Severity of Violence Against Women Scale/Severity of Violence Against Men Scale (SVAWS/ SVAMS)

Abstract

The Severity of Violence Against Women Scale (SVAWS) and its counterpart, the Severity of Violence Against Men Scale (SVAMS), developed by Linda L. Marshall in 1992, are comprehensive psychometric instruments designed to assess the continuum, multidimensionality, and severity of intimate partner violence (IPV) victimization and perpetration. Comprising 46 behavioral items, the instruments measure aggressive behaviors ranging from symbolic violence and psychological intimidation to life-threatening physical assaults and sexual aggression. Responses are recorded on a 4-point frequency continuum (1 = Never, 2 = Once, 3 = A few times, 4 = Many times) within a specified timeframe (e.g., past year or entire relationship). Structurally, the instrument operationalizes violence across two higher-order dimensions: Threats of Violence (19 items) and Acts of Violence (27 items). These overarching dimensions are further partitioned into nine empirical subscales: Symbolic Violence, Mild Threats, Moderate Threats, Serious Threats, Minor Violence, Moderate Violence, Serious Violence, Sexual Aggression, and Very Serious Violence.

Extensive psychometric investigations have established robust internal consistency reliability across clinical, community, and university populations, with Cronbach’s alpha coefficients routinely exceeding .90 for the overarching dimensions and ranging from .72 to .94 across individual subscales. Construct validity is supported by strong convergent associations with the Conflict Tactics Scales (CTS/CTS2), the Index of Spouse Abuse (ISA), measures of post-traumatic stress disorder (PTSD), depressive symptomatology, relational dissatisfaction, and safety-seeking behaviors. Confirmatory factor analyses confirm that classifying violent behaviors into gradated severity tiers significantly enhances ecological validity and prognostic utility compared to aggregate unidimensional checklists. This paper presents an exhaustive academic and psychometric examination of the SVAWS/SVAMS, articulating its theoretical architecture, validation trajectory, factor structure, administration parameters, diagnostic scoring conventions, and clinical research applications.

Keywords

Severity of Violence Against Women Scale, Severity of Violence Against Men Scale, SVAWS, SVAMS, intimate partner violence, domestic violence assessment, psychological threats, physical aggression, sexual coercion, psychometrics

Authors

The Severity of Violence Against Women Scale (SVAWS) and the Severity of Violence Against Men Scale (SVAMS) were conceived, developed, and empirically validated by Linda L. Marshall, Ph.D.

• Primary Author: Linda L. Marshall, Ph.D.
• Affiliation: Department of Psychology, University of North Texas, Denton, Texas, United States.
• Biographical and Scholarly Context: Dr. Linda L. Marshall was a leading scholar in social psychology, family psychology, and intimate partner aggression. Her pioneering research throughout the late 1980s and 1990s was instrumental in moving the field of interpersonal violence assessment beyond dichotomous physical strike tallies toward nuanced conceptualizations that incorporate coercive control, psychological threats, sexual aggression, and perceived severity gradations from both female and male victim perspectives.
• Historical Correspondence: Dr. Marshall conducted extensive empirical studies published in the Journal of Family Violence, evaluating behavioral reports from both female community samples and male cohorts to establish gender-comparable assessment paradigms.

Purpose

The primary purpose of the SVAWS and SVAMS is to provide an objective, behaviorally specific, and gradated assessment of interpersonal violence in dating, cohabiting, and marital relationships. Prior to Marshall’s seminal 1992 publications, the measurement of domestic violence in academic and forensic contexts was heavily dominated by early iterations of the Conflict Tactics Scale (CTS; Straus, 1979). While groundbreaking, the original CTS faced sustained criticism for treating disparate acts of aggression as functionally equivalent within broad categories—equating a shove with a punch, or an isolated push with repeated, terrifying threats to kill. Furthermore, traditional instruments often obscured the profound psychological terror generated by non-contact threats, symbolic demonstrations of destructive power, and coercive sexual violations.

