Primary Predictor Variable: Children’s Exposure to Chronic Community Violence
Because community violence is a significant variable in this project, it is important to assess it using multiple methods, informants, and measures. Community violence is related to other forms of violence, several of which are assessed as well (with the exception of parent–child violence, as there were substantial concerns regarding parents’ willingness to consent to participate if we were to assess child abuse). Children’s exposure to community violence is being assessed using both child and parent report of perceived events. Perceived exposure to violence to community violence may differ from objective accounts of events (e.g., police reports, indicators of neighborhood and social disorder). The Children’s Report of Exposure to Violence (CREV; Cooley et al.
1995) is a widely used self-report questionnaire developed to assess children’s lifetime exposure to community violence. Community violence is defined as deliberate acts intended to cause physical harm against persons in the community. The types of violent situations include being chased or threatened, beaten up, robbed or mugged, shot, stabbed, or killed. The original CREV has good two-week test–retest reliability (
r = .75), internal consistency (overall α = .78), and construct validity (Cooley et al.
1995).
An additional module was created in a previous project to assess youth’s exposure to war and terrorism. Following the terrorist attacks in the United States in September 2001 and the initiation of the Iraqi war, this “world violence” module was designed to assess the frequency of children’s perceived exposure to war and terrorism that may have occurred in their communities, their country, or elsewhere in the world (e.g., attacks on public transportation, chemical or biological attacks, bombs, war). As in the other CREV modules, frequency of exposure to world violence is through four modes (i.e., media, hearsay, direct witness, direct victimization). The CREV-Revised (CREV-R) is comprised of the original 29 items plus world violence items. Its Total score is derived by summing the responses (scored 0–4) on the 45 scored items for the Media, Reported/Hearsay, Witnessed, Victim, and World Violence subscales; higher scores indicate greater exposure. The potential range of scores is from 0 to 180. The lifetime version of the CREV-R was used at Wave 1, but the past-year version is being used in Waves 2 and 3 to determine chronicity/severity of violence exposure. The CREV-R has good reliability and validity as demonstrated in a preliminary study of school-based sample of third to fifth grade urban children using a paper–pencil version. Cronbach’s α’s for the computerized version of the lifetime CREV-R Total score with the World Violence module is 0.78, without it is 0.88. The past-year CREV-R Total score α is 0.89.
The Children’s Report of Exposure to Violence—Parent Report (CREV-P) is a modification of the CREV-R to obtain parent/caregiver’s report of their perception of their child’s exposure to community violence. Scoring procedures for the CREV-P are generally identical to the CREV, although the Media violence subscale is not asked of the parents. Separate past year and lifetime scores may be computed. Both have good internal consistency in a preliminary study (Cronbach’s αs = 0.93 and 0.91, respectively) and in the current study (Cronbach’s αs for the computerized version of the lifetime and past-year Total scores are 0.81 and 0.79, respectively).
Eight items of the short version of the
My Exposure to Violence (MyETV; Selner-O’Hagan et al.
1998) structured interview are included in the MORE Project child battery. The past-year portion assesses exposure to violent events that were either witnessed or personally experienced (e.g., shot at, heard gunfire, serious accident, seen dead body). The distinction between violence in the home versus in the community is important to make in community violence research with children (Horn and Trickett
1998), and is permitted by the addition of the MyETV. On a sample of 9- to 24-year-old participants from diverse racial/ethnic groups, the MyETV was found to have high internal consistency, test–retest reliability, and good construct validity (Selner-O’Hagan et al.
1998). Reliability estimates based on the sum of the eight individual items in the current study were low (Cronbach’s α = 0.45).
Risk Factors
Demographic characteristics, including socioeconomic status, are reported by parents/caregivers using the Household Structure and Demographics questionnaire. It was created by researchers in the Baltimore Prevention Program at Johns Hopkins University for use in large school-based community-epidemiological studies is asked of parents/caregivers to provide family socio-demographic characteristics for each of the members of the household. It includes level of education, occupational status, ethnicity, employment status, age, and relationship to the target child. Additional information assessed includes self-reported total family income, the child’s country of origin, the biological father’s and mother’s involvement in the child’s caregiving, and the number of moves the family has made since the target child was born.
