VON E. NEBBITT, TAQI M. TIRMAZI, AND TAREK ZIDANFROM SPRING 2005 THROUGH SUMMER 2008, cross-sectional data were collected from 898 African American adolescents living in public housing developments in four large U.S. cities: New York City (Queens), North Philadelphia, St. Louis, and Washington, DC. The data we collected and used for analyses in the following chapters are all self-reported and obtained using standardized instruments.
The purpose of this chapter is to explain the methodology and procedures used to conduct this study. The chapter is presented in four sections. First, we describe the process for community engagement. Second, we explain the research protocol and provide a brief description of the sample. Third, we describe all measures used to collect data, including scoring information and psychometric properties where applicable. Finally, we describe each housing development, including the socioeconomic and demographic characteristics of each research site.
COMMUNITY ENGAGEMENT
Community engagement was achieved with the use of a consistent, malleable, and sensitive plan. The consistency of the plan involved a methodological process that was built on community ties and that reliably culminated with the successful collection of data. The malleability of the plan required a willingness to make slight modifications to accommodate the unique characteristics of each public housing development. The sensitivity of the plan included an awareness that public housing communities are not monolithic; each housing site is a unique context in its own right, and within each context cultural norms may differ.
The community engagement plan consisted of five steps. First, the principal investigator (PI) contacted the local housing authority to garner its support. Second, the PI and research assistants (RAs) identified and contacted a community/recreation center or social service agency (i.e., community-based organizations) in or adjacent to the housing development. Third, the PI hired a staff member (normally a recreation aide) from the community-based organization and a resident from the housing development as a community liaison. Fourth, the RAs and community liaisons posted flyers throughout the housing development and in the surrounding community. Finally, the PI, RAs, and community liaison (i.e., the research team) convened potential youth participants to the community centers, screened them for the inclusion criteria, and administered the survey to youth who met the inclusion criteria.
Community engagement is the essential key to successful community-based research. In each city, the research team identified three key community stakeholders (public housing residents, community centers/social service providers, and local housing authorities) that were necessary to achieve the goals and objectives of community-based research. Public housing residents were, without a doubt, the primary stakeholders. The research team fostered genuine rapport with residents and asked for their involvement in all phases of the research project. Residents were considered to be the primary stakeholders because their involvement was a necessary and sufficient condition to the success of community-based research within public housing developments. It is our position that without their support and “buy in,” research in public housing would be impossible or extremely challenging and would yield unreliable data. Furthermore, we believe that data obtained by circumventing residents’ involvement should be interpreted with extreme caution.
The secondary stakeholders were the community/recreation centers and social service providers in or near the housing development. The service sectors were considered to be the secondary stakeholders because they were necessary but not sufficient for a program of research within public housing developments. Tertiary stakeholders were the local housing authorities. Local housing authorities were considered to be the tertiary stakeholders because their participation was neither a necessary nor sufficient condition to a successful program of research within public housing developments. This statement should not be interpreted as an endorsement to circumvent local housing authorities when conducting research in public housing. However, we believe that a program of research can be successfully initiated and completed without involvement from the local housing authority. It should be noted that letters of support were solicited from the local housing authorities from all cities in this study. Housing authorities in each city supported this research. In the next section, we outline the roles of community stakeholders.
ROLE OF LOCAL HOUSING AUTHORITIES
The PI sent a description of the study to the director of resident services in each housing authority. The write-up included a brief overview of the proposed study and requested a meeting to discuss the study in greater detail. Three of the four housing authorities responded favorably to our invitation. The housing authorities that agreed to meet and discuss the study also granted our request to move forward with the research. The housing authorities provided us with letters of support. In the one city where the local housing authority did not respond to our invitation, the invitation was extended to the director of a community center that served the two housing developments in that city. The director of the community center accepted our invitation to meet and discuss the study. The director then forwarded our request to the department of parks and recreation for approval. Approximately 1 week later, the director of the community center informed the PI that the request had been approved. The department of parks and recreation also provided a letter of support. Despite the endorsement from parks and recreation, the research team continued to notify the local housing authority of our research in their properties. Eventually, the local housing authority provided verbal support. They also provided the PI with descriptive information on residents in their housing developments.
To avoid placing a gratuitous burden on local housing authorities, they played the roles of advising and approving agents. Local housing authorities were not involved in recruitment or survey implementation. This approach was employed to increase the veracity of youths’ responses and to minimize systematic error in youths’ self-reports of their behavior within the housing development. Furthermore, because some housing authorities have zero-tolerance laws, we saw excluding housing authorities from data collection as part of our obligation to protect human participants in research.
