Segregation,Securitization,and Bullying: Investigating the Connections Between Policing,Surveillance,Punishment,and Violence

Introduction

Although school segregation was outlawed more than six decades ago by the U.S. Supreme Court in Brown v. Board of Education, U.S. public school systems remain heavily segregated across race and ethnicity. Numerous studies on school segregation discovered that it resulted in higher racial and ethnic disparities in education, employment, and criminal justice system outcomes (Kim et al. 2010Peguero & Hong, 2020; Reardon & Owens, 2014; Rios, 2011, 2017; Shedd, 2015). These findings have been brought to the attention of local policymakers and stakeholders to promote educational equity in order to improve the outcomes of students who belong to marginalized groups. Yet, racial and ethnic segregation is also linked to the uneven distribution of resources across public school systems (e.g., districts or cities), which in turn leads to greater teacher turnover rates, student dropout rates, social disorder, and violence in schools with larger shares of racial and ethnic minority students (Lopez-Aguado, 2018; Morris, 2016; Owens & McLanahan, 2020; Sewell, 2020). Thus, not only are racial and ethnic minority students in public schools at a significant disadvantage due to the empirical effects of school segregation, but they are also subject to greater scrutiny by school officials due to the concentrated levels of delinquency and violence.

Because of the hyper-criminalization of racial and ethnic minorities across the U.S., public schools with large populations of racial and ethnic minority students experience greater levels of over-policing, surveillance, and punishment (Irwin et al., 2013; Owens & McLanahan, 2020). This hyper-criminalization is fueled by widespread stereotypes and narratives of racial and ethnic minority youth, especially Black and Latina/o/x students, as truant, dangerous, delinquent, and violent. Not only does this public discourse, which is pervasive throughout the news and social media, support the hyper-criminalization of these students, but it also produces complex and long-lasting detrimental consequences for racial and ethnic minority youth. For example, scholars have argued that this hyper-criminalization facilitates the school-to-prison pipeline, disrupts educational progress and success, forms mistrust between students, teachers, and school administrators, and fosters bullying and aggressive behaviors among youth (Kim et al. 2010Morris, 2016; Rios, 2011, 2017; Shedd, 2015). Numerous studies have focused on the relationships between race/ethnicity, hyper-criminalization, and bullying in schools looking at individual-level occurrences (Bradshaw et al., 2013; Peguero, 2012; Peguero & Williams, 2013). Nevertheless, there is limited knowledge about these aforementioned patterns at the school-level.

Over the past half century, the number of students of color enrolled in U.S. elementary and secondary public schools has tripled from 8.3 million in 1968 to 25.5 million in 2016 (Frankenberg et al., 2019). Because this share of the population is projected to continue growing over the next decade, it is critical to develop a better understanding on the experiences of racial and ethnic minority students with regard to the criminalization of youth and bullying in public schools in order to ensure their future success. This study seeks to extend prior research on school segregation and criminalization by examining the associations between criminalization and bullying within majority Black, Latina/o/x, and White schools (Peguero & Hong, 2020; Reardon & Owens, 2014; Rios, 2011, 2017; Shedd, 2015). Using data from the 2015–2016 School Survey on Crime and Safety (SSOCS), which collects information from a nationally representative sample of U.S. public elementary and secondary schools, we address two overarching research questions. First, do racially/ethnically segregated schools have higher rates of policing, surveillance, and punishment? Second, do policing, surveillance, and punishment within segregated schools moderate the rate of bullying? Our findings reveal important connections between policing, surveillance, punishment, and bullying within segregated schools; however, important and distinctive nuances are presented and examined. Finally, we discuss major implications on the complex relationship between segregation and criminalization especially considering the importance of providing a safe and healthy educational learning environment for all students.

Segregation and Schooling

After the U.S. Supreme Court declared racial segregation in public schools unconstitutional in Brown versus Board of Education in 1954, efforts to desegregate public schools were met with massive resistance from local communities following a series of courtroom cases, widespread protests, and the passage of legislation. By 1968, the U.S. Supreme Court returned to this issue in Green versus County School Board of New Kent County and ruled that efforts to integrate schools must meaningfully decrease segregation levels. This decision hastened efforts to desegregate schools through court-ordered plans that integrated Black students into majority-White schools in the 1970s (McGrew, 2019). The racial and ethnic makeup of students in U.S. public schools has since changed dramatically over the last few decades. While the share of Black students enrolled in U.S. public schools remains at around 15 percent since 1968, the share of White students dropped from about 79 to 48 percent and the share of Latina/o/x students dramatically grew from about 5 to 26 percent (Frankenberg et al., 2019). The Asian student population also grew from less than 1 to about 5 percent of those enrolled in U.S. public schools since 1968. Future national projections indicate that the share of White students will continue to decline while the shares of Latina/o/x, Asian, and multiracial students will continue to rise in U.S. public schools over the next decade (National Center for Education Statistics, 2020).