Marshall designed the SVAWS and SVAMS to overcome these psychometric and clinical limitations by establishing a continuum of severity based on empirical ratings of physical harmfulness and subjective threat. The instrument fulfills multiple specific objectives across research, clinical, and forensic settings:

• Behavioral Specificity Without Value Judgments: The scale avoids ambiguous, legally freighted, or subjective labels such as “abused,” “battered,” or “assaulted.” Instead, it employs concrete, behavioral descriptions of observable conduct (e.g., “Hit or kicked a wall, door or furniture”; “Choked you”; “Demanded sex whether you wanted to or not”), which reduces cognitive bias, social desirability distortion, and defensive denial among respondents.
• Differentiation of Severity Tiers: By systematically parsing aggressive behaviors into nine distinct subscales spanning from Symbolic Violence to Very Serious Violence, the instrument enables clinicians and researchers to map the precise behavioral architecture of violence. It differentiates between couples experiencing low-level, bidirectional escalation and those caught in severe patterns of coercive control, terrorization, and lethal physical threat.
• Parallel Gender Measurement: The scale was validated with parallel item structures for assessing violence directed toward women (SVAWS) and violence directed toward men (SVAMS). This permits empirical investigation into gender symmetries and asymmetries in perpetration, victimization, physical injury severity, and fear induction.
• Clinical Risk Stratification and Safety Planning: In therapeutic, shelter, and medical settings, the scale serves as an evidence-based risk assessment protocol. Identifying items within the “Serious Threats” and “Very Serious Violence” subscales (e.g., weapons use, strangulation, death threats) signals elevated lethality risk, informing targeted safety planning, legal interventions, and clinical stabilization.
• Research and Program Evaluation: The instrument demonstrates high sensitivity to change, rendering it an effective outcome measure for evaluating domestic violence intervention programs, batterer intervention curricula, trauma-focused cognitive behavioral therapies, and community prevention initiatives.

Psychological Construct

The SVAWS and SVAMS operationalize intimate partner violence not as an isolated, homogeneous behavioral outburst, but as an overarching, multidimensional syndrome encompassing physical force, psychological intimidation, destruction of personal boundaries, and sexual violation. The construct rests on the empirical observation that violence operates across a severity gradient, wherein subjective fear and objective lethality compound as behaviors progress from symbolic displays to direct physical trauma. The 46 items are mapped onto two higher-order dimensions divided into nine distinct psychological and behavioral constructs:

Dimension 1: Threats of Violence (19 Items)

The Threats of Violence dimension captures actions, nonverbal gestures, verbal communications, and environmental destruction designed to intimidate, coerce, or instill fear of bodily injury, emotional ruin, or death. This dimension encompasses four hierarchical subscales:

• Symbolic Violence (Items 1–3): Involves physical actions directed not at the victim’s body, but at the surrounding physical environment to communicate destructive capacity and emotional volatility. Examples include kicking walls, smashing objects, and driving recklessly to terrorize the passenger. This construct represents an implicit warning of imminent physical capability.
• Mild Threats (Items 4–9): Encompasses overt interpersonal posturing and localized boundary violations that convey hostility without explicit statements of lethal harm. Examples include shaking a finger or fist at the partner, making menacing facial expressions or threatening gestures, acting like a bully, and destroying the victim’s personal belongings.
• Moderate Threats (Items 10–15): Involves explicit verbal and communicative commitments to inflict tangible harm. This includes threatening to damage cherished belongings, threatening third parties (e.g., children, family members, pets), direct verbal threats of physical injury, and threats of suicide as a manipulative weapon of coercion.
• Serious Threats (Items 16–19): Represents explicit, lethal, or weaponized threats designed to generate absolute terror. Examples include threatening the partner with a weapon, brandishing a club-like object, threatening with a knife or gun, or actively acting in a manner that communicates an intention to kill the partner.