Although local housing authorities were excluded from actual data collection, they were important in three specific ways. First, local housing authorities informed the PI of the dynamics within the housing developments, such as current crime trends and challenges faced by residents. Second, local housing authorities helped to identify young resident leaders, key social service agencies, and directors of community centers. Third, local housing authorities provided socioeconomic and demographic information on families and structural information on each housing development.
ROLE OF COMMUNITY CENTERS AND SERVICE PROVIDERS
After receiving the endorsements from local housing authorities, the PI identified and contacted community centers and social service agencies in or adjacent to the targeted housing developments. Priority was given to community centers and agencies in close proximity to the targeted developments and those identified by local housing authorities as frequented by residents. Once contacted, the PI or a senior RA met with directors to garner their support and to discuss their potential role in the study. During the meeting, directors were asked to provide space for data collection and to disseminate information on the study throughout the housing development. Strategies to increase the participation and accuracy of participant response were also discussed. In three cities, community/recreation centers provided space for data collection and disseminated information on the study. In one city, a social service agency provided space for data collection and disseminated information on the study. Prior to data collection, directors of community agencies provided feedback on the research protocol and questionnaire.
Once approval for data collection was secured, the research team identified spaces within the community agencies to facilitate data collection. It should be noted that the research sites could not provide similar accommodations for data collection. The physical layout of participating agencies differed significantly across research sites. Some agencies were able to provide more privacy than others. Some agencies were able to accommodate youth who requested privacy, whereas at other agencies this option was unavailable. Furthermore, in two housing developments (site 3 in St. Louis and site 2 in Washington, DC), community centers were unavailable during data collection. Data collection therefore occurred in an open-air common space located in the housing development. The diverse floor plans of participating agencies caused slight variations in data collection across developments.
In addition to hosting data collection, the participating agencies held information sessions for interested youth and community stakeholders. An RA facilitated the information sessions. Information sessions covered the researchers’ institutional affiliations, the purpose of the study, the risks and benefits of the study, rights as a human participant in research, participants’ right to withdraw from the study at any time and the right to refuse to answer any questions, and contact information for the PI and the institutional review boards at Washington University and Howard University. Furthermore, flyers were distributed at information sessions. Flyers contained the purpose of the study and the date/location for data collection. Flyers also contained contact information on the PI, an RA, and a contact person at the host agencies. Each community-based agency received an honorarium for their participation in this study.
At least one employee of the community agency acted as our contact person and agency liaison. Agency liaisons were familiar with the youth in the targeted housing developments, and the youth were familiar and fairly comfortable with the agency liaison. The primary duties of the agencies’ liaisons were as follows: 1) posting the flyers in the host agency, 2) collecting contact information on youth interested in the study, 3) ensuring the research team had access to the host agency during data collection, 4) maintaining order and keeping youth organized during data collection, and 5) minimizing repeat cases (i.e., preventing youth from completing the survey a second time only to obtain the cost incentive). Agency liaisons received a stipend for their service.
THE ROLE OF RESIDENTS
The PI hired two community liaisons at each housing development. Typically, community liaisons were African Americans in their early to mid-twenties who had lived in the housing development for 5 years or more and were well known throughout the housing development. Community liaisons differed from agency liaisons in that community liaisons were not employed at the community-based organizations. Each community liaison received brief training in the research protocol and various methods of recruitment. All community liaisons received a stipend for their service. In addition to official community liaisons, youth RAs were retained on an ad-hoc basis. Youth RAs assisted the research team with distributing flyers, recruitment cards, and, on rare occasions, letters to parents. Ad-hoc youth RAs were typically 13 to 15 years of age. They also received a small stipend per diem.
Community liaisons were essential to gaining access to this difficult-to-reach population of youth. Community liaisons were also critical to the research team’s legitimacy in the housing development. Prior to data collection, community liaisons posted flyers throughout their housing developments. They also distributed recruitment cards, which contained the purpose of the study and contact information for the host agency liaison.
During data collection, the community liaisons and ad-hoc youth RAs ensured that youth were residents in the housing development and that youth in the housing development knew the location of the data collection host agency. Community liaisons also helped the graduate student RAs with distribution of parental consent and youth assent forms. They were also helpful with distributing pens and blank questionnaires to youth who provided signed parental consent, youth assent, or informed consent. Community liaisons and ad-hoc RAs did not have access to completed surveys. Community liaisons were essential to the success of this project.
