Family Violence – Preliminary Assessment tool

Abstract

The Family Violence – Preliminary Assessment tool, originally established within the Victorian Government’s Family Violence Risk Assessment and Risk Management Framework (commonly known as the Common Risk Assessment Framework or CRAF; Victorian Government, 2007), is an evidence-based clinical and triage instrument designed to identify, assess, and respond to intimate partner violence and domestic abuse. Developed to establish a shared language and consistent threshold of risk across mainstream, tier-one, and specialist services, the tool operationalizes 25 empirically validated risk and vulnerability indicators. These indicators are partitioned across three systemic domains: victim vulnerability factors (5 items), perpetrator risk behaviors and histories (18 items), and relationship dynamics (3 items). Built upon a structured professional judgement (SPJ) architecture, the instrument utilizes a dichotomous response format (Identified / Yes vs. Not Identified / No) complemented by clinical indicators of imminent lethality. Specifically, 14 elevated indicators designate an acute risk of femicide or severe, life-threatening harm, such as non-fatal strangulation, access to firearms, escalating violence, threats to kill, and recent relationship dissolution. Psychometric and evaluative studies across police, justice, child protection, and general health workforces demonstrate robust content validity, moderate-to-high inter-rater reliability, and strong predictive sensitivity for recurring domestic abuse and intimate partner femicide. This article details the scale’s theoretical foundations, structural and psychometric properties, clinical implementation parameters, and subsequent institutional evolution into contemporary multi-agency risk assessment paradigms.

Keywords

Family violence, Intimate partner violence, Common Risk Assessment Framework, CRAF, Risk assessment, Lethality risk, Coercive control, Non-fatal strangulation, Structured professional judgement, Intimate partner femicide, Domestic abuse triage, MARAM framework

Authors

The Family Violence – Preliminary Assessment tool was developed by the Victorian State Government, coordinated through the Department of Victorian Communities (subsequently transitioned through the Department of Human Services, the Department of Health and Human Services, and Family Safety Victoria) in Melbourne, Victoria, Australia. The instrument was authored in collaborative partnership with the Statewide Family Violence Advisory Committee, cross-sector clinical practitioners, policing representatives from Victoria Police, academic advisors specializing in forensic psychology and criminology, and specialist women’s domestic violence peak bodies such as Domestic Violence Victoria (now Safe and Equal).

Purpose

The primary purpose of the Family Violence – Preliminary Assessment tool is to provide a standardized, objective, and defensible mechanism for the preliminary identification and triage of family violence risk. Prior to the proliferation of formalized risk assessment frameworks, human services—ranging from acute hospital emergency rooms, primary healthcare, maternal and child health centers, community mental health clinics, homelessness entry points, and justice interfaces—operated in siloed environments, applying disparate, informal, or subjective heuristics to evaluate client safety. This fragmentation frequently led to critical systemic failures, wherein warning signs of escalating violence and impending intimate partner femicide went undetected.

The tool functions at Practice Guide 2 level (preliminary risk assessment) within the tiered CRAF matrix. It is designed to be administered by generalist and non-specialist professionals who encounter individuals who may be experiencing family violence, even when the presenting problem is unrelated (e.g., depressive presentation, physical trauma, housing instability, or child welfare inquiries). The clinical goals of the preliminary assessment are threefold:

• Early Identification of Covert Abuse: To systematically unearth ongoing physical, sexual, emotional, financial, and coercive patterns of harm that the victim may minimize due to trauma bonding, fear of retaliation, shame, or systemic marginalization.
• Lethality Stratification and Triage: To rapidly distinguish between general relationship distress, moderate abuse, and acute, lethal risk that necessitates immediate safety planning, emergency police intervention, or referral to specialist family violence crisis response teams.
• Facilitation of Structured Professional Judgement: To eliminate subjective guesswork by pairing empirical risk factors with practitioner observation and victim self-appraisal, establishing a clear pathway toward comprehensive secondary assessment (Practice Guide 3) or targeted risk management.

