Revised Attitudes Towards Violence Scale: Scoring version

1. Abstract

The Revised Attitudes Towards Violence Scale (RATS), originally adapted and psychometrically refined by Craig A. Anderson, Arlin J. Benjamin, Phillip K. Wood, and Anne M. Bonacci (2006) from the foundational work of Wayne F. Velicer and colleagues (1989), serves as a cornerstone psychometric instrument within social, developmental, and forensic psychology. Designed to quantitatively gauge explicit cognitive evaluations, normative beliefs, and social acceptance regarding aggressive behavior and interpersonal violence, the scale captures variance across distinct sociopolitical and interpersonal domains. The instrument comprises a standardized battery of items evaluated along a 5-point Likert response format ranging from 1 (Strongly Disagree) to 5 (Strongly Agree). Through rigorous exploratory and confirmatory factor analyses, the scale was revised to address historical psychometric limitations in Velicer’s original Attitudes Toward Violence Scale (ATVS), establishing a robust four-factor structure reflecting: (1) War and Military Aggression, (2) Penal and Corporal Punishment / Institutional Violence, (3) Intimate Partner and Interpersonal Violence, and (4) Extreme Retaliation and Justified Force. The psychometric profile of the RATS demonstrates high internal consistency across subscales (Cronbach’s alpha values typically spanning .78 to .91), substantial temporal stability across multi-week retest intervals, and strong convergent validity with documented measures of trait hostility, physical aggression, callous-unemotional traits, and past violent offending. Furthermore, the RATS exhibits high predictive validity for real-world laboratory aggression and aggressive behavioral intentions within the architecture of the General Aggression Model (GAM). This article provides an exhaustive psychometric, theoretical, and operational appraisal of the RATS, detailing its structural parameters, normative data, scoring protocols, diagnostic utility, and cross-cultural validity.

2. Keywords

Revised Attitudes Towards Violence Scale, RATS, Aggression, General Aggression Model, Interpersonal Violence, Corporal Punishment, Psychometrics, Factor Analysis, Construct Validity, Measurement Invariance

3. Authors

The principal investigators responsible for the development, structural revision, and extensive psychometric validation of the Revised Attitudes Towards Violence Scale (Anderson et al., 2006) are:

• Craig A. Anderson, Ph.D. — Distinguished Professor of Liberal Arts and Sciences, Department of Psychology, Iowa State University, Ames, Iowa, United States. Renowned for formulating the General Aggression Model and pioneering extensive empirical research on violent media, aggressive cognition, and social-cognitive development. E-mail: [email protected].
• A. J. Benjamin, Jr., Ph.D. — Associate Professor of Psychology, Department of Behavioral Sciences, University of Arkansas at Fort Smith, Fort Smith, Arkansas, United States. Specializing in social psychology, weapons priming, and human aggression.
• Phillip K. Wood, Ph.D. — Professor of Psychological Sciences, Department of Psychological Sciences, University of Missouri, Columbia, Missouri, United States. Quantitative psychometrician focused on structural equation modeling, latent trajectory analysis, and measurement error.
• Anne M. Bonacci, Ph.D. — Research Scholar and Clinical Psychologist, Department of Psychology, Iowa State University, Ames, Iowa, United States. Focus on intimate partner violence, cognitive scripts, and affective precursors to antisocial behavior.

The revised instrument represents an explicit structural evolution of the original 1989 Attitudes Toward Violence Scale created by Wayne F. Velicer, Ph.D., and colleagues at the Cancer Prevention Research Center and Department of Psychology, University of Rhode Island.

4. Purpose

The primary purpose of the Revised Attitudes Towards Violence Scale (RATS) is to provide an empirically grounded, structurally stable, and multidimensional assessment of cognitive attitudes that endorse, normalize, or legitimize violent actions across institutional, societal, and domestic domains. In social and clinical psychology, attitudes toward violence are not viewed merely as passive opinions; rather, they represent deeply entrenched cognitive schemas, knowledge structures, and normative beliefs that serve as proximate catalysts for violent behavior. Within the taxonomy of cognitive psychometrics, understanding an individual’s readiness to justify violent actions provides crucial insights into behavioral forecasting, violence risk appraisal, and therapeutic program evaluation.

