Abstract
School violence and bullying are two public health concerns with consequences for youth in and out of school, for families, students, and community members. In this article, a social-ecological framework is briefly described as a way to understand bullying and school violence; then the National Center for Educational Statistics (NCES) longitudinal and cross-sectional data sets are described in detail. Data that assess bullying and/or school violence are described, and recommendations for additional items are proposed. In general, a longitudinal, multisite, multi-informant study is needed to address definitional and etiological issues related to school violence and bullying so that prevention efforts can be developed, implemented, and evaluated that incorporate multiple levels of the ecology, including peers, schools, communities, and neighborhoods.
A recent National Center for Educational Statistics (NCES) report indicated that about a third of students in Grades 9 to 12 reported they had been in a physical fight at least one time during the previous 12 months anywhere, and 12% said they had been in a fight on school property during the previous 12 months (Robers et al., 2013). Rates of victimization were similarly high. Approximately 28% of 12- to 18-year-old students reported they had been bullied at school during the school year, and victimization was highest among sixth graders (37%), compared to seventh or eighth graders (30% and 31%, respectively). Also, approximately 9% to 11% of youth report being called hate-related words having to do with their race, religion, ethnic background, and/or sexual orientation (Robers et al., 2013).
Although research on bullying and school violence has increased exponentially over the past decade, many questions remain about these phenomena that require additional research. In order to guide future research, the objectives of this article are to (a) briefly review etiological underpinnings of bullying and school violence using a social-ecological framework, (b) describe survey items/scales in each NCES data set and publication that addresses bullying and school violence, and (c) discuss data needs for selected emerging issues in bullying and school violence research and how NCES can or cannot address these issues. In order to determine how the data have been used by scholars (e.g., dissertations, peer-reviewed articles, chapters) across the disciplines of education, psychology, criminology, social work, and so on, I conducted searches in PsycINFO and MEDLINE. Given the space limitations here, none of these sections is exhaustive in nature, but each provides a brief commentary to provide guidance for scholars who might want to address their research questions with NCES data sets.
Etiological Underpinnings of Bullying and School Violence: Social-Ecological Perspective
School violence, bullying, aggression, and peer victimization are best understood from an ecological perspective in which individual characteristics of children interact with environmental factors to promote victimization and perpetration (Basile, Espelage, Rivers, McMahon, & Simon; 2009; Espelage, 2004, 2012; Hong & Espelage, 2012). An ecological framework has been used to explore the risk and protective factors of bullying and peer victimization whereby child and adolescent behavior is shaped by a range of nested contextual systems, including families, peers, and school environments (Bronfenbrenner, 1979; Espelage, 2015; Hong & Espelage, 2012). The ecological perspective provides a conceptual framework for investigating the independent and combined impact of these social contexts and dynamic, transactional influences on behavioral development. The microsystem includes structures with which children and adolescents have direct contact, such as families, peers, and schools. The mesosystem, another component of the ecological framework, comprises the interrelations among microsystems. This ecological framework has been applied to the conceptualization of bullying perpetration and victimization and school violence, and it highlights reciprocal influences on bullying behaviors among the individual (e.g., disability, race/ethnicity, immigrant status, biological sex, social skills), family (e.g., family violence, parental monitoring, family closeness), school (e.g., school climate, discipline policies, teacher-student relations, poverty, bully policies, prevention efforts, classroom factors), peer group (e.g., norms supportive of aggression, gender-based bullying, delinquency), community (e.g., collective efficacy in neighborhoods, safety issues, exposure to violence), and society (e.g., legislative policies around school violence, media influences). For exhaustive reviews, please refer to Espelage (2015) and Hong and Espelage (2012).