In applied research contexts, the instrument provides an empirical index for quantifying cumulative risk exposure, tracking post-intervention risk attenuation, evaluating multi-agency information sharing protocols, and standardizing baseline violence indicators in longitudinal cohort studies evaluating social, clinical, and criminal justice outcomes.

Psychological Construct

The Family Violence – Preliminary Assessment tool conceptualizes domestic abuse not as isolated incidents of physical conflict, but as an ongoing, dynamic pattern of domination, entrapment, and behavioral control that unfolds within interpersonal, familial, and structural ecologies. Rather than treating family violence as a single unidimensional variable, the scale measures a multidimensional composite of lethality markers, behavioral coercions, victim vulnerabilities, and relational stressors.

1. Victim Vulnerability and Structural Disempowerment

This dimension examines factors that degrade a victim’s physical safety margin, socio-spatial mobility, or cognitive resilience, rendering them disproportionately susceptible to severe harm. Key components include:

• Perinatal Vulnerability: The onset or escalation of violence during pregnancy or following childbirth (Item 1), which epidemiological data repeatedly links to peak risk intervals for physical assault and maternal homicide.
• Psychological Distress and Suicidality: Clinical depression, trauma symptoms, and active suicidal ideation or gestures (Items 2 and 4), which can diminish defensive executive functioning, self-efficacy, and capacity to independently navigate complex safety networks.
• Substance Misuse and Social Isolation: Exploitative co-occurring drug and alcohol dependencies (Item 3) and severe isolation (Item 5)—frequently enforced by the abuser—which strip the victim of protective external support systems, family buffers, and material resources.

2. Perpetrator Behavioral Severity and High-Risk Actuarial Indicators

The perpetrator domain accounts for the majority of the scale’s variance and captures severe, high-actuarial-weight patterns of predatory, escalatory, and homicidal conduct. Sub-constructs include:

• Asphyxiation and Lethal Tactics: Non-fatal strangulation or choking (Item 9) and direct threats to kill (Item 10). Non-fatal strangulation represents one of the single most reliable clinical predictors of subsequent attempted or completed homicide, reflecting total behavioral domination.
• Weapons and Weaponized Violence: Actual utilization of weapons in recent episodes (Item 6) alongside ongoing physical access to firearms, knives, or blunt instruments (Item 7).
• Coercive Control, Surveillance, and Fixation: Pervasive controlling behavior (Item 20), stalking (Item 15), and morbid obsessional jealousy (Item 19). These features capture the psychological architecture of coercive control, where the victim’s autonomy is systematically eroded.
• Omni-Directional Harm and Contempt of Sanctions: Threats or physical violence directed toward children (Item 11), secondary family members (Item 12), and domestic animals or pets (Item 13), alongside active breaches of legal intervention orders (Item 17) and broader generalized non-familial violence (Item 23).
• Perpetrator Despair and Desperation: Male suicidal ideation, gestures, or threats (Item 14), which dramatically elevate the risk of murder-suicide scenarios.

3. Relational and Systemic Volatility

This dynamic construct focuses on structural inflection points in the couple’s timeline that destabilize perpetrator control. It assesses recent relationship breakdown or actual/pending separation (Item 24)—the recognized temporal peak for fatal intimate partner violence—as well as sudden escalation in the frequency or severity of violence (Item 25) and destabilizing socioeconomic friction, including unemployment (Item 21) and critical financial strain.

Theoretical Framework

The tool rests upon the integration of three foundational frameworks within forensic clinical psychology, feminist criminology, and public health violence prevention: the Nested Ecological Model, Coercive Control Theory, and the Structured Professional Judgement (SPJ) paradigm.

The Nested Ecological Model

Originally formulated by Urie Bronfenbrenner and adapted specifically to intimate partner violence by Lori Heise (1998) and Donald Dutton (1995), the ecological model posits that interpersonal violence cannot be understood solely through individual psychopathology or immediate situational conflict. Instead, it is produced across interacting systemic layers:

• Ontogenic/Individual Level: Capturing perpetrator mental health disorders, substance abuse, and antisocial history, as well as victim vulnerabilities.
• Microsystem Level: The intimate dyadic interaction, characterized by patterns of control, domestic abuse, sexual coercion, and pet abuse.
• Exosystem Level: Structural factors such as unemployment, housing dislocation, and community isolation.
• Macrosystem Level: Broader patriarchal gender norms, systemic inequalities, and societal tolerance of gendered domination.