Despite the changing landscape in the student enrollment from a majority White student population to a majority-minority student population, U.S. public schools remain extremely segregated today. Although school segregation declined following the court-ordered desegregation plans in the 1970s, scholars report that this decline lasted through the mid-1980s before rising again (Reardon & Owens, 2014). In a national report on school segregation, for example, Frankenberg et al. (2019) noted that “the share of ‘intensely segregated minority schools,’ which are schools that enroll 90–100% non-White students, have more than tripled from 5.7% in 1988 to 18.2% in 2016” (p. 21). Using national data from the Department of Education, they found that about 40 and 41.6 percent of all Black and Latina/o/x students, respectively, were attending “intensely segregated minority schools” in 2016. This represented an increase by about 8 percent for both groups of students since 1988 in large part due to the absence of public policies aimed at encouraging integration and the release of court-ordered desegregation plans (Frankenberg et al., 2019; Reardon & Owens, 2014). Thus, the experience of Black and Latina/o/x students, who attend “intensely segregated minority schools,” does not mirror the racial and ethnic diversity of students enrolled in U.S. public schools.

School segregation is a major issue in the U.S. for several reasons. First, schools represent a pivotal agent of socialization in the lives of children and adolescents and therefore, racially segregated schools raise concerns across psychological, relational, and societal domains. Second, the lack of resources and funding that is disproportionately allocated to public schools attended by majority-Black and majority-Latina/o/x students is linked to poor educational achievement and attainment. Finally, the overlap between racially/ethnically segregated schools and concentrated poverty places students who attend these public schools at a greater disadvantage. For example, Frankenberg et al. (2019) reported that one-half of the “intensely segregated minority schools” had a large majority of students who qualified for free or reduced priced lunch. In contrast, less than one-third of U.S. public schools where Black and Latina/o/x students made up less than 10 percent of the student body had a large majority of students who qualified for free or reduced priced lunch. Using free or reduced priced lunch eligibility as a measure of poverty, Frankenberg et al. (2019) argue that “schools with racial segregation and schools with concentrated poverty both produce less academic success for students, and most of the schools that rank high on either measure have both” (p. 11).

Criminalization and Schooling

Since the 1980s, the number of people incarcerated in the U.S. has increased by about 500 percent with nearly 2 million individuals currently serving time behind bars. While the number of incarcerated individuals has started to decline over the last decade, the racial and ethnic disparities among those incarcerated remains. For example, Black individuals are incarcerated in state prisons at nearly 5 times the rate compared to White individuals and Latina/o/x individuals are 1.3 times more likely to be incarcerated than non-Latinx White individuals (Nellis, 2021). Because of the staggering racial and ethnic disparities among those imprisoned, numerous scholars have argued that the criminal justice system operates as a system of racialized social control that serves in the creation and maintenance of a racial under-caste (Alexander, 2010; Chin et al., 2019; Rabaka, 2010). Like mass incarceration, the concept of mass criminalization is conceptualized as “targeted racial discrimination, or ‘the interconnections and intersections of white supremacy within the criminal justice system and seemingly neutral social institutions” (Rabaka, 2010, p. 308). The “archetype” of the criminal is steeped in anti-Blackness that is veiled in race-neutral language drawn from carceral vocabulary and rationale. Thus, criminalization extends beyond the criminal act to include surveillance, discipline, and punishment for violations of White cultural norms, such as choices in dress, hairstyles, tone, language, behaviors. Although “Blackness” establishes a process for criminalization, this process is realized throughout the U.S. on people of color including Latina/o/x youth (Chin et al., 2019; Morris, 2016; Rios, 2011, 2017).