Culture of violence is a broad risk factor that includes indices of the community’s pervading attitude toward violence, aggression, and hostility, both as a means to an end and as a symptom, as well as whether it is accepted or not. Included among the measures to assess this construct are the child’s report of their Attitudes Toward Violence and Relational Aggression, parent STAXI and CTS1, as well as indices of violence from the neighborhood assessment, NIfETy. Youth’s self-reported attitudes toward violence were assessed via five items derived from the
Attitude toward Interpersonal Peer Violence Scale (Slaby and Guerra
1988). The scale indicates the perceived legitimization or appropriateness of aggressive responses to threat. Responses across the five items (e.g., “Its okay for me to hit someone if they hit me first”) are averaged with higher scores indicating greater support for aggressive behavior. Prior research with this measure reported α’s ranging from 0.75 (Dahlberg et al.
1998) to 0.85 (Bradshaw et al. in press). Parental trait anger is one of six scales from the
State-Trait Anger Expression Inventory-2 (STAXI-2, Spielberger
1999) used in the MORE Project to assess angry feelings as a personality trait. This scale is comprised of ten items assessed on a four-point Likert scale to indicate the frequency/intensity of anger over time. Evidence supports the validity and reliability of the STAXI-2 for adolescents and adults (Spielberger
1999). In the MORE Project, the Cronbach’s α for the STAXI-2 Trait Anger scale is 0.81.
Relational aggression is assessed through four items developed by Little et al. (
2003) to measure reactive relational aggression. Participants indicate the extent to which they agreed with statements such as, “If others have hurt me, I try to keep them from being in my group of friends,” and “When I am upset with others, I ignore them or stop talking to them.” Prior research with this measure reported an α of 0.63 and in the MORE Project is very low (α = 0.35). Parental trait anger is one of six scales from the
State-Trait Anger Expression Inventory-2 (STAXI-2, Spielberger
1999) used in the MORE Project to assess angry feelings as a personality trait. This scale is comprised of 10 items assessed on a four-point Likert scale to indicate the frequency/intensity of anger over time. Evidence supports the validity and reliability of the STAXI-2 for adolescents and adults (Spielberger
1999). In the MORE Project, the Cronbach’s α for the STAXI-2 Trait Anger scale is 0.81.
Familial conflict and violence is assessed using the
Conflict Tactics Scale—
Form R (CTS1-Form R; Straus
1979,
1987,
1988,
1990), which is a 13-item parent report of intrafamilial violence used to resolve conflicts. Items are rated on a 6-point Likert scale; higher scores indicate more family conflict and higher levels of coerciveness. There are three subscales (Reasoning, Verbal aggression, and Violence), each of which the parent respondent rates: (a) their own behavior toward their partner (i.e., “participant”); and (b) their partner’s behavior toward the participant (i.e., “partner”). No questions regarding parent/caregiver aggression toward their child were asked. The CTS-Form R has high internal consistency, face and concurrent validity, and acceptable construct validity (Straus
1979). In the current sample, the internal consistencies for the Reasoning, Verbal Aggression, and Violence subscales were α’s = 0.69, 0.76, and 0.76, respectively, and for the Participant and Partner scales were 0.72 and 0.69, respectively.
Characteristics of the neighborhood environment are assessed by the
Neighborhood Inventory for Environmental Typology (NIfETy; Furr-Holden et al.
2008). The NIfETy method uses independent evaluators who go to the residential blocks of Baltimore neighborhoods to systematically assess physical and social disorder; indicators of violence, alcohol, and other drug exposure; and positive neighborhood characteristics. Built upon previous methods that assessed neighborhood context to inform child and family health (e.g., Caughy et al.
2001; McDonnell
2007; Raudenbush et al.
2003; Sampson and Raudenbush
1999,
2005), the NIfETy method involves an epidemiological approach to evaluate characteristics of residential neighborhoods that might indicate a change in crime, violence, victimization, and alcohol and other drug exposure in a manner that is quantifiable, replicable, and designed to be longitudinal (Furr-Holden et al.
2008). For the MORE Project, the city unit blocks in which the consented families resided were given to the NIfETy project investigators who sent trained field assessors to make evaluations using Palm OS Zire 31 Personal Digital Assistants (PDAs) programmed with Pendragon Form 5.0 software. For Cohort 1, 98.1% of the families’ neighborhood blocks were assessed. The assessments were conducted in the daytime.