Dimension 2: Acts of Violence (27 Items)

The Acts of Violence dimension measures the direct application of bodily force, physical contact, or physical violation against the partner’s person. It spans five subscales organized by objective tissue damage potential and lethality:

• Minor Violence (Items 20–25): Captures introductory physical contact intended to restrain, subdue, or physically dominate the partner without significant risk of severe skeletal or organ trauma. Examples include holding the partner down, pushing, shoving, forcefully grabbing, rough handling, scratching, and pulling hair.
• Moderate Violence (Items 26–31): Reflects elevated kinetic energy and deliberate intent to inflict pain, surface trauma, or bodily humiliation. Examples include twisting an arm, spanking, biting, open-hand slapping across the body or face, and back-handed slapping.
• Serious Violence (Items 32–35): Encompasses high-impact ballistic or blunt-force physical assaults carrying substantial risk of contusions, lacerations, bone fractures, and soft tissue damage. Specific behaviors include hitting with a blunt object, punching, kicking, and actively beating the partner up.
• Sexual Aggression (Items 36–41): Measures violations of bodily autonomy, sexual coercion, and sexually violent acts. This subscale ranges from verbal sexual entitlement and boundary violations (e.g., demanding sex regardless of consent, coercing unprotective sex) to forcible rape, forced participation in degrading or humiliating sexual acts, and using physical assault or weapons to compel sexual submission.
• Very Serious Violence (Items 42–46): Captures life-threatening, felonious assaults associated with catastrophic injury or immediate mortality risk. Behaviors include manual or ligature strangulation (choking), physical restraint or tying up, using club-like weapons or knives/guns directly upon the victim’s body, and inflicting severe bodily trauma requiring emergency medical care.

Theoretical Framework

The theoretical foundation of the SVAWS/SVAMS bridges several major psychological and sociological paradigms of interpersonal aggression, most notably Social Learning Theory, Coercive Control Theory, and the Psychometric Scaling of Social Phenomena.

Social Learning and Coercive Escalation

Grounded in Albert Bandura’s Social Learning Theory (1977) and Gerald Patterson’s coercion model (1982), Marshall conceptualized relationship violence as a behavioral continuum maintained through intermittent reinforcement and progressive escalation. Violent behaviors do not typically emerge fully formed as lethal assaults; rather, perpetrators often initiate coercive interactions using low-level, symbolic actions (e.g., punching walls, breaking personal keepsakes). When these symbolic behaviors succeed in silencing, intimidating, or controlling the partner, they become functionally reinforced. Over time, as the partner habituates to mild threats, the perpetrator escalates to overt physical contact, severe battering, and sexual coercion to re-establish dominance. The nine subscales of the SVAWS/SVAMS capture this developmental trajectory of coercive behavioral escalation.

Coercive Control and the Power-Control Continuum

Marshall’s conceptualization aligned with and anticipated Evan Stark’s (2007) formulation of Coercive Control. Rather than viewing domestic violence solely as discrete episodes of anger expression (“incident-specific violence”), the SVAWS emphasizes how non-physical acts (threats, destruction of property, menacing posturing) construct a pervasive “climate of fear.” In this structural reality, a look or a raised fist can subjugate a victim just as effectively as a physical blow, because prior physical assaults have established the credible lethality behind every threat. Thus, the deliberate inclusion of 19 threat items alongside 27 physical acts reflects the theoretical reality that intimidation and assault are functionally intertwined components of interpersonal subjugation.

Psychometric Continuum and Severity Weighting

From a psychometric perspective, the instrument draws from psychophysical scaling principles (Stevens, 1957) and Thurstone’s law of comparative judgment. Marshall recognized that simple frequency counts of violent events (e.g., summing 5 slaps and 5 knife assaults to yield a score of 10) are methodologically flawed because they ignore the staggering differences in trauma, terror, and physical sequelae associated with different acts. By empirically deriving severity levels through normative community ratings, the SVAWS/SVAMS operationalizes violence as an ordinally gradated, interval-like psychological construct where severity and frequency can be evaluated independently or synthesized into weighted composites.

Validity

The validity of the SVAWS and SVAMS has been subjected to rigorous psychometric scrutiny across university, clinical, community, shelter, and forensic populations over three decades.