As noted earlier, criminalization extends far beyond the criminal justice system to other public institutions such as schools (Annamma, 2016, 2017; Irwin et al., 2013; Morris, 2016; Rabaka, 2010; Rios, 2011, 2017). While criminalization is found in the policies and practices of schools serving majority-Black and majority-Latina/o/x students, it is more prevalent in schools that exist within socioeconomically disadvantaged neighborhoods through security and surveillance mechanisms that facilitate and support the criminalization process (Irwin et al., 2013; Kupchik & Ward, 2014). The school-to-prison pipeline is a metaphor often used by researchers, activists, and the media to describe the criminalization of students in schools through a fusing of the carceral state and institutions of education (Alexander, 2010; Annamma, 2016, 2017; Meiners, 2011; Morris, 2016). As a concept, it works to “highlight a complex network of relations that naturalize the movement of youth of color from our schools and communities into under- or unemployment and permanent detention” (Meiners, 2011, p. 550). The school policies and practices focused on surveillance and punishment create an environment where delinquent labeling and contact with the criminal justice system are more likely to occur. These include physical security measures (e.g., security cameras, metal detectors, drug-detection dogs, security guards); the presence of police on campus (e.g., school resource officers, sworn law enforcement officers, on-campus police department); and exclusionary discipline practices (e.g., suspension, expulsion) (Annamma, 2017). Overall, the goal of surveillance “is not to surveil all bodies, but to socially and spatially monitor black and brown bodies” (Annamma, 2016, p. 3). Thus, it is Black and Latina/o/x students, their families, and their communities who carry the weight of this system.

The Potential Relationship Between Segregation, Criminalization, and Bullying

Over the last few decades, an increasing number of studies have examined the relationship between race, ethnicity, and bullying (Xu et al., 2020). School bullying, which is typically defined as intentional and repeated aggression against someone who cannot easily defend themselves, can take different forms such as physical, verbal, and relational (e.g., social exclusion). With the rising use of technology by students in school, bullying has also taken the form of cyberbullying, which is bullying through digital devices such as cell phones, computers, and tablets. Earlier studies have discovered that some racial and ethnic minority youth are exposed to more bullying than majority groups while others experience less. For example, DeVoe and Kaffenberger (2005) found that White and Black youth were more likely than Latina/o/x youth to report being victims of indirect bullying. On the other hand, some studies point out that Asian and Latina/o/x youth are more likely to be bullied or targeted in schools because of their ethnicity as well as the long-standing discrimination against “foreigners” in comparison to other youth such as White and Black youth (Hong et al., 2014; Zhang et al., 2021). Other studies focusing on cyberbullying, however, found that White and multi-racial youth were more vulnerable to cyberbullying victimization than Black and Latina/o/x youth (Edwards et al., 2016; Hinduja & Patchin, 2021; Pontes et al., 2018). Looking at bullying and cyberbullying, Wang et al. (2009) found that Black youth were less likely to be victims of verbal and relational bullying than White youth, but not physical or cyberbullying. These inconsistent findings suggest that race and ethnicity alone are not adequate predictors of bullying and cyberbullying, but rather other factors such as social, cultural, and environmental context in schools might better predict bullying and cyberbullying (Bradshaw et al., 2013; Huang & Cornell, 2019; Juvonen & Graham, 2014; Peguero, 2012; Peguero & Hong, 2020).

Although research on race, ethnicity, and school bullying has mostly focused on students’ experiences at the individual-level, studies have reported that school context may moderate this relationship (Xu et al., 2020). First, studies have shown that racial and ethnic minority youth enrolled in racially/ethnically segregated schools are more likely to be exposed to high aggression in the classrooms and therefore, may experience more bullying (Thomas et al., 1997). Second, some findings suggest that school officials’ tolerance for violence may be higher for bullying incidents involving Black and Latina/o/x students because violence is viewed as normative for these students while exposure to violence is less frequent among White youth (Peguero, 2012). Therefore, the attention and concern paid by school officials to Black and Latina/o/x youth for this behavior may be more lenient than the attention and concern provided to White youth. And yet, the hyper-criminalization of Black and Latina/o/x youth behavior in schools may lead officials to view bullying in racially/ethnically segregated schools, like other forms of student misbehavior, as requiring stricter control and punishment (Bradshaw et al., 2013; Huang & Cornell, 2019; Juvonen & Graham, 2014; Peguero & Hong, 2020). Thus, these inconsistencies in the literature require further investigation. Given the increased risk for negative mental health and behavioral outcomes following school bullying incidents, particularly those bias-based, there is a need for more research on the association between race, ethnicity, and school bullying. Many school bullying incidents involving derogatory comments from students about race/ethnicity in the classroom and students’ comments supporting the idea of White superiority have been documented in recent reports and may carry serious implications for their academic, physical, and psychological well-being (Mulvey et al., 2018; Rogers et al., 2017).