When youth gathered at the host agencies for information sessions, community liaisons helped to translate the confidentiality and the rights of human subjects in research clauses into the communities’ vernacular. During data collection, they were vital in conveying the message that the local housing authority had no role in data collection, that any information available to housing authorities would be deidentified, and that all questionnaires would receive an identification number. Without their support and community ties, this study would not be possible.
PROCEDURES
Data were collected from 898 youth ages 11 to 21 living in nine public housing neighborhoods located in four large U.S. cities. Three housing developments were in St. Louis; two housing developments were in Washington, DC; two housing developments were in New York City (Queens); and two housing developments were in North Philadelphia. At six of the nine sites, data collection took place in community centers located in or adjacent to the housing development. At two of the nine sites, data collection occurred in an open-air common space on the grounds of the housing development. At one of the nine sites, data collection took place in a nonprofit organization located between two housing developments.
SAMPLE POOL
Inclusion Criteria
Participation was restricted to youth who lived in family-only developments. Inclusion criteria included the following: 1) current residency in the targeted housing developments and 2) being between the ages of 13 and 21 years. The architectural structure of the housing developments included high-rise, low-rise garden style, and barrack-style developments.
Exclusion Criteria
Youth in elderly and disabled housings, HOPE VI developments, and Section 8 apartments were excluded from this study. Also, youth who could not demonstrate the capacity to give informed assent were excluded from the study. Youth were screened using the Capacity to Consent Screen (Zayas, Cabassa, & Pérez 2005). The Capacity to Consent Screen measures a participant’s capacity to give informed consent by assessing whether they understand the purpose of the research project, how data will be collected, their rights as a study participant, and that they can cease their participation at any time without consequences.
Only youths who responded correctly to eight or more questions participated in the study. The interviewer administrated the screen three times to obtain the necessary eight correct answers (cumulatively after providing informed consent instructions three times). If the youth failed to achieve a score of 8 after the third time, then the youth was escorted from the room and a member of the research team explained to the youth that he did not meet the requirements for participating. The youth was offered a snack and thanked for his or her willingness to participate. Only two youth, both males (ages 13 and 14), did not meet the capacity-to-consent criteria.
Recruitment consisted of flyers and announcements at local community centers. Specifically, members of the research team, agency liaisons, and community liaisons posted flyers in the housing developments, in community centers, and in agencies around the housing developments. In addition, recruitment cards were distributed to youth living in the communities. The flyers and recruitment cards included a brief overview of the study, the date and location for data collection, and contact information for the PI and RA and agency liaisons.
A different recruitment strategy was employed to recruit youth who were identified by community liaisons as “youth on the block” (i.e., youth who rarely visited community/recreation centers because they were heavily involved in delinquency, drug use, or the drug exchange). These youth on the block were also identified during a walk-through observation in each housing development. Once youth on the block were identified, the PI, an RA, and a community liaison approached what appeared to be the senior member (i.e., the O.G.) of the group. During our conversation with the O.G., the PI identified himself as a researcher and explained his institutional affiliation, disclosed his practice experiences with families in public housing, discussed the purpose of the study, and also shared the critical need to include the voices of “youth on the block” in public housing research. The PI also emphasized how the results of the study may have applications for integrating “youth on the block” back into the community. This approach consistently yielded fruitful recruitment across cities and housing developments. It should be noted, however, that the PI, RA, and community liaison were all African American males with social work practice experiences, personal life experiences, or both in urban public housing neighborhoods. If this approach is replicated in subsequent studies, it should be done with caution and include community liaisons and researchers who are familiar with the vernacular and cultural norms of life in urban public housing neighborhoods.
DATA COLLECTION
Prior to data collection, RAs underwent two trainings. First, RAs were trained in instrument implementation. All community liaisons also received this training. Second, RAs were trained in culturally and contextually appropriate behavior given the research settings. At each data collection session, trained African American graduate students (RAs) and a community liaison explained the purpose of the study, risks and benefits of the study, the purposes of informed consent and assent, confidentiality, and that participants were free to drop out of the study at any time without penalty or consequences. After reviewing the purpose of the study, trained community liaisons distributed parental consent/youth assent forms to potential participants who were 17 years of age and younger for their parents’ review and signatures. Potential participants aged 18 and older received informed consent for their review and signature. Once parental consent/youth assent or informed consent was obtained, groups of 8 to 10 youth were assembled to complete the Capacity to Consent Screen (Zayas, Cabassa, & Pérez 2005). Youth who demonstrated the capacity to give consent were gathered in small groups of 15 to 20 and directed to designated tables to complete the questionnaire. Once assembled at designated tables, members of the research team reiterated the purpose of the study and gave each participant the survey and a pen.