There are 114 quantitative and 15 qualitative items that comprise seven domains assessed by the NIfETy that include positive/healthy and negative indicators: (1) Physical layout of the block face (e.g., length/width of block, alleys present [that run through to next street], dwelling count); (2) Types of structures (e.g., single family/detached homes, liquor stores, churches); (3) Adult activity (e.g., adults watching youth, adults in work uniforms, [male] adults sitting on steps); (4) Youth activity (e.g., youth playing, “corner kids/boys,” dangerous youth activities; 5) Physical disorder and order (e.g., abandoned/vacant structures, new construction or renovation, police present); (6) Social disorder and order (e.g., outdoor community recreation outlets, homeless people, traffic); and (7) Violence and alcohol and other drug indicators (e.g., drug paraphernalia, memorials, obvious signs of drug selling). In an independent sample, internal consistency reliability for the Total NIfTEy scale was good (intra-class coefficient = 0.84); α coefficients ranged from 0.27 to 0.90 for the subscales; and inter-rater reliability and validity were in the acceptable to good range (Furr-Holden et al.
2008).
Adverse Life Events in the child’s life are assessed by both child and parent report on the MESA, respectively. The former is assessed using the
Multicultural Events Schedule for Adolescents (MESA; Gonzales et al.
1995). This scale was developed to assess major and minor life events that are specific to an inner city, multi-ethnic population (Gonzales et al.
1995). It was normed on African American and Caucasian youth, as well as English- and Spanish-speaking Mexican American adolescents. The MESA was derived from existing life events scales (e.g., Adolescent Perceived Events Scale, Compas et al.
1987; Adolescent Life Events Checklist, Johnson and McCutcheon
1982) and is comprised of 84 items that occur over the past 3 months. A Total life events score is based on the total number of events endorsed, with a higher score indicating more adverse life events and hassles. There are eight separate subscales: Family Trouble/Change; Family Conflict; Peer Hassles/Conflict; School Hassles; Economic Stress; Perceived Discrimination; Language Conflicts; and Perceived Violence/Personal Victimization. The MESA has concurrent validity and adequate test–retest reliability (two weeks:
r = 0.71; Gonzales et al.
1995). In the current study, the Cronbach’s α for the MESA total score is 0.90.
Parental reports of adverse events that occurred in their child’s life are assessed using a modified version of the
Multicultural Events Schedule for Adolescents (MESA; Gonzales et al.
1995) created for the MORE Project using five of the eight subscales. This Parent-MESA is comprised of 34 items that occurred over the past year and include the following subscales: Family Trouble/Change, Family Conflict, Economic Stress, and Violence/Personal Victimization. In the current study, the Parent-MESA Total score Cronbach’s α is 0.86.
Protective Factors
Children’s Prosocial Coping is assessed by the child’s self-report on the ACOPE and SSRS, as well as teacher reports on the SSRS and TOCA. The
Adolescent Coping for Problem Experiences (A-COPE; Patterson and McCubbin
1987) is a youth self-report measure that identifies major coping strategies and behaviors in dealing with general life stress (Schwarzer and Schwarzer
1996). The 54 items rated on a 5-point frequency scale comprise 12 subscales, although only seven of them are used in the MORE Project (i.e., Ventilating Feelings, Seeking Diversions, Solving Family Problems, Avoiding Problems, Seeking Spiritual Support, Investing in Close Friends, and Seeking Professional Support). The A-COPE has been validated within a longitudinal study investigating health-risk behaviors and is appropriate for research on youth stress and health-risk behaviors (Schwarzer and Schwarzer
1996). In the current study, the internal consistency reliability for the A-COPE Total score is α = 0.72.
The
Social Skills Rating System (SSRS; Gresham and Elliott
1990) assesses third through twelfth grade students’ social behaviors that may affect student-teacher relationships, peer acceptance, and academic performance (Gresham and Elliott
1990). There are five subscales, two of which are used in the MORE Project: Cooperation and Self-Control. The Cooperation scale includes behaviors such as helping others, sharing materials, and complying with rules and directions (Gresham and Elliott
1990). The Self-Control scale includes behaviors exhibited in conflict (e.g., teasing) and nonconflict (e.g., compromising) situations. Items are rated on a three-point Likert frequency scale. National norms are based on a very large, diverse (e.g., multi-racial, male, female) sample of youth, yielding a median Social Skills Total score coefficient α of 0.90 across all informants (i.e., parent, child, and teacher). Published internal consistency α coefficients for the Cooperation and Self-Control subscales ranged from 0.78 to 0.84, with acceptable test–retest reliabilities for the Cooperation and Self-Control subscales (Gresham and Elliott
1990). In the MORE Project, internal consistencies for the youth’s Social Skills Rating Scale Cooperation and Self-Control subscales are α = 0.74 and 0.58, respectively, and the combined SSRS Cooperation and Self-Control α is 0.78.
