessing whether the defendant could have held an honest and reasonable belief in consent, in light of his attitude and behaviour towards the complainant generally.

Applying propensity rules to relationship evidence: More guidance needed

The case studies suggest that further guidance might be necessary for judges applying the section 43 propensity analysis to relationship propensity evidence; it would be helpful to have the Supreme Court revisit the issue of relationship propensity evidence. The current approach is to consider whether there is ‘broadly similar’ offending evident in the propensity evidence and index charges. However, given the legacy of the common law's ‘strict similarity’ analysis of orthodox ‘similar fact’ propensity evidence, and given further that the section 43 probative value factors relate closely to similarity in a strict sense, it is unsurprising that in some cases, such as S v R and H v R above, the courts seem to have taken an unduly restrictive approach to ‘broad similarity’.

Importantly, the Supreme Court minority opinion in Mahomed highlighted that the section 43(3) factors were ‘plainly prepared with orthodox similar fact evidence in mind’. ⁷³ There is no guidance in section 43 for the assessment of the probative value of relationship propensity evidence. The concern is that, given this lack of guidance, and given the Court of Appeal's statement that relationship propensity evidence requires ‘broadly similar offending’, ⁷⁴ in some cases the section 43(3) factors may result in ‘similar fact’ analysis seeping into the assessment of relationship propensity evidence.

Rather, we submit that the idea of ‘broad similarity’ is best avoided when it comes to relationship propensity evidence; it is a distracting and confusing formulation in this context, and it is not a term found in the legislation. Rather, the focus should simply be on ‘probative value’, which is both the term used in the statute, and aligns with the common law's approach to ‘background’ character evidence. It would be helpful to have the Supreme Court revisit the assessment of probative value of relationship propensity evidence in order to clarify the relevance of ‘broad similarity’, and to set out further guidance for judges given the unsuitability of the section 43(3) factors to this end.

Conclusion

The common law has historically given men the benefit of assuming a woman is consenting to sexual contact based on her previous behaviour—for example, her agreement to sex with the defendant or with other parties on separate occasions, her reputation for promiscuity. Conversely, the law has historically been less inclined to allow women to explain their reaction to unwanted sexual contact by reference to the defendant's previous behaviour. For example, before the ‘resistance requirement’ was abolished, a woman's failure to resist could be fatal to a rape conviction, unless there was evidence the failure was due to a reasonable fear of violence. However, reasonableness was in many cases interpreted as requiring a complainant's fear to be connected with present and imminent threats, rather than previous violence or a broader context of abuse. ⁷⁵

The resistance requirement is no longer a feature of rape doctrine in New Zealand, but as a matter of credibility, a woman's behaviour during unwanted sexual contact is still very much under examination in rape trials (McDonald, 2020, 2023). Where sexual violation occurs within a relationship characterised by coercive control and abuse, it paints an incomplete and therefore artificial and misleading picture for juries if they are not made aware of that broader relationship dynamic. Courts should be slow to exclude relationship propensity evidence that explains the underlying dynamics contributing to acquiescence and submission in the context of intimate partner sexual offending. If physical and sexual violence are delinked as too dissimilar to be cross-admissible as evidence of a defendant's propensities within a relationship, then the complexities of abuse patterns that have unfolded within such a relationship can be obscured. Very often, such relationships, and the reactions and responses of complainants caught therein, are only properly understood when the broader dynamics of dominance, coercion and control are before the jury, as opposed to excluding disparate methods of achieving that dynamic.

In the context of intimate partner violence, discrete acts of abuse and control may differ with respect to the specific acts, omissions and circumstances involved. But where there is evidence of a broader power dynamic of coercive control or ‘intimate terrorism’, the perpetrator is in control, and their abusive acts will tend to reinforce this power dynamic, regardless of what specific methods of abuse are employed. As such, the dynamic of coercive control is what provides sufficient linkage between propensity evidence and index charges. By failing to apply a coercive control lens in appropriate cases, judges can inappropriately and artificially categorise different forms of violence as discrete, insufficiently similar, acts. But, as we have argued, it is vitally important that the law takes the ‘propensity to control’ seriously. Rather than slipping into an orthodox ‘similar fact’ analysis, a broad and flexible approach to probative value is needed in this context.