This is an overview of the types of constructs which one might look at to determine if a student is in a high risk category for acting out in a violent manner, and the types of tests which would measure those constructs. We will look at some of these predictors, the constructs they attempt to measure, and how this might aid in predicting future behavior.
There have been a lot of studies, interventions, programs, and models designed to reduce or predict violence among our youth. The strongest predictor being past violent behavior. Most of these studies have been linked to some type of deficiencies in the home environment and school environment. The overwhelming question facing America now is – Why would a student who has almost anything he desires, living in an upper middle class neighborhood, bring a gun to school with the purpose of killing his classmates and teachers? The question for researchers is – Can we predict which students are likely to engage in this type of behavior? The resounding answer so far seems to be negative.
There is not any test, inventory, or self-report scale which can tell us which students will act out in this manner. However, reviewing the literature there appears to be different types of measurement when looked at aggregately, might identify those students who would be at higher risk although they do not show a past history of violence and therefore fall outside of the previously researched areas.
Some of the things we would hope to assess in identifying violence-related attitudes, beliefs, and behaviors among youths would be broken into three categories:
1. Attitude and Belief Assessments
— aggression, couple violence, education and school, employment, gangs, gender roles, television, handguns
2. Psychological and Cognitive Assessments
— aggressive fantasies, role models, attributional bias, depression, psychological distress, fatalism, future aspirations, hopelessness, hostility, moral reasoning, perceptions of self, responsibility, self-efficacy, impulse control, self-esteem, empathy, and social consciousness
3. Behavior Assessments
— concentration, aggressive behavior, conflict resolution skills, drug and alcohol use, handgun access, leisure activity, parental control, social competence, social problem solving skills, victimization, disciplinary and delinquent behavior.
4. Environmental Assessments
— exposure to violence, family environment (adaptability, bonding, cohesion, relationships), quality of life, quality of neighborhood
Assessment of Self-Esteem
One of the psychological and cognitive assessments we choose to look at is self esteem. Self-esteem has been viewed in different ways. Block and Robins (1993) have viewed it as a global entity: “we view self-esteem as the extent to which one perceives oneself as relatively close to being the person one wants to be and/or as relatively distant from being the kind of person one does not want to be, with respect to person-qualities one positively and negatively values”. Self concept theory has stressed that self-esteem is an attitude about oneself as a whole (global self-esteem) as well as one’s functioning in specific areas of concern to oneself (specific self-esteem).
Relatively little is know concerning relationships between a child’s self-esteem and observations of the child’s behavior. Most have come to a clinical assumption that children with externalizing behavior suffer from poor self-esteem.
The other issue about self-esteem revolves around whether or not it is a stable trait or a fluctuating state. Heatherton and Polivy (1991) referred to the short-lived changes in an individual’s self-esteen as “state” self-esteem and developed a scale to measure it called the State Self-Esteem Scale (SSES) which is a 20-item Likert-type scale designed for measuring temporary changes in individual self-esteem. There are three self-esteem factors in the scale: Academic Performance, Social Evaluation, and Appearance. Coefficient Alpha for the scale equals 0.92.
Linton (1996) conducted a study to test its validity by comparing it with the Rosenberg Self-Esteem Scale. It consists of ten items answered on a four point scale from “strongly agree” to “strongly disagree”. The scale has a Guttman scale reproducibility coefficient of 0.92 and a test-retest correlation of 0.85. Her results showed a significant correlation between self-esteem measures on the Rosenberg Self-Esteem Scale and four components on the SSES. She demonstrated that SSES measures four distinct components within the state self-esteem construct and provides evidence that there is a fluctuating nature of self-esteem. It also supports the use of the SSES for study within the adolescent populations.
