CARE-2 is one of the most accurate identifiers of youth at risk for violence

Research on the CARE-2 Risk & Needs Assessment

An Analysis of the CARE-2's Predictive Ability PDF Icon

Attachment and the Development of Relationship Skills PDF Icon

CARE-2 Assessment Results, Year 2 Word Document

THE SAMPLE
To establish the statistical characteristics of the revised CARE-2 Assessment Form, the form was administered to 1,026 individuals. Of that number, 369 were children (ages 2 to 12), 636 were adolescent (ages 13 to 19), and 21 with missing data. From the sample, 677 were male and 328 were female, and 21 of the profiles were missing data. In terms of estimated intelligence, 72% of the sample (739) had average intelligence, 19% (195) had above average intelligence, 5% (51) had below average intelligence and 4% had missing data. Table 3.1 provides a summary of the number of individuals within each age group by gender. Data was collected from youth in outpatient treatment programs, residential settings, juvenile detention settings, group home facilities, and school settings. For comparison purposes data was also collected from a subset of youth with no reported behavioral problems. The main locations for data collection were the Mid-Atlantic and the Midwestern regions of the United States. Youth placement varied as well, with 72% (739) living at home, 6% (62) in detention, 7% (72) in residential treatment placement, 4% (41) split between all other placement locations (e.g., foster care, independent living, group home placement, hospital, or prison), and 5% (51) had data missing. In terms of ethnicity, the sample was primarily Caucasian (544, 53%) and African-American/Black (380, 37%). Other ethnic groups included Hispanic (26, 2.5%), Hispanic and Other ethnic (75, 7.5%), and 2.5% were missing data. To determine if the ethnic composition of this sample was similar to the U.S. population of youth with serious behavior problems, the national statistics were reviewed. Nationally, Caucasian juveniles constituted 71% of all juvenile arrests and African American juveniles constitute 26%. Other ethnicities account for an additional 3% of the population. Fifty-five percent of violent crimes were committed by Caucasian youth and 43% were committed by African American youth (Snyder and Sickmund, 1999).

The sample was then assessed for other characteristics including trauma and abuse history, family history, assault and delinquency history, school background information, mental health and substance abuse history, and presenting social and psychological traits. Of the 1,026 individuals, 57% (759) did not report a history of early abuse, 25% (256) reported a history of early abuse, and 1% (11) had no data. Fifty-four percent (554) of the youth sampled reported that they had experienced a childhood trauma. Forty five percent (462) did not report a childhood trauma, and one percent (10) had no data entered. Assessment of family history traits indicated that 60% (616) of the youth sampled have a family history of violence, 39% (400) did not report a family history of violence, and 1% (10) had missing data. In comparison, of the 1,026 individuals sampled, 61% (400) did not report a history of low warmth in their families, 59% (605) did report a history of low warmth in their families, and 2% (21) had data missing.

RELIABILITY
Reliability of the CARE-2 was computed using split half (.73), equal length Spearman Brown (.84), and unequal length Spearman Brown (.84). This level of reliability is considered acceptable for the development of clinical scales such as the CARE-2 (Nunnally, 1994).

VALIDITY
There are four norms for the CARE-2: pre-adolescent male, adolescent male, pre-adolescent female, adolescent female. Scoring is individualized for each group and will be reported separately.
PRE-ADOLESCENT MALES
The total CARE-2 score was significantly associated with chronic and severe behavior problems among pre-adolescent males. Table 3.6 provides the means and standard deviations for CARE-2 scores across the six severity groupings. The average CARE-2 score for individuals in the group with no history of behavior problems, including assaults, was 7. The average CARE-2 score then rises with the severity of the problem behaviors. The average scores for pre-teen males with chronic or severe assaultive behaviors and severe behavior problems was 56. The difference across the means was statistically significant (R = .77, F = 72.32, P = .00, ROC = .94). The CARE-2 score was significantly correlated with a history of severe behavior problems and assaultive behaviors. The CARE-2 score was significantly correlated with a history of severe behavior problems and assaultive behaviors among pre-teen females. The CARE-2 score was significantly correlated with a history of severe behavior problems and assaultive behaviors among teen males.

PREDICTIVE VALIDITY
To determine the degree to which scores on the CARE-2 were predictive of future violent behavior, a small sample (112) of youth was tracked after the CARE-2 was administered. The administrator of the study kept a confidential master list of youth identification codes and names. Therapists were asked to submit a sample of case files of youth who had and those who had not committed assaults after the administration of the CARE-2. An examination of the case files was conducted at 6 months after the initial CARE-2 Assessment. Assaults committed after the administration of the CARE-2 (or lack thereof) were recorded in the database. The total CARE-2 score for each individual was then correlated with the number of assaultive behaviors determined from the review of the case files. The total CARE-2 score was, as expected, positively and statistically significantly related to the likelihood of future assaultive behavior and other behavior problems. Youth with the more severe and chronic behavior problems and who had been assaultive had the highest CARE-2 scores, while youth with no or mild behavior problems had the lowest CARE-2 scores.

Discriminant Function Analysis
The items with numerical scores were subjected to discriminant function analysis. For males ages 6 to 12, there were four functions, listed below:

I. Aggression
  • Assaults: not chronic, chronic, severe, using a weapon, and against an authority figure
  • Believing in the legitimacy of aggression as a means to an end
  • Delinquency
  • Lacking remorse for one's misdeeds or victims
II. Behavior Problems
  • A history of behavior problems
  • Behavior problems before the age of 12
  • Poor anger management
  • Impulsivity
  • Poor problem solving skills
  • Poor social skills
  • Few pro-social peers
  • Chronic school behavior problems
III. Trauma Issues
  • Childhood trauma
  • Lacks commitment to school
  • Harms animals
  • Exhibits bullying behavior
IV. Attachment Issues
  • Attachment problems
  • Fire setting
  • Enuresis or encopresis
  • Favorable attitudes toward anti-social behavior

Function I accounts for 97% of the variance, while all four functions combined account for 98% of the variance.