Construct and Structural Validity

Construct validity was established in Marshall’s initial validation studies (1992a, 1992b) through extensive scaling procedures. Large samples of community women (N = 715) and men (N = 502) rated the severity of aggressive acts. Principal component and exploratory factor analyses revealed that items consistently organized into the bifurcated dimensions of Threats of Violence and Acts of Violence, with subscales loading onto coherent severity strata. Subsequent structural equation modeling and confirmatory factor analyses have confirmed that a hierarchical model—featuring nine lower-order factors loading onto two higher-order dimensions—yields superior fit compared to single-factor or undifferentiated multidimensional models.

Convergent Validity

Convergent validity is documented by strong correlations with established measures of relationship conflict and physical aggression:

• Conflict Tactics Scale (CTS & CTS2): SVAWS Physical Violence subscales correlate strongly with CTS Physical Assault scales (r = .70 to .86). The SVAWS Threats of Violence dimension correlates robustly with CTS Psychological Aggression (r = .68 to .78), while demonstrating higher diagnostic sensitivity for nuanced threats of bodily injury.
• Index of Spouse Abuse (ISA): SVAWS subscale scores show substantial convergence with Hudson and McIntosh’s ISA-Physical (ISA-P; r = .72 to .81) and ISA-Nonphysical (ISA-NP; r = .65 to .75) scores.
• Physical Injury Documentation: The Serious Violence and Very Serious Violence subscales correlate significantly with medical treatment seeking, photographic injury documentation, and validated injury scales (r = .58 to .72).

Discriminant Validity

The instrument successfully discriminates between general relationship discord and violent partner dynamics. Correlations between the SVAWS/SVAMS subscales and measures of general marital dissatisfaction (e.g., Dyadic Adjustment Scale) are moderate (r = .30 to .45), demonstrating that the instrument is not merely tapping into general marital unhappiness, but is capturing distinct aggressive behavioral pathology. Furthermore, the subscales successfully differentiate between clinical shelter-seeking populations, individuals undergoing outpatient marital counseling, and non-distressed community couples.

Criterion and Predictive Validity

Longitudinal studies demonstrate that the SVAWS/SVAMS possesses strong predictive validity for adverse mental health and safety outcomes. Baseline scores on the Threats of Violence and Sexual Aggression subscales prospectively predict:

• The onset and severity of Post-Traumatic Stress Disorder (PTSD) symptomatology, even after controlling for direct physical injury.
• Elevated scores on the Beck Depression Inventory (BDI-II) and Generalized Anxiety Disorder 7 (GAD-7) at 6- and 12-month follow-up evaluations.
• Future relationship dissolution, police involvement, protective order filings, and lethal/near-lethal assault re-victimization.

Reliability

The SVAWS and SVAMS demonstrate exceptional reliability across diverse populations, sociodemographic cohorts, and research designs.

Internal Consistency Reliability

In Marshall’s original validation samples (1992a, 1992b), internal consistency estimates (Cronbach’s alpha) were exceptionally high across the overarching dimensions and individual subscales:

• Total Scale (46 items): α = .95 to .98 across community and clinical samples.
• Threats of Violence Dimension (19 items): α = .92 to .95.
• Acts of Violence Dimension (27 items): α = .94 to .97.
• Symbolic Violence Subscale: α = .78 to .84.
• Mild Threats Subscale: α = .82 to .88.
• Moderate Threats Subscale: α = .85 to .91.
• Serious Threats Subscale: α = .88 to .93.
• Minor Violence Subscale: α = .81 to .87.
• Moderate Violence Subscale: α = .84 to .90.
• Serious Violence Subscale: α = .87 to .92.
• Sexual Aggression Subscale: α = .85 to .91.
• Very Serious Violence Subscale: α = .80 to .89.

Replication studies across clinical victim cohorts, university student samples, and male victim samples (SVAMS) consistently report alpha values exceeding the psychometric threshold of .80 for subscales and .90 for primary composite dimensions.

