Executive summary: reducing and preventing violence against women

Introduction

Overview and purpose

The review:

• aimed to synthesise global and Australian evidence on prevention interventions to identify whether – and under what conditions – they can reduce rates of perpetration and/or victimisation of violence against women (referred to as ‘impact’)
• examined evidence on combining interventions to mutually-reinforce each other for greater impact (known as a ‘saturation’ approach), and on how this might extend beyond those directly engaged to the broader population
• will inform a multi-year project to develop a place-based ‘saturation model’ for prevention of violence against women, led by Respect Victoria, and gave particular attention to place and community-based approaches.

Background and context

• Victoria has built an enabling policy environment and infrastructure to support place-based approaches (footnote 1). Previous projects have built foundational evidence for how a place-based saturation model might be designed and implemented (footnote 2), and many process, output and outcome evaluations of prevention activity in Victoria attest to their strength.  
• While globally a growing number of evaluations of prevention of violence against women interventions have demonstrated impact (reference 1), there is limited evaluation and research evidence demonstrating the impact of Victorian prevention work on the severity and prevalence of violence.  
• Further, to achieve and sustain impact across a broader population, research indicates that individual interventions are not enough (reference 2-4). There is a need to better understand how to combine and scale interventions to measurably reduce and prevent violence at the population level, and the conditions, factors, and processes that enable and sustain reductions.

Guiding questions

The following questions guided this review:

1. What does recent research and practice tell us about the effectiveness of prevention interventions, in terms of impact on rates of perpetration and/or victimisation of violence against women?
2. What do we know about the foundational conditions, variables or criteria that affect the extent of prevention practice impact?
3. What do we know about how and whether outcomes from individual interventions are strengthened because of how they interact when coordinated with other interventions, and what design, implementation or contextual conditions contribute to any ‘mutually-reinforcing effect’?

Review scope

The review:

• focused on the impact of prevention interventions – defined as reductions in victimisation and/or violence perpetration against women, whether with specific intervention groups or at the population level
• did not analyse in detail other outcomes of prevention activity – especially reductions in the drivers and reinforcing factors of violence – which are equally important, and likely precursors (reference 5) to such impacts
• included studies from low, middle and high-income countries internationally, and gave particular attention to place- and community-based approaches employing multiple components, and their impact on rates of violence against women
• included a range of interventions designed to prevent new violence against women, reduce recurring violence, and/or reduce risk of victimisation or perpetration (regardless of how source material described the type of intervention, as varying terminology is used).

Methodology

• This review was a scoping study, synthesizing evidence from a range of sources that met inclusion criteria based on the questions and scope above.  
• The process included a desk review of primary research and evaluations, meta-analyses and evidence summaries, non-empirical evidence, and ‘grey’ literature on practice evidence, accompanied by discussions with Respect Victoria staff, prevention sector stakeholders and other key informants.
• The review did not include studies relying on qualitative data only, nor quantitative evaluations measuring other types of ‘impact’ (to that defined in the Scope, above), such as of attitudinal, practice or normative shifts (reference 6) (footnote 3).  

Assumptions and limitations

• This review did not attempt to replicate the search and analysis processes of recent comprehensive or rigorous reviews, and instead analysed their findings in the light of current research objectives and questions. These were supplemented by primary research and evaluations that emerged after those reviews were published. It is possible some studies have been missed.
• Most source material focused on men’s physical and sexual intimate partner violence and non-partner sexual violence against women. This material, in most cases, did not examine in detail the impact of interventions on emotional, economic or other types of violence, for example, nor on any violence perpetrated within same-sex partnerships – a limitation effectively transferred to this review.
• The analysis covers interventions from low, middle and high incomes countries, and often from contexts that vary significantly from Victoria’s in terms of existing prevalence rates, norms around gender roles, and the extent of state and institutional support in creating an enabling environment for change. Caution is needed when drawing conclusions about their applicability in the Victorian context.
• Only a small number of interventions have been subject to longitudinal evaluation (footnote 4) that would illuminate, in the intimate partner violence example, what happens as participants’ relationships evolve, or as they enter new ones, due to an earlier or ongoing intervention. The paucity of evaluations that capture longer-term change limits this and other reviews, and indeed currently hampers a broader understanding of how to effectively prevent violence against women

Return to the top of this page

Findings

Evidence of the impact of prevention interventions

Well-crafted primary prevention interventions can reduce violence against women for groups they engaged, in programmatic timeframes