Trained graduate students were present to assist youth where needed and to minimize missing data while youth completed surveys. Also, a licensed clinical social worker was present at each data collection session to administer brief counseling and referrals if youth experienced discomfort while completing the questionnaire. Each participant received a cash incentive of $15. On average, it took approximately 40 minutes to complete the survey. All questions and answers were read aloud and participants circled their desired response. Participants who demonstrated acceptable reading level and high comprehension on the Capacity to Consent Screen completed their questionnaires independently. Youth were provided with a comfortable location to complete the surveys. Youth were provided a snack (pizza and soda) after completing the surveys. Debriefing sessions were held while youth ate their snacks.
Data collection was confidential but not anonymous. The PI obtained contact information on all participants who consented to provide contact information. This method was employed to facilitate subsequent studies and to help clean the data if needed. Human subjects guidelines were strictly observed to ensure the safety and protect the identity of study participants. Parental consent and youth assent were obtained using standardized informed consent and youth assent forms as approved by Washington University and Howard University.
The recruitment efforts yielded a sample of 898 youth living in nine public housing developments across four U.S. cities. The average age of the sample was 15.4 years, with a standard deviation of 2.3 years. Fifty-two percent of the sample was female. Tables 4.1 through 4.4 provide a full description of all demographic and study variables, as well as comparisons between gender and across the four cities.
TABLE 4.1 Descriptive Analysis of Demographic Variables and Comparison by Gender
NS, not significant; *p < .05, **p < .001, ***p < .000.
TABLE 4.2 Descriptive Analysis of Study Variables and Comparisons by Gender
NS, not significant; ATOD, alcohol, tobacco, and other drug use; *p < .05, **p < .001, ***p < .000.
TABLE 4.3 Descriptive Analysis of All Study Variables and Comparisons by City
ND, no data; NS, not significant; *p < .05, **p < .001, ***p < .000.
a Classes 1 and 2 are different.
b Classes 2 and 3 are different.
c Classes 1 and 3 are different.
d Classes 1 and 4 are different.
e Classes 2 and 4 are different.
f Classes 3 and 4 are different.
TABLE 4.4 Descriptive Analysis of All Study Variables and Comparisons by City
ND, no data; NS, not significant; ATOD, alcohol, tobacco, and other drug use; *p < .05, **p < .001, ***p < .000.
a Classes 1 and 2 are different.
b Classes 2 and 3 are different.
c Classes 1 and 3 are different.
d Classes 1 and 4 are different.
e Classes 2 and 4 are different.
f Classes 3 and 4 are different.
g Youth with scores over 3 SD above the mean were dropped.
MEASUREMENT INSTRUMENTS
Participants completed a survey composed of several standardized instruments previously used with minority youth. The survey assessed youths’ perceptions of the neighborhood domain (the housing development), the household domain, the peer domain, and a range of externalizing behaviors and internalizing symptoms.
PERCEIVED NEIGHBORHOOD
The Subjective Neighborhood (SN) scale was used to assess the youths’ subjective appraisal of their neighborhoods. The use of this scale builds upon previous studies (Aneshensel & Sucoff 1996). The SN scale is a 22-item Likert scale that list attributes of a neighborhood. Responses on this scale range from strongly disagree (1) to strongly agree (4). Aneshensel and Sucoff (1996) identified two subscales of the SN: ambient hazard and social cohesion. In their sample, ambient hazard had excellent reliability (α = .90) and social cohesion had acceptable reliability (α = .64) (Aneschensel & Sucoff 1996).
To build on this work, the two components (ambient hazard and social cohesion) of the SN scale were tested for reliability. The ambient hazard subscale asked youth to report on the following risk factors in their neighborhood: violent crimes, drive-by shootings, gang fights, drug use and dealing, property crimes, graffiti, police harassment for no reason, whether the houses are clean (reverse scored), whether the neighborhood is safe (reverse scored), and whether the houses are in good shape (reverse scored). This subscale demonstrated acceptable reliability (α = .64) with the current sample. The social cohesion subscale asked youth to report on three aspects of social cohesion among individuals living in their neighborhood: people are friendly here, youth know each other here, and adults know each other here. This subscale also demonstrated acceptable reliability (α = .61).