As detailed in this article, NCES data sets have been used to address several critical aspects associated with bullying and school violence (e.g., trends, immigrant populations, school safety issues). As it is becoming increasingly difficult to engage in school-based research, individual scholars are relying on nationally representative data sets to address many emerging issues in bullying and school violence. Thus, in order to understand how NCES data sets have informed and can inform (with proposed modifications) the social-ecological understanding of bullying and school violence, the next section will include a detailed discussion of the items/scales in each of the NCES data sets.
NCES Data Sets and Publications
NCES High School Data Sets and Publications
NCES has produced two longitudinal high school studies, the Educational Longitudinal Study–2002 (ELS:2002) and the High School Longitudinal Study–2009 (HSLS:09). The ELS:2002 is the fourth in a series of school-based longitudinal studies. Approximately 52 dissertations and 40 peer-reviewed articles have included these data in studies on achievement and school success, and six of the dissertations examined school safety as predictors of school success. ELS:2002 was conducted with 10th graders in 2002 and then assessed them as 12th graders in 2004, and two additional assessments were performed in 2006 and 2012. In the 10th-grade assessment, the student survey included individual items that assessed school violence: (1) “In class I often feel ‘put down’ by my students,” (2) “I don’t feel safe at this school,” (3) “There are gangs in school,” and (4) “Fights often occur between different racial/ethnic groups.” A bullied victimization item was embedded in a larger school violence scale: (1) “Had something stolen from me at school,” (2) “Someone offered to sell me drugs at school,” (3) “Someone threatened to hurt me at school,” (4) “I got into a physical fight at school,” (5) “Someone hit me,” (6) “Someone used strong-arm or forceful methods to get money or things from me,” (7) “Someone purposely damaged or destroyed my belongings,” and (8) “Someone bullied me or picked on me.” In addition, students were asked to name a few of their friends, and then they answered some questions about their friends’ behaviors, which would allow for studies on peer influence of behaviors, which are important in the socialization of bullying (Espelage, Holt, & Henkel, 2003). Only four of these items were included in the 12th-grade assessment (something stolen at school, someone offered to sell you drugs at school, someone threatened to hurt you, someone hit you at school). The parents’ assessment includes information about their child’s disability, questions related to language, and their level of school contact, and they also named some of their youth’s friends and asked questions about their friends. This data set has been used in cross-sectional analyses and published in peer-reviewed outlets to investigate victimization in after-school programs (Peguero, 2008; Ripski & Gregory, 2009), to examine violence and victimization experiences among children of immigrant parents (Peguero, 2009a, 2009b), and to examine school safety issues as predictors of teacher-rated academic achievement (Nelson & Gastic, 2009). Although this data set is limited in its measurement of bully victimization with one item at the 10th grade, additional longitudinal analyses could examine victimization and school safety issues and predicted adult outcomes. However, there is no assessment of sexual harassment or teen dating aggression—two phenomena that are particularly relevant for adolescence (Espelage, Low, Anderson, & De La Rue, 2013).
The HSLS:09 is ongoing and includes a nationally representative, longitudinal study of over 23,000 ninth graders from 944 schools. Approximately six dissertations and three peer-reviewed articles have included these data with studies focused on math and science achievement. This data set includes surveys of students, their parents, math and science teachers, school administrators, and school counselors. Assessments were conducted in 2009 and 2012 (data available), and the second follow-up is scheduled for 2016. The 2009 and 2012 student assessments include one item related to school safety: Students were asked to respond to the statement, “I feel safe at school.” No other items related to school violence or bullying were assessed, and parents and counselors were not queried on these issues.