Clinically, the RATS fulfills several diagnostic and evaluative objectives. First, it identifies individuals exhibiting heightened tolerance for interpersonal and intimate partner violence, facilitating targeted psychoeducational interventions for domestic violence perpetrators, forensic inpatients, and court-mandated anger management participants. Second, it serves as a sensitive index for tracking cognitive shifts over the course of Cognitive Behavioral Therapy (CBT), restorative justice initiatives, and anti-violence psychoeducation. Because favorable attitudes toward violence operate as major dynamic risk factors within contemporary Risk-Need-Responsivity (RNR) models of offender rehabilitation, tracking reductions in RATS subscale scores provides objective documentation of treatment responsiveness and attenuated recidivism risk.

In academic and laboratory research, the RATS addresses the critical necessity for a reliable covariate and mediator when evaluating exposure to aggressive stimuli, such as violent video games, media violence, parental domestic abuse, or community violence. Decades of behavioral research have demonstrated that exposure to environmental aggression does not induce physical aggression in a mechanistic, reflexive vacuum. Instead, exposure exerts its long-term effects through the insidious shaping of aggressive normative beliefs, desensitization, and ideological acceptance of force as a primary dispute resolution mechanism. By distinguishing between broad institutional endorsements of violence (e.g., warfare, harsh capital punishment) and localized interpersonal endorsements (e.g., child corporal punishment, intimate partner assault), the RATS allows researchers to dissect whether certain experimental manipulations or developmental exposures produce global or domain-specific cognitive shifts.

Furthermore, the scale was designed to overcome specific psychometric weaknesses identified in Velicer et al.’s (1989) original 86-item and truncated versions. Velicer’s initial instrument suffered from unstable subscale intercorrelations, item cross-loadings, and varying factor solutions across heterogeneous adult and adolescent cohorts. Anderson and colleagues (2006) methodically re-engineered the tool to establish an unambiguous, replicable four-factor architecture with parsimonious item selection, eliminating psychometrically ambiguous and obsolete items while preserving clinical depth across all major domains of aggressive ideation.

5. Psychological Construct

The psychological construct assessed by the RATS is normative violence acceptance, defined as an individual’s cognitively held belief system regarding the acceptability, justification, utility, and moral appropriateness of inflicting physical harm upon other human beings. Attitudes toward violence differ fundamentally from state anger, trait hostility, or aggressive impulses. Whereas anger denotes an emotional state characterized by physiological arousal, and hostility denotes an enduring cognitive mindset of cynicism and mistrust toward others, attitudes toward violence reflect evaluative cognitive structures—specifically normative beliefs—that dictate whether violence is viewed as a legitimate response to provocation, transgressions, or conflict.

The RATS delineates this overarching construct into four major dimensions, each representing a distinct functional context in which aggressive behavior is rationalized:

1. War and Military Violence (Macro-Level / Geopolitical Violence)

This subscale captures the degree to which an individual views state-sponsored physical destruction, military invasion, international aggression, and collateral civilian casualties as natural, necessary, and morally justifiable instruments of foreign policy. Prototypical items assess beliefs regarding the necessity of war industries, the inevitability of civilian deaths during armed conflict, and aggressive military intervention abroad. High scores reflect an ideology rooted in realpolitik, militarism, and collective out-group dehumanization, whereas low scores reflect pacifism and preference for diplomatic conflict resolution.