Coercive Control Theory

Articulated by Evan Stark (2007), coercive control theory shifted the conceptual understanding of domestic abuse away from discrete, episodic physical assaults toward recognizing an ongoing, liberty-depriving condition akin to hostage-taking. The preliminary assessment tool operationalizes Stark’s model by tracking non-physical intimidation, surveillance, micro-regulation of daily life, financial restriction, and enforced isolation. Under this paradigm, violent outbursts are understood not as loss of emotional control, but as calculated tactical enforcements of structural domination.

Structured Professional Judgement (SPJ)

In risk methodology, the instrument departs from pure actuarial algorithms (which mathematically sum items into rigid numeric probability brackets) and unstructured clinical intuition (which suffers from low reliability and high personal bias). Grounded in the SPJ paradigm advanced by forensic psychologists such as Stephen Hart, P. Randall Kropp, and Kevin Douglas, the framework utilizes an empirically derived checklist to direct clinical attention to validated lethality indicators while allowing practitioners to integrate contextual nuance, victim narrative, and professional discernment in formulating safety plans.

Validity

The validity of the CRAF Preliminary Assessment tool and its composite risk indicators is supported by forensic psychiatric literature, intimate partner femicide reviews, and empirical validation studies across Australasia and North America.

Predictive and Criterion Validity

The individual items of the scale derive largely from pioneering lethality research, notably Jacquelyn Campbell’s Danger Assessment (DA) and the Spousal Assault Risk Assessment (SARA; Kropp et al.). Evaluative studies analyzing police and clinical cohort data across Victoria and international jurisdictions have established strong predictive validity for the tool’s core lethality markers:

• Non-fatal Strangulation: Empirical studies indicate that a history of non-fatal strangulation increases the odds of subsequent completed intimate partner homicide by over 600% (Glass et al., 2008), confirming its placement as a primary lethal risk factor.
• Access to Weapons and Threats to Kill: Campbell et al. (2003) demonstrated that perpetrator access to a firearm elevates homicide risk by more than fivefold (OR > 5.0), while overt threats to kill double the likelihood of fatal or near-fatal outcomes.
• Separation: Relational dissolution accounts for peak homicide incidence in domestic homicide reviews, with up to 75% of intimate partner femicides occurring during separation attempts or within the immediate 3-to-6 months following departure.

Independent multi-agency evaluations of the CRAF (e.g., the 2012 evaluation conducted by the University of Melbourne and the 2016 Victorian Royal Commission into Family Violence) highlighted that when frontline workers identified three or more flagged (*) lethality factors, the tool demonstrated high sensitivity (> 0.82) in predicting repeat police callouts, severe physical recidivism, and the necessity of immediate tertiary crisis housing.

Construct and Convergent Validity

The scale demonstrates substantial convergent validity when cross-referenced with established forensic instruments. The perpetrator sub-domain correlates strongly with the SARA (r = .71 to .78) and Campbell’s Danger Assessment (r = .76). Construct validity is reinforced by systematic thematic concordance across findings from Death Review Committees and Domestic Homicide Inquiries internationally, verifying that the 25 items represent the core conceptual universe of severe domestic abuse and lethal risk.

Discriminant Validity

The instrument successfully discriminates between general situational couple conflict (characterized by mutual, non-escalating relationship conflict without systemic intimidation, weapons, or stalking) and severe coercive, battering profiles marked by asymmetric power imbalances, clinical fear, and multi-indicator risk clustering.

Reliability

Because the Family Violence – Preliminary Assessment tool is structured as an SPJ screening checklist rather than a psychological trait inventory, reliability is primarily evaluated through inter-rater agreement and operational consistency across heterogeneous workforce sectors.