The Teacher Form of the SSRS (Gresham and Elliott
1990) individually assesses student’s social skills and academic competence using this screening instrument that classifies social behavior in educational and family environments. The Teacher Form of the SSRS is comprised of subscales that assess social skills, problem behaviors and academic competence. The 57 items are rated using 3-point frequency and importance scales. The raw scores from the Social Skills Scale and the Problem Behaviors Scale are converted into age- and gender-normed standard scores (
M = 100;
SD = 15; Benes
1995) based on a large standardization sample that included regular and special education students, as well as a significant proportion of ethnic minority youth (Benes
1995). The teacher report is psychometrically sound and has good internal consistency, test–retest reliability, and validity (Gresham and Elliott
1990). The Cooperation, Assertion, Self-Control, Internalizing, Externalizing, and Hyperactivity subscales are used in the current study; the Cronbach’s αs for those scales range from 0.80 to 0.93 and the Cronbach’s α for the Academic Competence scale as 0.96.
The
Teacher Observation of Classroom Adaptation—
Revised (TOCA-R; Werthamer-Larsson et al.
1991) is a brief measure of each child’s adequacy of performance on the core tasks in the classroom as defined and assessed by the teacher. The teacher reports on the adequacy of each child’s performance on a six-point scale on six basic tasks: Accepting authority (aggressive behavior); social participation (shy or withdrawn behavior); self-regulation (impulsivity); motor control (hyperactivity); concentration (inattention); and peer likeability (rejection). In addition, the teacher reports on youth’s academic performance, behavior, education, substance use, and mental health services s/he perceives each child needs or is receiving. The TOCA-R subscales included in the MORE Project include: Concentration, Aggression, Shy Behavior, Likeability, Hyperactivity, Impulsivity, Proactive Aggression, Oppositional Defiant, and Conduct Problems. Excluding the Shy Behavior scale, whose α was 0.51, the subscale Cronbach’s αs in the MORE Project range from 0.78 to 0.91.
Cognition and Cognitive Ability are hypothesized as protective factors, including the WASI and Social Information Processing
. General cognitive ability is assessed in the MORE Project using two of four subtests of the
Wechsler Abbreviated Scale of Intelligence (WASI; Psychological Corporation
1999). The WASI was designed to provide a quick and accurate estimate of an individual’s intellectual functioning for screening purposes (Psychological Corporation
1999). The Vocabulary subtest assesses expressive vocabulary, expressive knowledge, verbal knowledge, and fund of information; Matrix Reasoning involves gridded patterns to assess nonverbal reasoning ability (Psychological Corporation
1999). The published WASI test–retest reliability coefficients for the children’s sample ranged from 0.86 to 0.93 for the Vocabulary subtest, 0.86 to 0.96 for Matrix Reasoning, and for the two-subtests combined were from 0.85 to 0.88 (Psychological Corporation
1999).
Social information processing is assessed via a modified version of Dodge and Frame’s (
1982) series of vignettes describing a hypothetical situation between the participant and a peer (Bradshaw et al. in press). The interviewer reads aloud each scenario and participants state their interpretation of the provocateurs’ intent (hostile attribution bias) and their own likely response (aggressive response generation). Using a seven-point coding scheme, independent coders rate intent and response such that higher scores indicate more aggressive hostile attribution biases and aggressive response generation, respectively. Prior research with this measure yielded αs of 0.62 (intent) and 0.63 (response). The correlations among the intent ratings assigned by the three coders for the MORE project ranged from
r = .94,
p < .001 to
r = .96,
p < .001, with an ICC of .95, indicating a high level of agreement. Similarly, the correlations among the response ratings ranged from
r = .91,
p < .001 to
r = .93,
p < .001, with an ICC of .91. Ratings were averaged to yield one score per item. The intent and response scores across all four vignettes were averaged, yielding one score for hostile attribution bias (α = .78) and one for response generation (α = .78), respectively. Having lower scores on the intent and response items is hypothesized as more protective.