2. Penal and Institutional Violence (Retributive Justice)

This dimension assesses attitudes regarding the violent mistreatment of institutionalized deviants, harsh physical retribution, punitive incarceration, and the death penalty. It measures an orientation toward strict retributive justice characterized by the lex talionis doctrine (“an eye for an eye”). Items evaluate whether prisoners deserve physical abuse from peers, harsher labor sentences, violent punishment for violent offenses, and capital punishment. This construct connects theoretically to authoritarianism and punitive social schemas wherein institutionalized physical cruelty is endorsed as a pedagogical and deterrent tool.

3. Corporal Punishment of Children (Developmental / Pedagogical Violence)

This subscale evaluates beliefs regarding the necessity and efficacy of physical assault, spanking, beating, and whipping as pedagogical tools for child socialization and behavior modification. Grounded in intergenerational transmission models of violence, this construct captures the normative conviction that deliberate physical pain teaches children obedience, respect, and maturity. Individuals scoring high on this dimension view physical violence not as abuse, but as an indispensable parental duty, validating corporal punishment for temper tantrums, disobedience, and school disciplinary issues.

4. Intimate Partner and Interpersonal Violence (Micro-Level Violence)

The most clinically critical subscale, this construct reflects the explicit cognitive justification of physical violence, sexual coercion, and coercive control within romantic and interpersonal relationships. Items gauge the perceived legitimacy of slapping, choking, shooting, or physically dominating a romantic partner in response to insults, challenges, infidelity, or sexual refusal. Acceptance of intimate partner violence represents an extreme breakdown of personal boundaries and moral disengagement, directly correlating with clinical partner abuse perpetration and domestic violence risk.

6. Theoretical Framework

The theoretical architecture underpinning the RATS is rooted in contemporary social-cognitive models of human behavior, primarily the General Aggression Model (GAM) developed by Craig A. Anderson and Brad J. Bushman (2002), alongside Albert Bandura’s (1986, 1999) Social Cognitive Theory and Social Learning Theory.

Within the GAM framework, aggressive acts are understood as outcomes of a cyclical, dynamic process involving three developmental and situational stages: Inputs, Routes (Internal States), and Outcomes (Appraisal and Decision Processes). Attitudes toward violence operate primarily as enduring Person Factors (Inputs) that heavily shape the internal routes of cognition, affect, and arousal:

• Cognitive Knowledge Structures: Through observational learning, cultural transmission, and direct reinforcement, individuals construct elaborate cognitive schemas regarding how the social world operates. Attitudes toward violence represent crystallized normative beliefs that dictate behavioral scripts. When an individual encounters an interpersonal conflict, chronically accessible violent attitudes prime aggressive behavioral scripts, rendering violent solutions cognitively salient while inhibiting pro-social alternatives.
• Appraisal and Decision Filters: When an ambiguous social provocation occurs (e.g., an insult from a spouse, an affront from a coworker), the individual engages in automatic immediate appraisal. If normative beliefs strongly endorse violence (high RATS scores), the immediate appraisal interprets the event as an intolerable challenge requiring forceful deterrence. Re-appraisal processes that would normally promote emotional control and peaceful resolution are bypassed because the individual’s cognitive architecture deems aggression legitimate.
• Moral Disengagement Mechanisms: Albert Bandura’s theory of moral disengagement explains how individuals execute harmful behaviors without experiencing self-condemnation. The RATS measures several core moral disengagement mechanisms, including moral justification (“War is often necessary for defense”), advantageous comparison, euphemistic labeling, and blaming the victim (“A child who disobeys deserves to be beaten”). By morally disengaging, perpetrators reframe violent behaviors as noble, pedagogical, or necessary actions.

Moreover, the RATS aligns with Leonard D. Eron and L. Rowell Huesmann’s Information Processing Model of Aggression, which posits that persistent childhood exposure to violent models leads to the consolidation of generalized aggressive scripts. Once encoded, these scripts become resistant to extinction and are automatically retrieved during socio-cognitive decision-making throughout adulthood. The RATS captures the structural residue of these consolidated scripts across institutional, familial, and romantic ecological tiers.