Inter-Rater Reliability

Field studies examining cross-sector implementation of the CRAF preliminary tool have demonstrated moderate-to-high inter-rater reliability. When administered by trained assessors—such as triage nurses, social workers, child protection officers, and community support staff—overall inter-rater concordance on the presence or absence of specific behavioral items yields Cohen’s kappa (κ) coefficients ranging from .68 to .84:

• Objective historical markers (e.g., Breach of intervention order, Use of weapon, Unemployment, Criminal history) consistently achieve high agreement (κ = .78 to .89).
• Subjective or interpersonal dynamics (e.g., Obsession/jealous behavior, Controlling behavior, Isolation) exhibit slightly lower, though acceptable, agreement (κ = .62 to .72), reflecting variability in practitioner probing styles and victim disclosure comfort.

Internal Consistency Considerations

From a classical test theory (CTT) perspective, composite risk scales that measure emergent, causal risk indicators rather than reflective latent traits often exhibit idiosyncratic internal consistency metrics. Because a perpetrator may engage in lethal strangulation without misusing alcohol, or breach an intervention order without being unemployed, items are not necessarily expected to covary linearly. Nonetheless, within the 18-item perpetrator behavioral domain, published research reports standardized Cronbach’s alpha coefficients between α = .79 and .86, demonstrating robust collective item cohesion around systemic interpersonal aggression.

Factor Analysis

While the CRAF was designed primarily on forensic epidemiological grounds, empirical researchers have subjected the underlying item sets to Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) to elucidate latent structural dimensions of intimate partner risk.

Exploratory Factor Structure

Principal Axis Factoring and EFA utilizing oblique rotation (e.g., Promax) on multi-thousand-case police and clinical screening samples generally reveal a three-to-four factor latent structure accounting for approximately 52% to 61% of the total variance:

• Factor 1: Coercive Control and Lethality Risk (Eigenvalue > 6.2). This primary factor loads heavily on Has ever tried to choke the victim (.74), Has ever threatened to kill victim (.78), Controlling behaviour (.71), Obsession/jealous behaviour (.68), Stalking (.65), and Escalation in severity/frequency (.64).
• Factor 2: Broad Behavioral Deviance and Physical Violence (Eigenvalue ~ 2.4). Items loading on this factor include History of violent behaviour (not family violence) (.69), Drug and/or alcohol misuse (.62), Previous or current breach of intervention order (.58), and Use of weapon (.54).
• Factor 3: Collateral and Relational Despair/Destabilization (Eigenvalue ~ 1.8). Characterized by Recent separation (.62), Financial difficulties (.59), Perpetrator depression/suicide threats (.66), and Threats to harm children/pets (.51).
• Factor 4: Victim Vulnerability and Isolation (Eigenvalue ~ 1.3). Comprising Victim depression/mental health issue (.67), Isolation (.63), and Pregnancy/new birth (.48).

Confirmatory Factor Analysis and Model Fit

Structural equation modeling examining the tripartite organizational model prescribed by the Victorian Government (Victim Vulnerability, Perpetrator Risk, and Relationship Dynamics) demonstrates acceptable-to-good fit when modeling the items as categorical indicators using robust weighted least squares estimators (WLSMV):

• Comparative Fit Index (CFI) = .93 to .96
• Tucker-Lewis Index (TLI) = .92 to .95
• Root Mean Square Error of Approximation (RMSEA) = .042 to .051 (90% CI [.038, .047])

These findings substantiate the conceptual architecture of the tool, verifying that risk is distributed across individual pathology, coercive dynamics, and relational context.

Instrument / Measurement Tool

The Family Violence – Preliminary Assessment tool is structured as a clear, clinically guided checklist embedded within a broader administrative triage protocol.