NCES Early Childhood Data Sets and Publications
NCES has three longitudinal studies within the Early Childhood Longitudinal Study (ECLS) program (http://nces.ed.gov/ecls) that examine early childhood experiences and later academic outcomes. The three longitudinal studies within the program include an ecological examination of how the home, school, and community environments shape the development of externalizing behaviors (e.g., disruptive, aggressive) in children. These studies are designed with a strong theoretical base and have allowed for the direct examination of the potential mechanisms by which youth become involved in school violence or aggression (the kindergarten class of 2010–2011 [ECLS-K:2011], specifically). First, the birth cohort of the ECLS (born in 2001) study was followed from birth through kindergarten entry (at 9 months of age, 2 years, preschool, and kindergarten). Approximately 33 dissertations and 55 peer-reviewed articles have used these data to examine food security, math, literacy, externalizing behavior, attention-deficit disorder, school readiness, father involvement, and so on. This study included direct assessments of the children (nurse, behavioral observations), self-administered teacher surveys, parent interviews, and self-administered parent surveys. A wide range of constructs was assessed across the ecology of the child, including prenatal care, developmental milestones (including social-emotional development), exposure to violence, community support, economic support, neighborhood quality, physical abuse of child, and relationship quality of parents. Teachers rated the child’s friendships (liked, annoys other children).
Second, the kindergarten class of 1998–1999 (ECLS-K) cohort is a sample of children followed from kindergarten through eighth grade. This is a longitudinal study that allows researchers to study how a wide range of family, school, community, and individual factors affected school performance. Approximately 115 dissertations and 129 peer-reviewed journal articles have included these data focused on a range of academic outcomes, and two papers examined externalizing behaviors. Parents, teachers, and students completed measures in kindergarten and first, third, fifth, and eighth grades. The Social Skills Rating Scale (Gresham & Elliott, 1990) was modified and completed by multi-informants to assess the youth’s social skills and problem behaviors (externalizing, internalizing). Third, the ECLS-K:2011 cohort is a sample of children followed from kindergarten through fifth grade (each year). One dissertation used these data to examine externalizing behaviors among rural youth (Sisson, 2015). These youth are currently in the third grade and will be assessed as fourth graders and fifth graders in spring 2015 and 2016, respectively. Like the ECLS-K, in the ECLS-K:2011, parents, teachers, and students complete measures to assess home, school, and community influences on academic and social outcomes. Parents, teachers, and students report on peer victimization experiences (adapted from Espelage & Holt, 2001; e.g., “other students called names, other students made fun of, other students picked on child”), and social skills, externalizing, and internalizing behaviors are also assessed via the Social Skills Rating System (Gresham & Elliott, 1990). Teachers also report on parental involvement for each student. These three studies demonstrate excellent examples of the importance of studying how early childhood experiences at home, in school, and in the community all interact to influence later academic and social engagement. It would be important to keep the victimization scale in the 2015 and 2016 assessments.
NCES Cross-Sectional Data Sets
In addition to these longitudinal studies, NCES publishes the Indicators of School Crime and Safety every 2 years (see Robers et al., 2013, for more details), which includes a summary of a series of cross-sectional surveys of students, teachers, and principals. Sources for this report include the School-Associated Violent Death Study, funded by the Department of Education, the Department of Justice, and the CDC; the National Crime Victimization Survey (NCVS; U.S. Department of Justice, 2011) and School Crime Supplement (SCS; http://nces.ed.gov/surveys/); the Youth Risk Behavior Survey (www.cdc.gov/yrbss); the Schools and Staffing Survey (SASS; http://nces.ed.gov/surveys/); and the School Survey on Crime and Safety (SSOCS; http://nces.ed.gov/surveys/).