Current Study

As demonstrated in prior research, it is evident that racially/ethnically segregated schools have stringent and higher levels of policing, security, surveillance, and disciplinary practices (Bracy, 2011; Irwin et al., 2013; Rios, 2011, 2017; Shedd, 2015). Of course, ensuring the safety of all youth is paramount; however, it is important to understand the correlates (i.e., segregation and criminalization) that may be compounding educational inequity for an already marginalized segment of the student population. Majority-Black and majority-Latina/o/x schools are often embedded in a high concentration of low-income areas, situated in communities with increased disorder and violence, allocated limited resources, and underserved in the tools that would facilitate educational progress and success for students (Bracy, 2011; Bradshaw et al., 2013; Rios, 2011, 2017; Shedd, 2015). On the contrary, majority-White schools are often characterized as safer, better organized, affluent, having increased access to resources, and healthier school climates (Calarco, 2020; Hagerman, 2018; Sulak, 2016). Even though increased policing, security, surveillance, and zero-tolerance punishment practices have become common across all schools, the securitization, surveillance, and criminalization of students are disproportionately impacting majority-Black and majority-Latina/o/x schools (Kupchik, 2010, 2016; Muschert et al., 2013). What remains unknown, however, is how segregation and criminalization are contributing to rates of bullying in majority-Black, majority-Latina/o/x, and majority-White schools.

This study focuses on addressing the relationship between racial/ethnic segregation, criminalization, and bullying, which remains unclear as shown in previous literature. First, do racially/ethnically segregated schools have higher rates of policing, surveillance, and punishment? Second, do policing, surveillance, and punishment within segregated schools moderate the rate of bullying? This study seeks to contribute to racial/ethnic educational inequality by exploring whether there is an association between youth control complex and bullying in racially/ethnically segregated schools.

Method

Dependent Variables

This study includes two dependent variables. The first is an ordinal variable that measures the frequency of bullying in the school. Principals were asked, “To the best of your knowledge, how often do the following types of problems occur at your school? Student bullying.” SSOCS defined bullying as “any unwanted aggressive behavior(s) by another youth or group of youths who are not siblings or current dating partners that involves an observed or perceived power imbalance and is repeated multiple times or is highly likely to be repeated.” Our second variable is also an ordinal variable that measures the frequency of cyberbullying. Principals were asked, “To the best of your knowledge, thinking about problems that can occur anywhere (both at your school and away from school), how often do the following occur? Cyberbullying among students who attend your school.” In the questionnaire, SSOCS stated that cyberbullying “occurs when willful and repeated harm is inflicted through the use of computers, cell phones, or other electronic devices.” For both questions, principals were asked to select one of five response options: (1) happens daily, (2) happens at least once a week, (3) happens at least once a month, (4) happens on occasion, and (5) never happens. These categories were reverse coded so that a higher number represents more frequent occurrences of student bullying and cyberbullying (see Appendix Table 1 for weighted distributions for both outcomes).

OPEN IN VIEWER

Table 1.

Weighted Means for Independent Variables by School Racial Context.

	Majority White	Majority Black	Majority Latinx	Diverse
Surveilance measures	4.76	5.31*	5.06*	4.85
SROs	.60	.74	.55	.75
Disciplinary actions	6.68	22.81*	12.52*	14.27*
Elementary school	.55	.57	.68*	.64*
Middle school	.19	.20	.15*	.20
High school	.16	.16	.14	.13
Combined school	.10	.07	.03*	.03*
School enrollment	536.38	529.75	676.52*	678.13*
Student-teacher ratio	17.04	27.57	21.14*	18.38*
Community crime	.10	.65*	.57*	.34*
% Disadvanatged	43.02	83.16*	81.06*	61.16*
% Special education	13.79	14.78	12.71	13.86
% Male	49.22	49.23	49.85	49.53
% Underachievement	12.17	26.95*	25.15*	18.17*
% Daily attendance	93.81	92.95	93.39	94.43
Urban	.12	.62*	.49*	.46*
Suburb	.34	.16*	.33	.40
Town	.16	.08*	.10*	.07*
Rural	.39	.14*	.08*	.07*
Northeast	.18	.17	.16	.11*
Midwest	.34	.16*	.06*	.18*
South	.30	.64*	.34	.46*
West	.18	.02*	.44*	.25

Data Source: School Survey on Crime and Safety, 2015–16, N = 2,060.