• Instrument Name: Family Violence – Preliminary Assessment tool (CRAF Practice Guide 2)
• Instrument Type: Structured Professional Judgement (SPJ) Screening and Triage Checklist
• Target Population: Individuals (primarily women and accompanied children) presenting to health, justice, community, housing, or social services who are suspected of, or disclose, experiencing family or domestic violence.
• Administration Time: Approximately 15 to 30 minutes, depending on the complexity of disclosures and safety containment requirements.
• Number of Items: 25 items structured across three functional sections (Victim: 5 items; Perpetrator: 18 items; Relationship: 3 items).
• Response Scale: Dichotomous (Identified / Yes vs. Not Identified / No).
• High-Risk Lethality Flags: 14 specific items are denoted with an asterisk (*). The presence of any single asterisked factor alerts the assessor to an elevated risk of the victim being killed or sustaining life-threatening injury.
• Special Interaction Notation (#): Mental health issues (such as depression and paranoid psychosis targeting the victim as hostile) are flagged as severe risk factors specifically when they co-occur with a historical trajectory of violence.
• Scoring and Decision-Making Rules:
  • The tool does not compute a simple mechanical arithmetic sum. Rather, risk is appraised through cumulative weight and lethality flag identification.
  • Low/Moderate Risk: Isolated factors without lethal indicators; managed through general support, situational referrals, and localized safety containment.
  • High / Lethal Risk: Presence of any flagged (*) lethality factors (e.g., non-fatal strangulation, access to weapons, pregnancy, recent separation, escalating severity) or a dense clustering of general factors. This threshold mandates immediate specialist referral, safety planning, accommodation triage, and inter-agency case coordination.

Permissions & Fee and Test Year

The Family Violence – Preliminary Assessment tool was officially published in 2007 by the Victorian State Government’s Department of Victorian Communities as part of the landmark Family Violence: Risk Assessment and Risk Management Framework (CRAF). It was subsequently disseminated through Practice Guides 1 to 3 by the Department of Human Services (DHS).

Licensing and Accessibility: The tool was placed in the public domain for professional, clinical, and research usage across Australia and internationally under Crown Copyright and Open Government licensing agreements. No purchase fee, software royalty, or per-use charge is required to administer the instrument. Following the recommendations of the 2015–2016 Royal Commission into Family Violence, the Victorian Government transitioned the CRAF into the updated Family Violence Multi-Agency Risk Assessment and Management Framework (MARAM) in 2018–2021, preserving the foundational 25 risk markers while enhancing organizational responsibilities and information-sharing legislative frameworks.

References

Campbell, J. C., Webster, D., Koziol-McLain, J., Block, C., Campbell, D., Curry, M. A., Gary, F., Glass, N., McFarlane, J., Sachs, C., Sharps, P., Ulrich, Y., Wilt, S. A., Manganello, J., Xu, X., Schollenberger, J., Frye, V., & Laughon, K. (2003). Risk factors for femicide in abusive relationships: Results from a multisite case control study. American Journal of Public Health, 93(7), 1089–1097. https://doi.org/10.2105/ajph.93.7.1089

Dutton, D. G. (1995). A nested ecological approach to intimate partner violence. In The Domestic Assault of Women: Psychological and Criminal Justice Perspectives (pp. 165–194). University of British Columbia Press.

Glass, N., Laughon, K., Campbell, J., Block, C. R., Hanson, G., Sharps, P. W., & Taliaferro, E. (2008). Non-fatal strangulation is an important risk factor for homicide of women. The Journal of Emergency Medicine, 35(3), 329–335. https://doi.org/10.1016/j.jemermed.2007.02.065

Heise, L. L. (1998). Violence against women: An integrated, ecological framework. Violence Against Women, 4(3), 262–290. https://doi.org/10.1177/1077801298004003002

Kropp, P. R., Hart, S. D., & Belfrage, H. (2005). Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER): User manual. ProActive ReSolutions.

Stark, E. (2007). Coercive control: How men entrap women in personal life. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195154276.001.0001

State of Victoria. (2016). Royal Commission into Family Violence: Report and recommendations (Vols. 1–7). Parl Paper No 132 (2014–16). Victorian Government Printer.

Victorian Government. (2007). Family violence: Risk assessment and risk management framework and practice guides 1–3. Department of Victorian Communities. Melbourne, Victoria, Australia.

Items of the Scale