SCS to the NCVS
The SCS was created as a supplement to the NCVS and was codesigned by the NCES and Bureau of Justice Statistics. The SCS survey collects information about victimization, crime, and safety at school. The SCS is a national survey of approximately 6,500 students ages 12 through 18 in U.S. public and private elementary, middle, and high schools. The SCS was conducted in 1989, 1995, 1999, 2001, 2003, 2005, 2007, 2009, and 2011. Constructs include alcohol and drug availability; fighting, bullying, and hate-related behaviors; fear and school avoidance behaviors; gun and weapon carrying; and gangs at school. Approximately 25 peer-reviewed articles and six dissertations have used these data. A particular strength of this survey is the inclusion of promotive and protective factors, including involvement in student government, clubs, and athletic teams; perceptions of school rules and equity; perceptions of teacher-student relations; and neighborhood scales. The survey also assesses avoidance behaviors associated with school safety issues, including locations (e.g., hallways, cafeteria), activities, classes, and truancy. Nine items address security measures (e.g., metal detectors, code of conduct), and students are asked if they reported bullying to a teacher or some other adult. This survey also includes an assessment of gang activity or involvement with three questions: (1) “Are there any gangs at your school?” (2) “During this school year, how often have gangs been involved in fights, attacks, or other violence at your school?” (3) “Have gangs been involved in the sale of drugs at your school during this school year?” NCES should consider assessing gang membership and ask if students were recruited but did not join a gang to identify what protective factors might lead to resistance of gang membership (De La Rue & Espelage, 2014). Three items for cyberbullying were added to the 2009–2010 survey. This survey is limited in that it is cross-sectional and self-report, and data cannot be linked to school-level or community-level characteristics. In reality, these data are useful for surveillance of school violence and bullying only and offer little in the way of addressing some of the major pressing research issues in the fields of school violence and bullying.
SSOCS
The SSOCS was the primary source of school-level data on crime and safety for NCES. It includes a nationally representative cross-sectional survey of about 3,500 public elementary and secondary public schools. SSOCS is administered to over 3,000 public primary, middle, high, and combined school principals in the spring of even-numbered school years. Data collection occurred in the spring of the 1999–2000, 2003–2004, 2005–2006, 2007–2008, 2009–2010 (data not available yet), and 2011–2012 school years (questionnaire and data are not on website). The survey includes items related to school practices and programs; parent and community involvement at school; school security; staff training; limitations on crime prevention; frequency of crime and violence at school; frequency of incidents reported to police or law enforcement; frequency of hate crimes, gang-related crimes, and gang-related hate crimes; disciplinary problems and disciplinary actions; and so on. Approximately six peer-reviewed articles and five dissertations have used these data. Furthermore, the SSOCS includes an assessment of school security practices and policies related to technology use, and written plans and drills for violence incidents. Finally, detailed questions regarding the role and involvement of school resource officers, security guards, or sworn law enforcement officers are included. These data provide descriptive information about what schools are doing to address school safety and security efforts, and how the security officers are involved in efforts to implement and enforce these efforts and how they are involved in teacher training efforts. Principals are provided with definitions of gang, hate crime, rape, sexual battery, sexual harassment, theft, and violence. No definition of bullying is provided, but cyberbullying is defined in the body of the survey. One item for bullying (“How often does student bullying occur at your school?”) is assessed. Again, this survey is limited by its cross-sectional nature, use of a single informant, and limited psychometric evidence for the constructs, including the one bullying item. NCES should consider how these data could be linked to other data sources.
NCES SASS
The SASS is a system of related questionnaires that provides descriptive data on the context of elementary and secondary education and provides policymakers a variety of statistics on the condition of education in the United States. It covers a wide range of topics, including teacher demand, teacher and principal characteristics, general conditions in schools, principals’ and teachers’ perceptions of school climate, and problems in their schools. The four components are the School Questionnaire, the Teacher Questionnaire, the Principal Questionnaire, and the School District Questionnaire. It includes comprehensive assessments of training, professional development training, and violence directed toward teachers. Approximately 80 peer-reviewed articles and over 100 dissertations have used these data. The most recent data collection was for 2011–2012, and it is now being changed to the National Teacher and Principal Survey. Of note is that the Teacher Questionnaire was instrumental in drawing national attention to the substantial number of teachers and paraprofessionals who had experienced violence in their classrooms and schools, but it failed to uncover those aspects of schools, communities, or school leadership that predicted the prevalence of teacher violence. Thus, the American Psychological Association appointed a task force that used the prevalence data to design a national study of National Education Association members (Espelage, Anderman, et al., 2013). NCES should consider a longitudinal, multilevel examination of violence directed toward teachers, staff, and principals that would examine what factors predict violence and what factors are associated with less violence, and it should consider developing interventions to reduce the growing teacher attrition that is linked to disruptive behaviors in the classroom.