Note: Significant differences compared to majority white schools denoted with asterisks. Significance is based on Welch's t-tests.

* p < .05 (two-tailed).

Control Variables

This study also includes several variables as model controls. First, a categorical measure of the school's grade level is used to identify schools as elementary, middle, high, or combined school (reference category). Next, school enrollment and student-teacher ratio are continuous variables that measures the total number of students in the school and the number of students per teaching staff (based on full-time equivalency), respectively. To measure the level of crime in the community (“community crime”) where the school is located, we use a dichotomous variable based on crime reported by school principals (0 = low crime, 1 = high/moderate crime). ¹ The following four continuous variables characterize the composition of students in the school: percent disadvantaged measures the percentage of students in the school who are eligible for free or reduced-price lunch; percent special education measures the percentage of students who receive special education services under the Individuals with Disabilities Education Act; percent male measures the percentage of students who are male; and percent underachievement measures the principals’ best estimate of the percentage of students who are below the 15^th percentile on standardized tests. Percent daily attendance is a continuous variable that measures the school's average daily attendance based on the percentage of students present. Lastly, categorical variables measure whether the school is in an urban, suburban (reference category), town, or rural locale, and in the Northeast, Midwest, South (reference category), or West region (see Appendix Table 2 for descriptive statistics of independent variables).

OPEN IN VIEWER

Table 2.

Ordinal Logistic Regression Models Predicting Bullying Among Students.

	Model 1		Model 2		Model 3
	B	SE	B	SE	B	SE
Majority Black	−.318	(.253)	−.408	(.255)	−.534	(.606)
Majority Latina/o/x	−.761**	(.247)	−.763**	(.246)	−1.094	(.627)
Diverse	−.264	(.196)	−.266	(.197)	−.482	(.489)
Surveilance measures			.108**	(.040)	.068	(.055)
x Majority Black					.102	(.108)
x Majority Latina/o/x					.086	(.103)
x Diverse					.093	(.092)
SROs			−.008	(.016)	.011	(.019)
x Majority Black					−.258	(.133)
x Majority Latina/o/x					−.017	(.043)
x Diverse					−.047	(.029)
Disciplinary actions			.008*	(.003)	.022***	(.005)
x Majority Black					−.016**	(.006)
x Majority Latina/o/x					−.010	(.007)
x Diverse					−.018**	(.005)
Elementary school	−.516*	(.261)	−.456	(.258)	−.451	(.259)
Middle school	.515*	(.250)	.399	(.248)	.381	(.248)
High school	−.210	(.261)	−.337	(.259)	−.363	(.260)
School enrollment	.001***	(.000)	.001***	(.000)	.000**	(.000)
Student-teacher ratio	−.001	(.001)	−.001	(.001)	−.001	(.001)
Community crime	.365*	(.176)	.306	(.177)	.315	(.177)
% Disadvanatged	.015***	(.003)	.013***	(.003)	.012***	(.003)
% Special education	−.003	(.014)	−.003	(.014)	−.002	(.014)
% Male	.015*	(.007)	.016*	(.007)	.016*	(.007)
% Underachievement	.006	(.005)	.005	(.005)	.006	(.005)
% Daily attendance	−.006	(.007)	−.006	(.007)	−.005	(.006)
Urban	.160	(.162)	.124	(.164)	.113	(.166)
Town	.596**	(.202)	.592**	(.202)	.552**	(.204)
Rural	.409*	(.188)	.403*	(.188)	.406*	(.189)
Northeast	.085	(.202)	.107	(.199)	.136	(.203)
Midwest	.464**	(.168)	.419*	(.170)	.443**	(.171)
West	.860***	(.163)	.917***	(.164)	.931***	(.164)
Threshold 1	−1.790	(.769)	−1.368	(.763)	−1.495	(.741)
Threshold 2	2.611	(.779)	3.058	(.772)	2.944	(.748)
Threshold 3	3.896	(.784)	4.361	(.776)	4.257	(.752)
Threshold 4	5.482	(.793)	5.970	(.784)	5.878	(.762)

Data Source: School Survey on Crime and Safety, 2015–16, N = 2,060.