EXPOSURE TO COMMUNITY VIOLENCE
Community violence was assessed using the Survey of Exposure to Community Violence: Self-Report Version (Richters & Martinez 1990). This 26-item Likert scale measures several types of violence. Examples of items included in the scale are “How many times have you yourself actually been threatened with serious physical harm by someone?” or “How many times have you seen someone else being sexually assaulted, molested, or raped?” Both direct (victimization) and indirect (witnessing) community violence were assessed. Items ask youth how many times each violent event occurred; responses range from never (1) to many times (4). Items are summed so that higher scores represent greater exposure to community violence. The victimization and witnessing subscales demonstrated acceptable internal consistency with the current sample (α = .84 and .85, respectively).
Household Domain
PARENTAL MONITORING
To assess parental monitoring, warmth, and involvement, youth completed the Parental Attitude Measure (PAM; Lamborn et al. 1991). This 17-item scale assesses two aspects of parenting behaviors: monitoring and encouragement. The five-item parental monitoring subscale asks youth: “How much do your parents really know who your friends are?” Items are scored on a four-point Likert scale ranging from don’t know (1) to know exactly (4). The PAM scale was modified into a 10-item scale to measure both paternal and maternal monitoring. The subscale demonstrated acceptable reliability (α = .76) with the current sample. Jordan (2003) reported adequate reliability (α = .69) when using this subscale with a sample of urban African American youth.
The 12-item parental encouragement subscale assesses both maternal and paternal encouragement. This subscale asks youth, “Does your father/mother, stepfather/stepmother, or the man/woman who takes care of you push you to do your best in whatever you do?” Items are scored on a four-point Likert scale ranging from never (1) to always (4). The 12-item scale demonstrated good reliability (α = .88) with the sample of inner-city African American adolescents in the current study. PAM is scored by summing the items, with higher values indicating higher levels of supervision and encouragement.
PARENT–CHILD RELATIONSHIP
Parent–child relationship was measured using a four-item subscale from the Quality of Parental Relationship Scale from the National Youth Survey (Elliot 1987). The subscale assesses youths’ perception of the quality of their relationships with their parents. For example, respondents are asked, “How satisfied are you with your relationship with your parents?” Responses range from very dissatisfied (1) to very satisfied (5). Another example is, “How much warmth and affection do you receive from your parents?” Responses range from very little (1) to a great deal (5). The measure is summed so that higher scores represent a higher-quality parent–child relationship. The subscale demonstrated acceptable internal consistency with the current sample (α = .76).
FAMILY/HOUSEHOLD CONFLICT
Family/household conflict was assessed using the Family Conflict Scale (Barbarin, Richter, deWet 2001), which is a subscale of the Family Relations Scale. The Family Conflict Scale measures exposure to violence and conflict within the household. Responses are rated on a four-point Likert scale ranging from never (1) to many times (4). Items are summed so that higher scores represent greater exposure to domestic violence and conflict. Acceptable psychometric properties have been reported, with a Cronbach’s alpha of .85 and test-retest reliability coefficient of .90 (Barbarin, Richter, & deWet 2001). Item analyses indicated that questions measuring exposure to violence were highly correlated with questions for illicit drug use and victims of crime and violence. In this study, this measure was used to assess the degree to which the youth experience violence and conflict in their current households.
CAREGIVER’S STATUS
Caregiver’s status was assessed using two items. The first item states, “In my household, the person I consider to be my mother is…” The answer choices were as follows: my biological mother (e.g., your natural mother who had you; score of 1), my stepmother (2), a relative (e.g., uncle, grandmother; 3), an adult who is not related to me (e.g., foster parent, or none of the people I live with; 4), and I live alone (5). The second item states, “In my household, the person I consider to be my father is…” The answer choices were as follows: my biological father (e.g., your natural father; score of 1), my stepfather (2), a relative (e.g., uncle, grandfather; 3), an adult who is not related to me (e.g., foster parent; 4), and I live alone (5).
Peer Domain
EXPOSURE TO DELINQUENT PEERS
The Exposure to Delinquent Peers scale from the National Youth Survey (Elliot 1987) was used to measure respondents’ friends’ involvement in delinquent behavior. This scale asked youth the number of their close friends who have engaged in various types of delinquent behaviors over the past year. The delinquent behaviors assessed by this measure ranged from “alcohol use” to “pressured someone to have sex with them.” The responses categories for the items ranged from all of them (5) to none of them (1). The reliability for this 14-item scale was excellent (α = .93). Items are scored so that a higher score indicates greater exposure to delinquent peers (Elliot 1987).