Selected Emerging Research Issues on School Violence and Bullying
Whereas some research indicates that extreme forms of school violence are decreasing in prevalence (Robers, Zhang, & Truman, 2010), bullying and school violence in general and their aftermath continue to be significant problems for students, teachers, staff, and schools. Thus, there continues to be many pressing issues in the research literature on school violence and bullying, some of which are discussed next.
Definition of Bullying and Peer Victimization
Almost four decades of research has been conducted on bullying and peer victimization, ranging from prevalence studies and etiological investigations to systematic reviews and evaluations of prevention and intervention programs (see Espelage & Holt, 2012, for review). However, a rigorous debate has only recently emerged about how best to define bullying and how to distinguish it from other forms of aggression and/or peer victimization (American Educational Research Association, 2013; Rodkin, Espelage, & Hanish, 2015). This is an important issue to resolve as most states are required to have a bully definition, a policy, and a prevention plan. Definitions of bullying emphasize observable or nonobservable aggressive behaviors, the repetitive nature of these behaviors, and the imbalance of power between the individual or group perpetrator and the victim (Espelage & Holt, 2012; Ybarra, Espelage, & Mitchell, 2014). An imbalance of power exists when the perpetrator or group of perpetrators have more physical, social, or intellectual power than the victim. In a recent examination of a nationally representative study, early and late adolescents who perceived that their perpetrator had more power than them reported greater adverse outcomes (e.g., hopelessness) than victims who did not perceive a power differential (Ybarra et al., 2014). Finally, repetition should be evaluated carefully, given that youth who are victims of bullying often change their behaviors in order to minimize the probability of it happening again (Batanova, Espelage, & Rao, 2014). For example, they might stop riding the bus, stop attending lunch, or avoid school and other places where the victimization is occurring.
The SCS to the NCVS 2011 is the only source of data from students about their experiences of bullying. Students are presented with the following question to assess offline (face-to-face) bully victimization: “Now I have some questions about what students do at school that make you feel bad or are hurtful to you. We often refer to this as being bullied. During this school year, has any student bullied you?” Then students respond yes or no to seven items: name calling, rumor spreading, threatened, physical, coercion, excluded, and destroyed property. Online bully victimization is assessed with the following item, to which the students respond yes or no: Now I have some questions about what students do that could occur anywhere and that make you feel bad or are hurtful to you. We often refer to this as being bullied. You may include events you told me about already. During this school year, has another student posted hurtful information; threatened by email, instant messaging, text messaging, or through games; or excluded?
This assessment is not consistent with the national research definition of bullying. Of note, an expert panel convened in the fall of 2013 to provide suggestions for revisions of this supplement, and I suggested the following recommendations: (a) Assess the power differential between the perpetrator and victim (When you are bullied, is it by a person bigger, older, stronger, more popular?), (b) add response options to assess frequencies (never, one or two times, three or four times, and so on) for each item, and (c) make offline and online items parallel to allow direct comparisons of prevalence; perpetration should be added across all constructs.
Bullying, Peer Victimization, Academic Achievement, and Engagement
Several national and international research studies relying on cross-sectional data have documented that experiences of being victimized or bullying other students are associated with decreased academic achievement. Findings from a sample of seventh, ninth, and 11th graders in an urban public school district also revealed that for each one-point increase in grade point average, the odds of being a victim versus a bystander decreased by 10% (Glew, Fan, Katon, & Rivara, 2008). These associations are also found when students are followed over time in longitudinal studies (e.g., Juvonen, Wang, & Espinoza, 2011; Schwartz, Gorman, Nakamoto, & Toblin, 2005). Juvonen and colleagues (2011) documented that peer victimization can account for an average 1.5-point letter grade decrease in one academic subject (e.g., math) across 3 years of middle school. Moreover, the researchers found that greater self-reported victimization was associated with lower grades and lower teacher-rated academic engagement. However, a recent meta-analytic review of 33 studies conducted by Nakamoto and Schwartz (2010) reported that empirical research on this association has produced an incongruent pattern of findings and modest correlations. Much more research is needed to understand how victimization impacts academic engagement, school avoidance, grades, and ultimately, career placement. The ELS:2002 could be used to examine some of these associations because bully victimization, school safety issues, and school violence were assessed at baseline.