• As of 2020, at least 96 randomised controlled trials or quasi-experimental studies had been conducted globally, assessing interventions for their impact on men’s perpetration or women’s victimisation for physical or sexual intimate partner violence, and non-partner sexual violence (reference 1).  
• 41 (including 14 in high-income countries) of these showed ‘positive’ (statistically significant) impact at the end of the intervention, and 18 (including three in high-income countries) were ‘promising’ (e.g. significant for a sub-group, or a non-significant trend).  
• 37 (including 13 in high-income countries) showed ‘no impact’ on the forms of violence described above. However, they may have had statistically significant reductions on other forms of violence, such as emotional or economic violence against women (footnote 5).
• No comparable impact evaluations were found in Victoria (or Australia), though several Victorian interventions are similar to examples shown to be impactful in similar contexts elsewhere.
• ‘Positive’ and ‘promising’ interventions in high-income countries included schools and college-based interventions, couple and parenting interventions, and ante- or-post-natal interventions, working with general (as opposed to ‘higher-risk’) populations, and addressing the drivers of violence.  
• A smaller number of interventions reviewed in high-income countries aimed to reduce individual (higher) risk. Most targeted behaviours of young women (e.g. with regards to alcohol consumption), with some showing impact on victimisation (reference 1). Only one intervention targeted men with substance abuse issues, however it did not show impact on perpetration of violence (reference 7). Another supporting young people leaving out-of-home care did show impact (reference 8).  

Limited evidence suggests impact can result from new incidents of violence being prevented, or recurring violence being reduced, or do both

• A recent study examined in detail the data from six randomised controlled trials. All interventions were deemed impactful in low- and middle- income contexts (NB: with high past-12-month prevalence levels relative to Victoria’s).  
• The interventions were conceptualised as primary prevention, and included activities aimed at shifting gender norms, building skills, engaging leaders, etc. However, they did not specifically target ‘at risk groups’ or adopt early intervention approaches.
• Impact was found to result from:
  1. new incidences of violence being prevented from emerging (against statistical expectations), or
  2. reductions in recurring violence (i.e. that was already existing at baseline), or  
  3. a combination of the two (reference 9).
• There is currently insufficient research to indicate which factors determine whether an intervention will have more impact on ‘new’ or ‘recurring’ violence. Recommendations included the need for interventions to have a clear theory of change and specific impact pathways for desired outcomes.

Existing evidence is too limited – and skewed – to enable definitive conclusions about effectiveness by type of intervention

• Attempts have been made to classify interventions by strength of available evidence, but these show what interventions are getting evaluated for impact on violence against women, and how they compare against the indicators those evaluations measured. They do not provide an objective assessment of everything that works, or could work, among all the prevention interventions that exist.  
• The quality of design and implementation determine how effective an intervention is more so than the ‘type of intervention’ per se (reference 10).
• Most existing evaluations are of Interventions directly targeting/engaging individuals to promote changes in attitudes, beliefs and behaviours. Equally important is the work to shift environments or build capacity for prevention, but it is difficult, and arguably undesirable, to directly attribute changes in victimisation/perpetration rates to such interventions through evaluative processes (therefore such evaluations are rare).
• Even less is known about how different intervention types might enhance the impact of other interventions when implemented as part of a coordinated, multi-component approach (reference 10).

Evidence of impact at population level

Few evaluations seek to measure population-level impact, but there is some emerging evidence

• The evaluations of most interventions aiming to prevent violence against women are limited to measuring impact on target groups specific to that intervention’s scope, such as college students or first-time parents.
• The interventions that have demonstrated population-level impact fall into two broad categories (described further below):
  1. policy, legislative and environmental interventions, and
  2. place-based, multi-component community mobilisation interventions.

Certain policy, legislative and environmental interventions have been correlated with reductions in violence against women

• Two of these involved a suite of initiatives: the US Violence Against Women Act (VAWA) (reference 11) correlated with reductions in rates of, and homicides due to, intimate partner violence, in jurisdictions receiving VAWA grants; and, in a local Nicaragua municipality, a number of initiatives implemented over a 20-year period correlated with reductions in women’s reported experience of physical intimate partner violence (footnote 6).
• Other policy and welfare system changes correlated with reductions in victimisation, including reforms to parental leave provisions, firearms control, off-premises alcohol outlet density and school-based restraining orders (all in high-income countries) (reference 6); and two social protection programs, in Peru and Colombia (reference 12). However, the strength of evidence varies for different types of policy change (e.g. evaluations of similar interventions in other contexts may show conflicting results).

Three multi-component, place-based community mobilisation interventions demonstrated population-level impact

• All three (references 13-15) were implemented in low- and middle-income African countries (two in Uganda and one in Ghana). The population size was limited by the anticipated reach of the interventions, but all interventions involved several thousand people across one or more administrative areas of a local municipality.  
• Impact in all three was measured through randomised controlled trials, using surveys in ‘control’ and ‘treatment’ communities.  
• The programs involved ‘a set of interventions in which multiple components are deployed (reference 1).’  