Note: Variable reference categories include majority White school, combined school, suburban locale, and South region. Models present weighted estimates.

*p < .05, **p < .01, ***p < .001 (two-tailed).

Analysis Plan

Because this study examines patterns of bullying in racially segregated schooling contexts, the analytic sample includes schools that are majority-White, majority-Black, and majority-Latina/o/x, as well as schools with diverse student bodies. Schools with a majority of Asian/Pacific Islander, American Indian, or multiracial youth (30 in total) were excluded from analyses due to their small subgroup sample sizes. This resulted in an analytic sample of 2,060 schools. To address problems associated with missing data, SSOCS used imputation procedures, including aggregate proportions, hot deck, and clerical imputation to create values for all questionnaire items with missing information before releasing their data to the public. Accordingly, there are no cases with missing data in this sample. Analyses include the survey weight (FINALWGT) created by NCES (via Stata 16's svyset command and svy prefix) to produce population-based estimates and standard errors that account for SSOCS's complex sampling design. Additionally, we checked for but found no evidence of multicollinearity in our models, as both tolerance indices and variance inflation factors (reported in Appendix Table 3) failed to reach even conservative thresholds (i.e., VIF>5, TI<.20) (Chatterjee & Simonoff, 2013; O’Brien, 2007).

OPEN IN VIEWER

Table 3.

Marginal Effects Predicting Bullying Among Students by School Racial Context.

	Frequency	Majority White	Majority Black	Majority Latina/o/x	Diverse
Surveilance measures	Never	−.003	−.001	−.007	−.005
	Occassionally	−.010	−.016	−.020	−.021
	Monthly	.006	.007	.012	.014
	Weekly	.005	.007	.011	.008
	Daily	.002	.003	.004	.004
SROs	Never	.000	.003	.000	.002
	Occassionally	−.001	.068*	−.001	.008*
	Monthly	.000	−.029*	.000	−.005*
	Weekly	.000	−.029*	.000	−.003*
	Daily	.000	−.013	.000	−.002*
Disciplinary actions	Never	−.001**	.000	−.001*	.000
	Occassionally	−.003***	−.001	−.002***	−.001
	Monthly	.002***	.000	.001**	.000
	Weekly	.001***	.000	.001**	.000
	Daily	.001**	.000	.000*	.000

Data Source: School Survey on Crime and Safety, 2015–16, N = 2,060.

Note: Average marginal effects are calculated from ordinal logistic regression analyses predicting likelihood of bullying among students. Models also control for school level, total enrollment, student-teacher ratio, community crime, % disadvantaged, % special education, % male, % underachivement, % daily attendance, locale, and region.

*p < .05, **p < .01, ***p < .001 (two-tailed).

Data analyses were completed in several stages. To address the first research question, we began by estimating weighted means, disaggregated by school racial context, for each independent variable in the study. We also tested for significant differences in the means (compared to the majority-White schools) using Stata's lincom command. Next, we used ordinal logistic regression to predict the frequency of bullying and cyberbullying among students. Here, we estimated three regression models for each outcome. ² Model 1 includes the dichotomous measures of school segregation and school controls. In Model 2, we added independent variables measuring the number of surveillance measures employed, number of SROs at the school, and the number of disciplinary actions taken by administrators. To address the second research question, Model 3 includes the interactions between school racial contexts and the three surveillance and disciplinary measures. We also estimated separate ordinal regression models predicting the frequency of bullying and cyberbullying in majority-White, majority-Black, majority-Latina/o/x, and diverse schools. To facilitate interpretation and comparison across models, average marginal effects for each of the three variables of interest (surveillance measures, SROs, and disciplinary actions) were produced using Stata's margins command.

Results

Segregation and Bullying

Table 2 presents estimates from ordinal logistic regression models predicting the frequency of bullying at school. This study focuses on the role of school context in Model 1. According to the results, majority-Latina/o/x schools have significantly lower odds of reporting bullying than majority-White schools (b =  − .761, p < .01; OR = .467) net school controls. The coefficients for majority-Black schools and diverse schools are both smaller in magnitude and not significant, which means that there are no differences in the frequency of bullying between these schools and majority-White schools. The covariates in the other models also emerge as significant predictors of bullying in school. Bullying is significantly less likely to occur in elementary schools than in combined schools but significantly more likely to occur in middle schools. Furthermore, schools with higher student enrollment are more likely to have frequent occurrences of bullying, as are schools with a higher percentage of disadvantaged or male students. Schools in communities with either moderate or high levels of crime are more likely to have frequent occurrences of bullying among students than those in communities with low levels of crime. Bullying is also more likely to occur in schools located in towns and rural locales (vs. the suburbs) and in the Midwest and West regions (vs. the South).