Bullying Among Students With Disabilities
Research indicates that students with disabilities are twice as likely to be identified as perpetrators and victims than students without disabilities (Rose, 2010; Rose & Espelage, 2012). Disability data were collected in the NCES longitudinal studies from parents and have been used in a limited way in the literature. NCES should create a mechanism for greater mining of these data, perhaps through a call for proposals to use these data or creating postdoctoral positions that require working with the NCES data on disability issues.
Bullying and the Lesbian, Gay, and Bisexual (LGB) Community
A large percentage of bullying among students involves the use of homophobic teasing and slurs, called homophobic teasing or victimization (Espelage, Basile, & Hamburger, 2012; Poteat & Espelage, 2005; Poteat & Rivers, 2010). Bullying and homophobic victimization occur more frequently among LGB youth in American schools than among students who identify as heterosexual (Espelage, Aragon, Birkett, & Koenig, 2008; Kosciw, Greytak, & Diaz, 2009; Robinson & Espelage, 2011, 2012). Some LGB youth report greater depression, anxiety, suicidal behaviors, and truancy than their straight-identified peers (Espelage et al., 2008; Robinson & Espelage, 2011). However, peer victimization does not appear to explain all of the mental health disparities between LGB and heterosexual youth (Robinson & Espelage, 2012). It is imperative that NCES conduct a longitudinal study of the experiences of victimization associated with sexual orientation, gender expression, or gender identity.
Summary and Recommendations
NCES has two longitudinal high school studies (ELS:2002 and HSLS:2009). The ELS:2002 data set includes an assessment of bully victimization (one item) and school violence (scale) at baseline. Papers have appeared in peer-reviewed outlets examining victimization experiences among children of immigrant parents and victimization experiences in after-school programs. Much more work could be conducted with this data set, but the single-item bully victimization assessment is a major limitation. The HSLS:2009 included only one item related to school violence: Students were asked how safe they felt at school. The three longitudinal studies within the ECLS program include an ecological examination of how the home, school, and community environments shape the development of externalizing behaviors in children. These studies are designed with a strong theoretical base and allow for the direct examination of the potential mechanisms by which youth become involved in school violence or aggression (ECLS-K:2011 specifically). It would be important that NCES follow the ECLS-K:2011 youth and families into early and late adolescence. This would allow the addition of different types of violence (perpetration and victimization), including bullying with more precise measurement, teen dating violence, and sexual violence. Furthermore, other at-risk behaviors that have direct and indirect influences on academic success (e.g., gang involvement, drugs and alcohol use, exposure to violence) as well as protective factors (e.g., extracurricular activities, positive youth activities, community support, and neighborhood collective efficacy) should be added to these later measurements.
That said, many questions remain as NCES attempts to address the public health concerns of school violence and bullying. It would be important within the surveys used for surveillance and trend analysis to distinguish bullying (power dynamic) from student aggression. Additional items should be added to those available without compromising the ability to track trends over time. Perhaps, within these cross-sectional studies, multi-informant data could be collected that could be matched to determine how perceptions of teachers, schools, parents, and community members influence youth outcomes. At the very least, these data could be analyzed using a multilevel framework. Also, objective measures of discipline should be tied to student outcomes more specifically to determine if the increase in security measures has an impact of reducing school violence and promoting school safety.