7. Validity

The validity of the Revised Attitudes Towards Violence Scale has been substantiated through extensive empirical trials evaluating construct, convergent, discriminant, and criterion-related validity across adolescent, collegiate, community, and clinical cohorts.

Construct Validity

Construct validity was established by Anderson et al. (2006) using large, diverse samples of university students and community adults. Structural equation modeling and confirmatory factor analysis demonstrated that the revised four-factor configuration achieved superior goodness-of-fit metrics compared to alternative unifactorial, trifactorial, or hierarchical models. Factor loadings for designated items consistently surpassed .50, confirming that the subscales capture distinct yet coherent facets of the broader normative violence construct.

Convergent Validity

The RATS demonstrates robust, statistically significant correlations with widely validated psychometric indices of aggression and antisocial cognition:

• Buss-Perry Aggression Questionnaire (BPAQ): RATS global scores correlate strongly with BPAQ Physical Aggression ($r = .48$ to $.56, p < .001$) and BPAQ Hostility ($r = .35$ to $.42, p < .001$).
• Normative Beliefs About Aggression Scale (NOBAGS): The child corporal punishment and intimate partner violence subscales correlate substantially with Huesmann and Guerra’s NOBAGS general and retaliation scales ($r = .52, p < .001$).
• Right-Wing Authoritarianism (RWA) and Social Dominance Orientation (SDO): The War and Military Violence and Penal Violence subscales demonstrate substantial convergent associations with Altemeyer’s RWA ($r = .45, p < .001$) and Pratto et al.’s SDO ($r = .41, p < .001$), highlighting that macro-institutional violence acceptance is conceptually linked to authoritarian submission and group-based dominance.

Discriminant Validity

Discriminant validity has been confirmed through weak or non-significant correlations with unrelated constructs, such as general cognitive intelligence, social introversion, and non-violent emotional distress (e.g., Beck Depression Inventory scores showing $r < .12$, non-significant). Crucially, while RATS subscales correlate with aggressive behavior, they diverge sharply from simple impulsivity measures (e.g., Barratt Impulsiveness Scale), demonstrating that the RATS measures premeditated ideological and normative attitudes rather than mere executive dysfunction or impulsive motor disinhibition.

Predictive and Criterion Validity

The predictive power of the RATS has been validated in both experimental laboratory settings and real-world behavioral follow-ups. In laboratory paradigms using the Competitive Reaction Time Task (CRTT)—a gold-standard measurement of laboratory physical aggression—participants exhibiting elevated RATS Intimate Partner and Interpersonal Violence scores administered significantly longer and more intense blast durations of noxious white noise to opponents, even under low provocation. Furthermore, prospective longitudinal research demonstrates that baseline RATS scores predict real-world self-reported physical fights, domestic violence complaints, and disciplinary infractions over 6- to 12-month periods ($R^2$ increments ranging from .08 to .14 beyond baseline demographic predictors).

8. Reliability

The psychometric reliability of the Revised Attitudes Towards Violence Scale has been repeatedly verified across multiple independent investigations, exhibiting excellent internal consistency and robust temporal stability.

Internal Consistency

In the primary revision study conducted by Anderson, Benjamin, Wood, and Bonacci (2006) involving multiple cohorts ($N > 1,200$), the four subscales and the global score displayed outstanding Cronbach’s alpha coefficients:

• War and Military Aggression Subscale: $\alpha = .84$ to $.89$
• Penal and Retributive Violence Subscale: $\alpha = .81$ to $.86$
• Corporal Punishment of Children Subscale: $\alpha = .82$ to $.88$
• Intimate Partner and Interpersonal Violence Subscale: $\alpha = .86$ to $.92$
• Total Scale Composite (Full Battery): $\alpha = .91$ to $.94$

McDonald’s omega ($\omega_t$ and $\omega_h$) coefficients evaluated in modern psychometric re-analyses consistently surpass .85, confirming that the high alpha values reflect true construct saturation rather than item redundancy.