PEER INFLUENCE
Peer influence was assessed using the Peer Influence subscale from the National Youth Survey (Elliot 1987). This scale measures the degree to which respondents’ peers have influenced their thinking and behavior. This six-item subscale asks youth, “How much have your friends influenced what you think?”, “How much would you like to be like your friends?”, and “How much shared interest and activities do you have with your friends?” Items are rated on a five-point Likert scale ranging from not very much (1) to very much (5). Higher scores indicate more influential peers. The scale demonstrated accepted internal consistency with this sample (α = .76).
Individual Domain
DELINQUENCY
The Self-Reported Delinquency scale is a 20-item subscale from the National Youth Survey (Elliot 1987). Youth respondents were asked to report the frequency with which they engaged in a variety of delinquent behaviors in the last year. In the original survey, each item consisted of two parts: raw frequency and rate (number of times per day). In the present study, only raw frequencies were collected, with the highest category scored as 12 or more times. This approach builds on other studies (Warr & Stafford 1993). To assess the extent to which youth engage in delinquent behavior, respondents were asked the question, “From January 2004 until today, did you ever [commit the act in question]?” Delinquent acts ranged from “steal something worth less than $5” to “attack someone with the idea of seriously hurting or killing them.” Responses were 12 or more times (4), 3 to 11 times (3), 1 or 2 times (2), and never (1). The Self-Reported Delinquency scale demonstrated excellent reliability with the present sample (α = .95). Items are scored so that a higher score reflects greater involvement in delinquent behavior (Elliot 1987).
Drug use and sexual behavior were assessed using the Centers for Disease Control and Prevention (CDC) Youth Risk Behavior Survey (YRBS). The CDC developed the YRBS in collaboration with federal, state, and private-sector partners for use in a national survey for the Youth Risk Behavior Surveillance System. The questions, which were tied to national health objectives for 2010, focus on priority health-risk behaviors established during youth that result in the most significant mortality, morbidity, disability, and social problems during both youth and adulthood (Centers for Disease Control and Prevention 2001). Modifications of the questions on alcohol use, tobacco use, illegal drug use, suicidality, and acquired immunodeficiency syndrome risk behaviors were used in this study. The YRBS was administered in 1990 and has been widely used with youth of color. The CDC provides coding, scoring, and comparative data at the local and national levels for this survey instrument (Centers for Disease Control and Prevention 2011).
SELF-EFFICACY
Self-efficacy was measured using the General Perceived Self-Efficacy Scale (Schwarzer & Jerusalem 1995). This 10-item scale measures a broad and stable sense of personal competence to deal with a variety of life situations. Respondents are asked how true the following statements are: “When I am faced with a problem, I can find several solutions” or “I am confident that I could handle unexpected events.” Responses range from not true at all (1) to true all of the time (4). Items are summed with higher scores representing greater general perceived self-efficacy. The measure demonstrated acceptable internal consistency with this sample (α = .92).
DEPRESSIVE SYMPTOMS
Depressive symptoms were assessed using the Center for Epidemiologic Studies Depression Scale (CES-D). The CES-D has been used extensively for assessing depression and psychiatric epidemiology (Murphy 2002; Naughton & Wiklund 1993; Nezu et al. 2002; Snaith 1993). CES-D is a 20-item scale that assesses mood, somatic complaints, social interactions with others, and motor functioning. Responses are rated on a four-point Likert scale ranging from rarely or none of the time (less than one day; score of 0) to most or all of the time (5 to 7 days; score of 3). The final score spans from 0 to 60, with a higher score indicating greater impairment. Respondents with a final score of 16 or higher are typically identified as having depression.
Although a cutoff point of 16 has been used in adult samples, this definition has yielded an estimated prevalence of adolescent depression of more than 50 percent (Nebbitt & Lombe 2007; Rushton, Forcier, & Schectman 2002). Roberts, Lewinsohn, and Seely (1991) suggested a cutoff of 24 for adolescents’ score on the basis of improved ability to detect depression as defined by the Diagnostic and Statistical Manual of Mental Disorders. Among community samples, internal consistency estimates range from .80 to .90 (Devins et al. 1988; Nebbitt & Lombe 2007; Radloff 1977). The CES-D demonstrated acceptable reliability with the current adolescent sample (α = .88).
ANXIETY
Anxiety was assessed using the Anxiety Sensitivity Index (ASI). The ASI is a 16-item self-report scale that assesses threatening beliefs about arousal (Peterson & Reiss 1987). Items are rated on a Likert scale ranging from 0 (very little) to 4 (very much). The ASI yields a total score by summing the ratings across all items, with higher scores reflecting higher levels of anxiety sensitivity.
