The Rogers Criminal Responsibility Assessment Scales (R-CRAS) is a structured decision model for quantifying relevant psychological variables that are salient for the retrospective evaluation of insanity. The R-CRAS was validated to address specifically the American Law Institute (ALI) insanity standard that requires an assessment of a defendant’s cognitive and volitional impairment at the time of the alleged offense. In addition to the ALI standard, the R-CRAS provides clinical data relevant to the M’Naghten insanity standard and the Michigan-based guilty but mentally ill (GBMI) standard.

The R-CRAS decision process combines an appraisal of general diagnostic categories with an assessment of cognitive and behavioral (i.e., volitional) abilities at the time of the offense. Three rationally constructed scales evaluate diagnostic issues: (1) Patient Reliability, which includes malingering or involuntary interference with accurate recall; (2) Organicity, which addresses the likely effects of brain damage or mental retardation; and (3) Psychiatric Disorders, which examines the effects of key Axis I symptoms. Two scales address legally relevant impairment: (1) Cognitive Control, which evaluates impairment in verbal abilities, awareness of the criminal behavior, and capacity for planning; and (2) Behavioral Control, which considers level and focus of criminal activity, as well as the defendant’s capacity to control criminal behavior and engage in responsible behavior. Finally, the GBMI items address general domains of impairment that are not specific to criminal behavior.

Description and Development

The R-CRAS was developed by a study group of five experienced forensic psychologists and psychiatrists who reached consensual agreement for the inclusion of core psychological and situational variables relevant to insanity. This operationalization of the ALI standard requires forensic experts to make a series of professional judgments based on anchored ratings. These ratings provided descriptions of impairment for each level that typically ranged from “none” and “slight” to “moderate,” “severe,” and sometimes “extreme.”

The primary standardization sample was drawn from two well-established outpatient forensic centers: the Isaac Ray Center in Chicago and the Court Diagnostic and Treatment Center in Toledo. Using eight forensic psychologists and eight forensic psychiatrists, R-CRAS data were collected on a total of 157 insanity evaluations with test-retest reliability on 76 cases. A secondary sample of 103 insanity referrals was collected from two inpatient and two outpatient forensic facilities.


The reliability of the R-CRAS is challenging to establish, given the retrospective nature of insanity evaluations. As a rigorous test of its reliability, the R-CRAS was administered by independent evaluators on separate occasions with an average interval of 2.7 weeks. For individual variables, the mean reliability coefficient was .58, which is very acceptable given the rigorousness of the retrospective test-retest design.

One critical issue is the reliability of the ALI decision model to demonstrate the reproducibility of R-CRAS decision variables by independent experts at separate times. Average agreement between experts was very high with values ranging from 85% to 100% (M = 91%). Kappa coefficients were generally excellent with an average of .81. Because kappas are affected by low base rates, the kappa for malingering was modest (.48) despite the high level of agreement (85%). For the final decision regarding insanity, the agreement between independent experts was almost perfect (97%; k = .94).


The R-CRAS used Loevinger’s model of construct validation, which is conceptualized in terms of substantive, structural, and external validities. The development of the R-CRAS addressed substantive validity in its selection and operationalization of key variables relevant to insanity. The structural validity used a formulation of insanity that could be tested as a series of hypotheses. In comparison with sane defendants, clinically evaluated insane defendants would manifest (a) a relative absence of malingering, (b) greater psychological impairment (i.e., organicity and mental disorders), and (c) greater impairment (i.e., cognitive and volitional). Marked differences were observed in the predicted direction. For example, very large effect sizes were found for the role of hallucinations (Cohen’s d = 1.80) and delusions (Cohen’s d = 3.15) in criminal behavior. Discriminant analyses were also used to demonstrate differentiating patterns between sane and insane defendants on calibration and cross-validation samples. As evidence of external validity, R-CRAS decisions were compared with legal outcomes. The primary samples yielded high concordance rates that were nearly identical—88.5% for the Isaac Ray Center and 88.1% for the Court Diagnostic and Treatment Center.

Forensic Applications

The R-CRAS is the only well-validated decision model for the assessment of criminal responsibility. Its model requires forensic psychologists and psychiatrists to quantify key variables related to the severity of Axis I symptoms and elements of criminal behavior. This model appears to be generalizable to defendants with different sociodemographic (e.g., gender, race, age, and education), criminal (e.g., prior arrests and delinquency), and clinical (e.g., prior diagnoses and hospitalizations) variables.

Forensic psychologists may prefer to use the R-CRAS for insanity evaluations as a structured guide rather than a formal test. This use appears warranted, especially in jurisdictions that do not use the ALI insanity standard. For GBMI consultations, the Michigan-based R-CRAS criteria are used in many jurisdictions. However, forensic clinicians should carefully check its relevance to their particular jurisdiction. Finally, experts are likely to be divided on whether to render conclusory opinions with insanity cases. Forensic psychologists avoiding conclusory opinions should use the final decision point for sanity as simply advisory and not document it in their forensic reports.


  1. Grisso, T. (2003). Evaluating competencies: Forensic assessments and instruments. New York: Kluwer.
  2. Rogers, R. (1984). Rogers criminal responsibility assessment scales (R-CRAS) and test manual. Odessa, FL: Psychological Assessment Resources.
  3. Rogers, R., & Sewell, K. W. (1999). The R-CRAS and insanity evaluations: A re-examination of construct validity. Behavioral Sciences and the Law, 17, 181-194.
  4. Rogers, R., & Shuman, D. W. (2000). Conducting insanity evaluations (2nd ed.). New York: Guilford Press.

Return to Criminal Responsibility assessment overview.