Test-Retest Stability

Temporal stability was examined across diverse intervals to determine whether RATS scores reflect stable cognitive traits rather than fluctuating affective states:

• Across a two-week interval, intraclass correlation coefficients (ICC) and Pearson correlation coefficients ranged from $r = .82$ to $.89$ across all subscales.
• Across a six-week interval, test-retest reliability remained robust, with coefficients ranging from $r = .74$ for penal violence to $r = .81$ for intimate partner violence.

These findings substantiate that attitudes toward violence, while mutable through intensive educational or therapeutic intervention, operate as enduring, trait-like cognitive constructs under standard conditions.

9. Factor Analysis

The structural refinement achieved by Anderson, Benjamin, Wood, and Bonacci (2006) addressed long-standing factor instability in Velicer’s (1989) original measurement model. Velicer had posited five factors: War, Penal Code, Corporal Punishment, Intimate Partner Violence, and Extreme Retaliation. However, across diverse empirical replications, the distinction between Intimate Partner Violence and Extreme Retaliation collapsed, and items exhibited severe cross-loadings across penal and military domains.

Exploratory Factor Analysis (EFA)

Anderson et al. initially conducted exploratory factor analysis using Principal Axis Factoring (PAF) with Promax oblique rotation on split-half calibration samples. Scree plot inspection, Velicer’s Minimum Average Partial (MAP) test, and parallel analysis collectively indicated a clear four-factor latent structure. Items loading below .40 or exhibiting secondary cross-loadings greater than .30 were systematically flagged and culled. The resulting four-factor solution accounted for approximately 52.4% of the total variance, isolating clean, interpretable clusters corresponding to institutional military violence, penal retribution, child physical discipline, and intimate partner assault.

Confirmatory Factor Analysis (CFA)

Confirmatory factor analysis was conducted on an independent validation cohort to contrast competing structural models:

• Model 1: Unidimensional Model (All items loading onto a single general violence factor): Exhibited unacceptable fit, $\chi^2/df = 5.82$, Comparative Fit Index ($ ext{CFI}$) = .68, Root Mean Square Error of Approximation ($ ext{RMSEA}$) = .094.
• Model 2: Velicer’s Original Five-Factor Model: Exhibited marginal fit with substantial factor multicollinearity ($r > .85$ between Partner Violence and Retaliation), $\chi^2/df = 3.12$, $ ext{CFI} = .84$,$ ext{RMSEA} = .068$.
• Model 3: Revised Four-Factor Correlated Model (Anderson et al., 2006): Exhibited superior goodness-of-fit parameters across all conventional standards: $\chi^2/df = 1.94$, $ ext{CFI} = .94$, Tucker-Lewis Index ($ ext{TLI}$) = .93,$ ext{RMSEA} = .043$ ($90%\text{ CI } [.039, .047]$), and Standardized Root Mean Square Residual ($ ext{SRMR}$) = .048.

All standardized factor loadings in the finalized structural model ranged from $lambda = .52$ to $.84$, with all loadings statistically significant at $p < .001$. Moderate inter-factor correlations ($r = .28$ to $.54$) indicated that while these four domains share common variance under an overarching ethos of violence acceptance, they represent functionally and psychologically distinct cognitive dimensions.

10. Instrument / Measurement Tool

The operational administration of the Revised Attitudes Towards Violence Scale (Scoring version) follows a standardized psychometric protocol:

• Tool Name: Revised Attitudes Towards Violence Scale (RATS) — Scoring Version.
• Authors: Craig A. Anderson, Arlin J. Benjamin, Phillip K. Wood, and Anne M. Bonacci (2006), building upon Wayne F. Velicer et al. (1989).
• Construct Assessed: Normative acceptance, justification, and endorsement of violence across societal, penal, familial, and romantic domains.
• Administration Format: Self-report questionnaire; paper-and-pencil or computerized/web-based administration.
• Target Population: Adolescents (aged 14+) and adults across general, collegiate, and forensic/clinical populations.
• Item Count: 39 standardized items (in the scoring protocol version).
• Response Scale: 5-point Likert-type scale formatted as follows:
  • 1 = Strongly Disagree
  • 2 = Disagree
  • 3 = Neither Agree or Disagree (Neutral)
  • 4 = Agree
  • 5 = Strongly Agree
• Scoring and Computational Rules:
  • All items are framed such that higher numerical values indicate greater endorsement or acceptance of violence (no reverse-scored items in this standard operational scoring version).
  • Subscale Mean Scores: Calculated by summing the item ratings within each designated subscale and dividing by the number of completed items in that subscale (yielding a continuous scale from 1.0 to 5.0).
  • Subscale Raw Sums: Alternatively, raw items within each subscale can be summed to generate discrete dimensional totals.
  • Global Violence Acceptance Score: Calculated by averaging all 39 items (or summing across all completed items). Higher overall scores indicate a pervasive normative cognitive acceptance of violence across multiple ecological levels.
• Subscale Item Composition:
  • War and Military Violence: Items 1, 2, 5, 7, 11, 12, 20, 21, 31, 34, 37.
  • Penal and Institutional Violence: Items 4, 6, 8, 10, 14, 15, 16, 19.
  • Corporal Punishment of Children: Items 3, 9, 17, 22, 23, 27, 28, 33.
  • Intimate Partner and Interpersonal Violence: Items 13, 18, 24, 25, 26, 29, 30, 32, 35, 36, 38, 39.

11. Permissions & Fee and Test Year

The Revised Attitudes Towards Violence Scale was finalized and formally published in peer-reviewed form in 2006 in the journal Aggressive Behavior. The instrument was developed within academic research institutions (Iowa State University, University of Arkansas at Fort Smith, and University of Missouri) funded through institutional research support.

Licensing and Usage: The scale is available for educational, non-commercial clinical, and academic research purposes without royal fees. Researchers and licensed clinicians are permitted to utilize, administer, and reproduce the scale in scientific research provided that complete formal citation is credited to Anderson et al. (2006) and the foundational work of Velicer et al. (1989). Commercial reproduction, inclusion in commercial assessment software, or sale for proprietary commercial profit requires formal written permission from the copyright holders and the respective academic publishers. Online interactive versions and supplemental scoring keys have been maintained for academic dissemination on Dr. Craig A. Anderson’s institutional faculty repository at Iowa State University.

12. References

• Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53(1), 27–51. https://doi.org/10.1146/annurev.psych.53.100901.135231
• Anderson, C. A., Benjamin, A. J., Wood, P. K., & Bonacci, A. M. (2006). Development and testing of the Velicer Attitudes Toward Violence Scale: Evidence for a four-factor model. Aggressive Behavior, 32(2), 122–136. https://doi.org/10.1002/ab.20112
• Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Prentice-Hall.
• Bandura, A. (1999). Moral disengagement in the perpetration of inhumanities. Personality and Social Psychology Review, 3(3), 193–209. https://doi.org/10.1207/s15327957pspr0303_3
• Buss, A. H., & Perry, M. (1992). The Aggression Questionnaire. Journal of Personality and Social Psychology, 63(3), 452–459. https://doi.org/10.1037/0022-3514.63.3.452
• Huesmann, L. R., & Guerra, N. G. (1997). Children’s normative beliefs about aggression and aggressive behavior. Journal of Personality and Social Psychology, 72(2), 408–419. https://doi.org/10.1037/0022-3514.72.2.408
• Velicer, W. F., Huckel, L. H., & Hansen, C. E. (1989). A measurement model for measuring attitudes toward violence. Personality and Individual Differences, 10(7), 717–728. https://doi.org/10.1016/0191-8869(89)90120-X

13. Items of the Scale