Features of place-based interventions demonstrating population-level impact

• The three prevention interventions presented above were the only randomised controlled trial-evaluated interventions found by the review that demonstrated impact on perpetration rates or victimisation of violence against women at the population level (footnote 7). They all met quality principles discussed below, and shared several other features:
  1. The demonstrated impact was limited to intimate partner violence against women, including when the intervention itself aimed to prevent broader forms of violence against women. In this latter case, programmers concluded that prevention of other forms, such as non-partner sexual violence, likely requires a specialised approach (reference 16).
  2. They were multi-component and included activities at multiple levels of the social ecological model – including institutional and policy advocacy, organisational development in locally-based sectors (such as education, police and health), direct participation programs, communications and social marketing.
  3. They adopted a ‘people-powered’ and benefits-based approach, supporting individuals and groups to develop skills and help guide actions and behaviours that positively shift drivers of violence across different settings where people live, work and play (reference 17).  
  4. They were designed and phased around theories of behaviour change, not simply in terms of violence perpetration behaviour, but also the behaviours associated with the drivers of that violence. They planned activities across multiple levels of the social ecology in a way that aligned with a sequential understanding of the behaviour change process (reference 18).

Conditions and factors that influence impact

Foundational conditions that underpin primary prevention intervention effectiveness

• Recent reviews (reference 19) of interventions evaluated as ‘impactful’ highlighted several conditions that are foundational for achieving impact. These are:  
  1. A prevention infrastructure: the core components, facilities, services and mechanisms considered essential for enabling and sustaining effective prevention of violence against women.
  2. Adherence to quality design principles: address the identified drivers, use an explicit gender analysis, follow a program logic, tailor to context, setting and population (footnote 8).
  3. Effective partnerships with, and connections to, response services including high-level support for victim survivors.
• Good practice in each of these three categories is well-documented elsewhere across international, national and Victorian literature, and not reproduced in this review (footnote 9).

Factors that make a difference to the scale and sustainability of violence against women impact

• In addition to the above, recent reviews show engagement ‘intensity’ with participants was a determining factor for the level of impact, particularly in interventions directly engaging individuals and groups. The intensity to impact relationship was one of ‘more is more’ (reference 20) and interventions had no impact below a certain level of intensity (reference 10).
• The following factors were identified as indicators of intensity:
  1. An adequate number of skilled and supported practitioners
  2. Appropriate program length
  3. Number, duration and frequency of sessions allowing time for reflection and experiential learning.

Combining interventions to maximise or expand impact

Rationale for combining interventions: the mutually-reinforcing effect

• Health promotion theory and practice suggests combining interventions in a coordinated way across settings and the social ecological levels to achieve a ‘mutually-reinforcing effect.’ (reference 21)
• This might be achieved through combining multiple strategies or techniques (reference 21), in multiple settings/sites (reference 22) (footnote 10), and/or at multiple social ecology levels (reference 23). In practice, a combination is common (given, for example, different techniques tend to be more applicable to certain levels of the social ecology).

Mutual reinforcement in prevention of violence against women

• A mutually-reinforcing effect has been demonstrated in many different health promotion areas (reference 23).
• In the prevention of violence against women, several evidence reviews noted that multi-component interventions tend to have more impact that single-component ones (references 2-3, reference 10, reference 19).
• A recent meta-analysis (reference 24) of impact evaluations of violence against women interventions globally (from prevention to response) concluded that combination or multi-component interventions were on average 60 per cent more effective than single-component ones – though caution is needed in interpreting this conclusion for specific programs and contexts. (footnote 11)
• Exactly how the ‘mutually-reinforcing effect’ operates is poorly understood. As previously noted, most existing evaluations are of individual rather than multi-component interventions.  

Initial principles for maximising impact by combining interventions

• The reviewed research makes clear that a direct relationship between ‘more interventions’ and ‘more impact’ should not be assumed (reference 21), even if the interventions meet quality standards and other conditions for impact.
• Some guidance for achieving mutual reinforcement in prevention of violence against women activity is emerging. The combination of interventions should be:
  1. Based on the principle of addressing multiple drivers and reinforcing factors – the most recent rigorous evidence review suggests that ‘multiple components’ are not, in and of themselves, the key to mutual reinforcement, but rather ‘more components’ means greater reach and range of influence to address multiple drivers, which is crucial (reference 10).
  2. Sequenced to support the phases of behaviour change – while a range of approaches to sequencing have been suggested across health promotion research, the evaluated successful place-based, multi-component interventions have phased activity across levels and strategies in accordance with behaviour change theories.
  3. Synchronised across settings and levels – ensuring that ‘intervention components and activities that are implemented at the different levels are synchronized in terms of optimal timing’ and with ‘a sense of coherence in the themed intervention activities (reference 25).’ It also requires a certain intensity of activity, not just within a single intervention but between them: ‘creating repetition of program activities and messages’ through ‘a multiple-exposure approach (reference 25).’