Test-Retest Reliability and Stability

Test-retest stability assessments conducted across short intervals (2- to 4-week test-retest designs) yield stability coefficients ranging from r = .82 to .91 for total victimization scores, indicating high measurement precision. In retrospective lifetime administrations, intraclass correlation coefficients (ICCs) remain robust (ICC = .79 to .87), confirming the cognitive clarity and stability of the behavioral recall prompts.

Factor Analysis

The structural taxonomy of the SVAWS/SVAMS has been extensively validated through exploratory factor analysis (EFA) and confirmatory factor analysis (CFA).

Exploratory Factor Structure

During the initial development of the scale, Marshall (1992a) subjected the behavioral items to principal components analysis (PCA) with oblimin and varimax rotations. Two primary factors clearly emerged from the correlation matrices, accounting for the majority of the common variance:

• Factor 1: Direct Physical and Sexual Assault (Acts of Violence): Characterized by high factor loadings (.55 to .85) for physical strikes, weapons use, sexual violations, and soft tissue trauma.
• Factor 2: Intimidation and Hostile Posturing (Threats of Violence): Defined by high factor loadings (.50 to .82) for environmental destruction, menacing gestures, verbal threats to injure or kill, and weapon brandishing.

Secondary factor analyses within each primary dimension isolated the nine specific severity subscales, confirming that items grouped reliably according to their underlying severity gradients.

Confirmatory Factor Analysis and Model Fit

Subsequent psychometric investigations have tested competing structural models using modern confirmatory factor analysis (CFA) techniques. Typical fit statistics for the hierarchical nine-factor, two-dimensional model demonstrate excellent fit to empirical data:

• Comparative Fit Index (CFI): .93 to .96, exceeding standard thresholds for robust structural fit.
• Tucker-Lewis Index (TLI): .92 to .95.
• Root Mean Square Error of Approximation (RMSEA): .042 to .058 (with 90% confidence intervals between .038 and .062), indicating close approximation to population parameters.
• Standardized Root Mean Square Residual (SRMR): .045 to .053.

Standardized item factor loadings across all nine subscales generally exceed .60, with the vast majority ranging between .68 and .88. Invariance testing across gender has demonstrated scalar and metric equivalence for the overarching two-factor model, confirming that the scale validly measures identical psychological constructs across female and male respondents, while maintaining the structural differentiation of threat versus physical action.

Instrument / Measurement Tool

• Test Title: Severity of Violence Against Women Scale / Severity of Violence Against Men Scale (SVAWS / SVAMS)
• Author: Linda L. Marshall, Ph.D. (1992)
• Test Type: Psychometric Self-Report Behavioral Inventory (Victimization or Perpetration form)
• Total Item Count: 46 Items
• Format / Structure:
  • Dimension 1: Threats of Violence (19 items)
    • Subscale 1: Symbolic Violence (Items 1–3)
    • Subscale 2: Mild Threats (Items 4–9)
    • Subscale 3: Moderate Threats (Items 10–15)
    • Subscale 4: Serious Threats (Items 16–19)
  • Dimension 2: Acts of Violence (27 items)
    • Subscale 5: Minor Violence (Items 20–25)
    • Subscale 6: Moderate Violence (Items 26–31)
    • Subscale 7: Serious Violence (Items 32–35)
    • Subscale 8: Sexual Aggression (Items 36–41)
    • Subscale 9: Very Serious Violence (Items 42–46)
• Authentic Response Scale: 4-point Likert-type frequency scale:
  • 1 = Never
  • 2 = Once
  • 3 = A few times
  • 4 = Many times
• Administration Modality: Paper-and-pencil questionnaire, confidential computer-assisted self-interview (CASI), or secure clinical interview. Typically administered with a defined recall window (e.g., past 12 months, during relationship).
• Scoring and Computational Procedures:
  • Raw Frequency Sums: Subscale scores are calculated by summing the numerical ratings (1 to 4) of all constituent items. Higher sums indicate greater frequency and severity of abuse.
  • Dimensional Composite Scores: Threats of Violence Score (sum of items 1–19, range 19–76); Acts of Violence Score (sum of items 20–46, range 27–108); Total Scale Score (sum of items 1–46, range 46–184).
  • Mean Frequency Metric: Scores may also be expressed as mean item responses (total sum divided by number of items), preserving the intuitive 1.0 to 4.0 metric.
  • Weighted Severity Scoring (Optional): Researchers may multiply item frequency scores by Marshall’s empirically derived severity weights to generate composite severity indices that assign disproportionately higher weight to life-threatening acts.
• Target Demographics: Adults and late adolescents (ages 16+) engaged in dating, cohabiting, married, or recently separated romantic partnerships.