Another study by Frankel (1996) compared Piers-Harris Self-Concept Scale(PHS) and the Child Behavior Checklist Inventory (CBCL) with the Pupil Evaluation Inventory (PEI) to get a better understanding of why children with internalizing problems (withdrawal, somatic complaints and sadness) consistently demonstrate low self-esteem, while results of children with externalizing behaviors (aggression, poor impulse control, and non-compliance) have been inconsistent. Externalizing behaviors have been demonstrated to be stable over time, in the absence of treatment (McMahon, 1994). Schneider and Leitenberg (1989) found that externalizers reported higher self-esteem than internalizers which seems to be inconsistent with Olweus (1978) findings that children who bully others have lower self-esteem than well adjusted children.
The PHS is an 80-item yes-no self report measure which takes about 20 minutes to complete. The PHS manual provides factor scores on six scales measuring specific self-esteem and a global score which is a weighted composite of items from the specific self-esteem factors (composite scale). It also contains a behavior scale, an intellect scale, a physical appearance scale, a popularity scale, and an anxiety scale. The manual states an internal consistency of the scales ranging from 0.88 to 0.93 and test-retest correlations were 0.81 across a 5-month interval.
The CBCL consists of 118 behavioral items. Frankel’s study (1996) used the revised Externalizing broad-band scale and the two narrow-band scales (Social Problems and Social Competence–Social) found to tap social competence. Frankel found that the mean Social Problems scale score was above the cutoff ofr clinical significance (98 percentile) while mean Externalizing and Social Competence–Social scale scores were in the problematic direction but within normal limits.
The PEI consists of 35 items, each rated as “describes child” or “does not describe child”. Development of withdrawal, likability, and aggression scales were based on peer ratings. Correlations between peer and teacher ratings have exceeded 0.54.
The results from Frankel’s study (1996) demonstrated that self-esteem in boys with peer problems was associated with a combination of social competence and externalizing problems. They showed that all the PHS scales except popularity and appearance were related to social competence. Both scales were related to externalizing problems. Therefore the boy without friends who is perceived as aggressive by his mother tends to report higher self-esteem in relation to peer acceptance that the non-aggressive boy without friends.
Assessment of Depression
One of the most common and widely used assessments for depression is the Beck Depression Inventory (BDI). The BDI consists of 21 items which cover a range of affective, behavioral, cognitive, and somatic symptoms that are thought to be indicative of unipolar depression. The subject can select from among four alternative responses for each item to reflect increasing levels of severity of depressive symptomatology. Scores can range from 0 to 63. The higher the score the more reported depression.
Carter (1996) conducted research on hospitalized adolescents, to compare the validity of the BDI, the Minnesota Multiphasic Personality Inventory (MMPI), and the Rorschach in assessing adolescent depression. Although these assessment scales have been researched throughly over the years with adults, the validity of these scales with adolescents has been conflicting.
The primary means of assessing depressive symptomatology on the MMPI is the depression (D) scale. It consists of 60 items with the subject either agrees or disagrees, allowing for a range of scores from 0 to 60. The items are associated with clinical symptoms that characterize feelings of hopelessness, despair, discouragement, and basic personality features like high personal standards and intrapunitiveness. MMPI-D was able to correctly identify 69% of a sample of depressed individuals using a T-score of 70 or above for its criterion.
The Rorschach Depression Index (DEPI), is comprised of five variables (vista responses, color-shading blends, egocentricity index, achromatic color responses, and morbid responses) and the subject can receive scores ranging from 0 to 5.
The results of Carter’s research (1996) showed a statistically significant relationship between the BDI and the MMPI-D scale. However, there was not a significant correlation between the DEPI with the BDI or the MMPID. This concurrent validity was assessed by computing Pearson correlation coefficients for the depressed and non-depressed groups. Both the BDI and the MMPI-D were statistically significant in discriminating depressed and non-depressed samples. The DEPI as a sole predictor variable did not yield a significant discriminant function.
When looking at the three assessments scales as predictor variables in varying combinations, no combination increased the classification accuracy rates produced by the MMPI-D scale alone.