Reiss (1991) posited that anxiety sensitivity constitutes a disposition to developing anxiety and does not require the experience of clinically significant anxiety in its own development. Telch, Lucas, and Nelson (1989) hypothesized that ASI is a four-factor scale, for which it would be more appropriate to regard the four factors as components of a single construct; the most parsimonious view of ASI is as a single-factor index (Taylor et al. 1992).
To test the validity of the ASI on a community sample, Schmidt and Joiner (2002) conducted an items analysis. Using corrected item-total correlations, they found that the 16-item scale had a coefficient of .86. However, they found that items 1, 5, and 7 produced questionable item-total correlations (e.g., correlations less than .20). After deleting these items, the internal consistency of the scale improved (α = .88). In the current sample, a Cronbach alpha for the 16 items yielded acceptable internal consistency (α = .90).
ATTITUDES TOWARD DELINQUENCY
Attitudes toward delinquency were assessed using the National Youth Survey’s attitudes toward delinquency subscale. Questions included “How wrong is it for someone your age to steal something worth less than $5” or “How wrong is it to attack someone with the idea of seriously hurting or killing them?” Responses range from very wrong (4) to not wrong at all (1). The 14-item Attitude Toward Delinquency subscale demonstrated acceptable reliability with the current sample (α = .94). Items were scored so that a higher score represents a greater perceived wrongness (Elliot 1987).
POSTTRAUMATIC STRESS DISORDER
Symptoms often associated with posttraumatic stress disorder were measured using the Impact of Events Scale (IES; Horowitz, Wilner, & Alvarez 1979). The IES assesses levels of subjective posttraumatic psychological distress and provides specific measures of event intrusion and event-related avoidance. The IES is a 15-item measure assessing the frequency with which experiences of intrusions, avoidance, and emotional numbing related to stressful events were experienced in the last week.
Respondents were asked to specify the frequency with which they had intrusion or avoidance-related thoughts on a Likert scale ranging from 0 (not at all) to 5 (often). Intrusion and avoidance were the two subscales, and a total score was calculated by summing across all items. Reliability scores were quite high, ranging from .78 to .92 in this study. Respondents were asked to use a specific violent event (of their choosing) as a reference point in answering the questions.
RESEARCH SETTINGS
Socioeconomic and Demographic Characteristics
During the time of data collection, the median annual household income across the nine housing developments ranged from $6,000 to $20,000, with a mean annual income of slightly less than $6,000. Incomes were adjusted to their 2008 values. The heads of households in the nine housing developments were disproportionately African American (95 percent) and female (97 percent). Latino families represented one-third (33 percent) of the households in public housing in New York City. The number of children per household ranged from 2 to 2.5 children.
CITY 1: ST. LOUIS
City 1 included three housing developments: one mixed high-rise and low-rise development and two barrack-style developments. The 2005 median household income in the three housing developments was $6,864. More than 75 percent of the residents had incomes below the official poverty line. Approximately 90 percent of the households were female-headed. The three developments housed approximately 3,500 residents, with 47 percent of residents being under the age of 18. A total of 96 percent of the residents were African American (St. Louis Housing Authority 2006).
The first development in City 1 included three high-rise buildings totaling 242 units and 12 low-rise barrack-style buildings totaling 92 units. The developments occupied six city square blocks. The second development in City 1 included 53 low-rise barrack-style buildings hosting 657 units. The development occupied nine city square blocks. The third development in City 1 consisted of 16 two-story townhouse buildings totaling 148 units built on five city square blocks.
Recruitment efforts in City 1 yielded a sample of 238 African American adolescents age 13 to 19, with a mean age of 15.6 years and a standard deviation of 2 years. Males (52 percent) had a slightly higher, but not significantly higher, representation than females (48 percent). In all, 37 percent lived in the first housing development, 32 percent lived in the second housing development, and 31 percent lived in the third development. Data collection was completed in City 1 in the fall of 2005.
CITY 2: WASHINGTON, DC
City 2 included two low-rise barrack-style public housing neighborhoods. A total of 223 residents lived in the first neighborhood and 426 residents lived in the second neighborhood. Ninety-eight percent of the families were African American, and 58 percent of the residents were under the age of 18. In 2006, the median household income was approximately $10,200 (Office of Resident Services 2006).