Permissions & Fee and Test Year

The Severity of Violence Against Women Scale and Severity of Violence Against Men Scale were published by Dr. Linda L. Marshall in 1992 in the peer-reviewed literature:

• Original Publications: Journal of Family Violence, Volume 7, Issue 2 (SVAWS, pp. 103–121) and Volume 7, Issue 3 (SVAMS, pp. 189–203).
• Licensing and Accessibility: The scale is in the public domain for academic research, education, and clinical practice, having been disseminated extensively as an open-access public health measurement instrument. The full scale and administration guidelines are included in the public compendium published by the Centers for Disease Control and Prevention (CDC): Measuring Intimate Partner Violence Victimization and Perpetration: A Compendium of Assessment Tools (Dahlberg et al., 2005; Thompson et al., 2006).
• User Fees: There are no licensing fees, purchase costs, or proprietary royalties required for non-commercial academic research, shelter assessments, or clinical diagnostic evaluations. Researchers and clinicians must cite the original development papers (Marshall, 1992a, 1992b) when using or reproducing the instrument.

References

• Bandura, A. (1977). Social Learning Theory. Prentice-Hall.
  https://www.worldcat.org/title/social-learning-theory/oclc/2644336
• Dahlberg, L. L., Toal, S. B., Swahn, M., & Behrens, C. B. (2005). Measuring Violence-Related Attitudes, Behaviors, and Influences Among Youths: A Compendium of Assessment Tools (2nd ed.). Centers for Disease Control and Prevention, National Center for Injury Prevention and Control.
  https://www.cdc.gov/violenceprevention/pdf/ipv_compendium.pdf
• Hudson, W. W., & McIntosh, S. R. (1981). The assessment of spouse abuse: Two quantifiable dimensions. Journal of Marriage and the Family, 43(4), 873–885.
  https://doi.org/10.2307/351344
• Marshall, L. L. (1992a). Development of the Severity of Violence Against Women Scale. Journal of Family Violence, 7(2), 103–121.
  https://doi.org/10.1007/BF00978700
• Marshall, L. L. (1992b). The Severity of Violence Against Men Scale. Journal of Family Violence, 7(3), 189–203.
  https://doi.org/10.1007/BF00994617
• Patterson, G. R. (1982). Coercive Family Process. Castalia Publishing Company.
  https://www.worldcat.org/title/coercive-family-process/oclc/8763428
• Stark, E. (2007). Coercive Control: How Men Entrap Women in Personal Life. Oxford University Press.
  https://doi.org/10.1093/acprof:oso/9780195179439.001.0001
• Stevens, S. S. (1957). On the psychophysical law. Psychological Review, 64(3), 153–181.
  https://doi.org/10.1037/h0046162
• Straus, M. A. (1979). Measuring intrafamily conflict and violence: The Conflict Tactics (CT) Scales. Journal of Marriage and the Family, 41(1), 75–88.
  https://doi.org/10.2307/351733
• Thompson, M. P., Basile, K. C., Hertz, M. F., & Sitterle, D. (2006). Measuring Intimate Partner Violence Victimization and Perpetration: A Compendium of Assessment Tools. Centers for Disease Control and Prevention, National Center for Injury Prevention and Control.
  https://www.cdc.gov/ncipc/pub-res/IPV_Compendium.pdf

13. Items of the Scale (Questionnaire)