Parent Psychiatric Health is assessed using the
Symptom Checklist-90-Revised (SCL-90-R; Derogatis
1977), a widely used self-report measure designed to assess a broad range of adult psychiatric symptom patterns. There are nine symptom scales and three global scales (i.e., Global Severity Index, Positive Symptom Distress Index; Positive Symptom Total). The nine symptom scales are: Somatization, Obsessive–Compulsive, Interpersonal Sensitivity, Depression, Anxiety, Hostility, Phobic Anxiety, Paranoid Ideation, and Psychoticism. The SCL-90-R has excellent reported reliability and validity (Derogatis and Savitz
2000). In the MORE Project sample, the SCL-90-R total score Cronbach’s α is 0.98. Lower scores on the SCL-90-R are hypothesized as protective.
Healthy Family Environment is being assessed by parent reports on the FES and the PPS. The
Family EnvironmentScale (FES; Moos and Moos
1986), which assesses social and environmental characteristics of the family. There are 10 nine-item subscales including: Cohesion, Expressiveness, Conflict, Independence, Achievement Orientation, Intellectual–cultural orientation, Active-recreational orientation, Moral-religious emphasis, Organization, and Control. The FES is used to measure the parent/caregiver’s perceptions of the family environment, deemed an important potential protective factor. In the current study, the Cronbach’s α for the Total FES score is 0.77. The
Parenting Practices Scale (PPS) assesses parental involvement, monitoring, and discipline using a five-point Likert scale. Parents/caregivers are asked how frequently over the past month they engaged in positive, developmentally appropriate interactions and communications with their child (i.e., 10-item Parental Involvement subscale), and applied discipline/punishment and its effectiveness (i.e., eight-item Discipline subscale). In the MORE Project, the Cronbach’s αs are 0.79 and 0.63, respectively.
Healthy School Environment is assessed via school administrator, teacher, and child assessments. A background questionnaire is completed by each principal, vice-principal, and teacher with MORE Project students to provide information on their educational career and teaching certification (e.g., highest degree, year graduated, type of certification).
School Climate is measured by a questionnaire that uses a four-point Likert scale to assess the atmosphere of the overall school environment including safety and availability of resources and support from the principal, vice-principal, and teacher’s perspective. In the current sample, the Cronbach’s α’s for the School Climate items for the principals and teachers are 0.92 and 0.94, respectively. Teachers also completed the
Classroom Climate scale that indicates the general skill level and behavior of all students in their class (in aggregate) on a five-point Likert scale; Cronbach’s α = 0.78. Children’s attitudes toward school are assessed via four items administered to the students from the
Sense of School Membership Scale (Goodenow
1993). Youth indicate on a four-point Likert scale the extent to which they agree with statements such as, “I feel like I belong at this school” and “The teachers here respect me.” Prior research on this measure reported α’s ranging from 0.77 to 0.88 (Dahlberg et al.
1998; Goodenow
1993); the current α reliability coefficient is only 0.57.
Healthy Neighborhood Environment is the last level of hypothesized protective factor, which includes parent reports, child self-reports, and assessments of positive neighborhood indicators. Positive aspects of the neighborhood environment are assessed using the
Collective Efficacy Scale (Sampson et al.
1997) as indicated by the parent/guardian. Collective efficacy is defined as “social cohesion among neighbors combined with their willingness to intervene on behalf of the common good” (Sampson et al.
1997). An extremely large survey of Chicago neighborhood residents showed that collective efficacy has a strong negative association with violence and high between-neighborhood reliability (Sampson et al.). The five-item Informal Social Control subscale α reliability coefficient in the MORE Project is 0.79. Students’ perceptions of safety are assessed through the following three items: “I feel safe at my school,” “I feel safe in my house,” and “I feel safe in my neighborhood.” Students indicate the extent to which they agree with each statement on a four-point Likert scale. Prior research on this measure reported an α of 0.63 (Dahlberg et al.
1998) and in this project the reliability coefficient is 0.54. Indicators of healthy neighborhoods as assessed by the NIfETy Method (Furr-Holden et al.
2008) are also hypothesized to protect children from adversities associated with exposure to community violence.