Return to the top of this page

Discussion

Considerations around impact and its measurement in the Victorian context

• The evidence distilled here challenges a long-held assumption that primary prevention work (only) lowers risk gradually and in a diffuse manner (reference 26). Primary prevention interventions for violence against women can indeed have relatively short-term impact on individuals, including for those who may already be perpetrating or experiencing violence (reference 1).
• The fact that impact can be due to the prevention of ‘new’ violence (i.e. before it occurs), and/or reductions in recurring violence, supports the premise in Change the Story that primary prevention can contribute to the response to existing violence by stopping or reducing it (while providing the social norms and structures that maximise the possibility of maintaining changed behaviours and rebuilding lives after violence has occurred) (reference 4).
• The research reviewed does not account for the specific impact an intervention might have on higher risk groups who are not already experiencing or perpetrating violence. These would be subsumed under the category of those not experiencing/perpetrating violence at baseline in the studies from which the above evidence is drawn. More research is needed to fully understand the factors that influence the type of impact a primary prevention intervention might have, and on which sub-sets of the population.  
• None of the above implies, of course, that separate response and early intervention activities are not needed. Far from it: all three are crucial elements of a comprehensive and integrated system.  

Considerations for how to maximise impact in the Victorian context

• Whether prevention activity does achieve impact, and to what extent, is a matter of design quality, delivery intensity and appropriate contextualisation of the work, criteria that align with and reinforce those in existing frameworks used in Victoria, most notably Change the Story (reference 4).
• Effective and available response systems, and an enabling legal, policy and institutional environment, can themselves magnify prevention activity impact (reference 24), which supports the approach used in Victoria, and emphasises the need for continued investment.
• However, prevention interventions need to be delivered at a certain level of intensity to have impact. This implies attempting to ‘stretch’ programming to reach more people, but with fewer opportunities for direct engagement, for example, or with larger practitioner-to-population ratios, can undermine the overall utility of the program.
• While exact figures vary (i.e. for ‘adequate’ dosage, duration, or number of skilled practitioners), depending on the setting and context, some ‘threshold’ estimates are emerging for specific types of interventions or settings. International analyses can provide useful up-to-date guidance to prevention activity in Victoria on this (reference 10).
• The few interventions worldwide that have demonstrated population-level impact have been place-based, ‘people powered’ community mobilisation programs, employing multiple techniques across several levels of the social ecology. Coordination mechanisms for such approaches exist in Victoria (reference 27), but further research is needed to understand how to contextualise and ensure adequate intensity of community mobilisation work for prevention of violence against women in the Victorian context.

Initial implications for a place-based saturation model in Victoria

• In planning and coordinating multi-component prevention work, a direct relationship between ‘more components’ and ‘more impact’ cannot be assumed. Rather, increased impact appears to result from addressing multiple drivers and reinforcing factors (in a quality way).
• A deep understanding of the nature of the current (past 12-month) violence against women prevalence is therefore needed, in the context where the intervention will take place, to identify specific drivers and reinforcing factors, and their relative weight. Addressing these multiple factors should guide the choice of components employed – techniques, settings and levels – in a multi-component prevention program.  
• This is a departure from approaches that seek mutual reinforcement by simply multiplying techniques, settings or levels of action alone.  
• The development of a Victorian place-based model could learn from multi-component community mobilisation models globally in terms of sequencing activity to support the phases of behaviour change, and synchronising implementation of different activities across settings and levels.  
• However, expectations around the possible impact of any Victorian multi-component, place-based community mobilisation model should also be assessed with caution due to differences in context, policy and institutional support, and past 12-month prevalence rates, between Victoria and the sites of evaluated examples. 

Return to the top of this page

Conclusion

• This review has shown that multi-component prevention work is indeed more impactful than single component interventions when certain criteria and conditions are met. These have been identified in the report, and include adherence to quality design principles, adequate implementation intensity, and attention to addressing multiple drivers through an intentionally phased approach.
• The review found examples of coordinated, multi-component, place-based approaches meeting these conditions – that have resulted in population-level impact.
• There is good reason to expect that impact on perpetration and victimisation of violence against women could be achieved at population level through a Victorian multi-component, place-based model that draws on the learnings above.
• Further research and consultation are needed to determine foundational design and implementation parameters for a place-based saturation model, along with the conceptual approach and key elements that should be included in the model and its design. 

Return to the top of this page

Footnotes