The first public housing neighborhood in City 2 consisted of 58 low-rise barrack-style buildings with 234 units occupying four city square blocks. The second public housing neighborhood in City 2 consisted of 108 low-rise barrack-style buildings housing 432 apartments occupying five square city blocks.
Recruitment efforts in City 2 yielded a sample of 164 African American adolescents age 11 to 19, with a mean age of 15.7 with a standard deviation of 2.3 years. Males (60 percent) had a slightly higher representation than females (40 percent). Fifty-seven percent lived in the first housing development and 43 percent lived in the second housing development. Data collection was completed in City 2 in the fall of 2006.
CITY 3: NEW YORK CITY (QUEENS)
City 3 included the largest high-rise housing development in the United States. It is divided into northern and southern sections, which[em dash]at the behest of residents—were treated as independent developments. Together, these two developments included 96 six-story buildings and housed 3,142 apartments. It occupied approximately 10 city square blocks and was home to slightly less than 3,000 families. In 2006, the median household income was slightly more than $20,000. Minority families (60 percent African American and 33 percent Latino) represented more than 90 percent of the families in this housing development. Seventy-five percent of the population was under the age of 18 years, with 60 percent being between the ages of 10 and 18 (Department of Resident Services 2007).
Recruitment efforts in City 3 yielded a sample of 237 African American adolescents age 11 to 19, with a mean age of 14.9 and a standard deviation of 2.4 years. Males (48 percent) had a slightly lower representation than females (52 percent). All participants lived in one housing development. Data collection was completed in City 3 in the fall of 2007.
CITY 4: PHILADELPHIA
City 4 included two housing developments: one two-story barrack-style housing development and one development with two 17-story high-rises buildings. The first development in City 4 consisted of 43 buildings containing 535 units; the two 17-story high-rise buildings in the second development hosted 499 units. The two developments had approximately 2,230 residents, with 63 percent under the age of 21. Ninety-eight percent of the residents were African American. The 2007 median household income in the two housing developments was approximately $7,500.
Recruitment efforts in City 4 yielded a sample of 149 African American adolescents age 12 to 20 years, with a mean age of 15.6 and a standard deviation of 2.4 years. Fifty percent of the sample was female. Youth resided in two housing developments. Data collection was completed in City 4 in the fall of 2007.
Sample Characteristics
Our recruitment efforts yielded a total sample of 898 youth living in public housing neighborhoods across four major U.S. cities: 26.5 percent from St. Louis, 18.3 percent from Washington DC, 38.6 percent from New York City (Queens), and 16.6 percent from Philadelphia. The sample reported an average age of 15.4 years, with a standard deviation of 2.3 years. Females represented 52.3 percent of the sample. Eighty-seven percent of the sample reported being African American, whereas the next largest group (9.5 percent) reported being of mixed race. The sample reported 4.7 family members in their household at the time of data collection.
Eighty percent of participants reported that their maternal caregiver was their biological mother, and 58 percent reported that their paternal caregiver was their biological father. Forty-three percent reported having both parents in their household, 80 percent reported having their mother in their household, and 56.4 percent reported having their father in their household. The sample reported that their nuclear family has lived in their public housing neighborhoods for an average of 13 years, with a standard deviation of 12 years. See tables 4.1 to 4.4 for descriptive statistics on the sample and comparisons among cities and between genders.
Limitations of the Data
These data will advance knowledge on a difficult-to-reach and scarcely studied population, but the limitations should be acknowledged. First, the subsample for each city may not include all measures, which inevitably determines what variables are included in the empirical chapters. The St. Louis subsample does not include measures of community and domestic violence, health-risk behaviors (sex and drug use), and parental behavior by the gender of the parent. Only the subsample for Philadelphia includes measures on emotional and practical support and help-seeking behavior.
Second, the empirical chapters are based on cross-sectional data. The chapters, therefore, only portray a snapshot of the lives of adolescents living in urban public housing. Causality and effects over time cannot be assessed on the basis of these data.
Third, the empirical chapters are based on a nonprobability sample. That is, sampling techniques included purposive and convenient sampling. We also used snowball sampling to gain access to youth who are not connected to mainstream institutions (e.g., schools, community centers, social service agencies). It should also be noted that, due to the difficulties associated with recruiting urban African American adolescents living in public housing in research, rigorous probability methods might not have yielded adequate sample size. Also, in light of the fact that this population has a near absence in empirical research, this nontraditional recruitment strategy was employed.
Finally, the results are based on a sample of youth living in public housing in large urban cities. Therefore, generalizing these findings to youth living in nonurban public housing and urban youth not living in public housing should be done with caution.