Model 2 includes variables measuring surveillance, SROs, and disciplinary actions taken. We found positive associations between both surveillance measures and disciplinary actions and school bullying. However, we did not find any association between the number of SROs in the schools and bullying. Each additional surveillance measure added was associated with an 11 percent increase in the odds of having more frequent bullying in the school. Likewise, each additional disciplinary action taken increased the odds of having more frequent bullying among students by just under 1 percent. Adding these measures to the model rendered differences by school-level not significant. Otherwise, these variables appeared to have a little impact on the relationships identified in Model 1.

To address the question of the moderators, this study turns to results from interaction terms included in Model 3. We find no evidence of moderation in the effect of surveillance measures as both main and interaction terms. Main and interaction coefficients for the effect of SROs are likewise not significant. This study, however, finds significant variation in the association between disciplinary actions taken and bullying among students across school racial/ethnic contexts. The main effect reveals a positive association between disciplinary actions and bullying in majority-White schools (b = .022, p < .001). The non-significant interaction term suggests that the positive association between disciplinary actions and bullying was likely also present in majority-Latina/o/x schools whereas negative interactions for majority-Black schools (b =  − .016, p < .01) and diverse schools (b =  − .018, p < .01) largely cancel out the main effect of disciplinary actions. As such, it was unlikely to find a significant relationship between disciplinary actions and bullying in either majority-Black or diverse schools.

Table 3 presents marginal effects for the key independent variables within majority-White, majority-Black, majority-Latina/o/x, and diverse schools (see Appendix Table 4 for full models). Like Table 2, we find no significant marginal effects for surveillance measures across any of the four racial/ethnic school contexts in Model 3. This was not the case for SROs. Prior models indicate no relationship between SROs and bullying, and no significant differences in this relationship across school types. However, within-group results reveal a negative association between the number of SROs and bullying in both majority-Black and diverse schools. In majority-Black schools, each additional SRO reduces the likelihood of bullying occurring weekly or monthly by 3 percent each and increases the likelihood of bullying occurring occasionally by around 7 percent. The relationship in diverse schools is much weaker, as each additional SRO reduces the likelihood of bullying occurring daily by .2 percent, weekly by .3 percent, and monthly by 5 percent, and increases the likelihood of bullying occurring only occasionally by .8 percent. Average marginal effects for disciplinary actions appear to mirror patterns from Model 3 in Table 2. In both majority-White and Latina/o/x schools, each additional disciplinary action taken increases the likelihood of bullying taking place monthly, weekly, or daily, and decreases the likelihood of bullying occurring never or occasionally. Disciplinary actions taken were not associated with the frequency of bullying in either majority-Black or diverse schools.

OPEN IN VIEWER

Table 4.

Ordinal Logistic Regression Models Predicting Cyberbullying Among Students.

	Model 1		Model 2		Model 3
	B	SE	B	SE	B	SE
Majority Black	.272	(.293)	.125	(.291)	.578	(.851)
Majority Latina/o/x	−.476*	(.241)	−.458	(.241)	−.979	(.595)
Diverse	−.233	(.180)	−.234	(.180)	−1.087*	(.466)
Surveilance measures			.089*	(.037)	.036	(.048)
x Majority Black					−.016	(.131)
x Majority Latina/o/x					.139	(.102)
x Diverse					.223**	(.084)
SROs			−.011	(.016)	.005	(.020)
x Majority Black					−.233	(.127)
x Majority Latina/o/x					.009	(.038)
x Diverse					−.050*	(.024)
Disciplinary actions			.012***	(.003)	.027***	(.005)
x Majority Black					−.015*	(.006)
x Majority Latina/o/x					−.018*	(.007)
x Diverse					−.020*	(.008)
Elementary school	−1.379***	(.243)	−1.354***	(.242)	−1.349***	(.246)
Middle school	.717**	(.220)	.582**	(.221)	.572*	(.226)
High school	.397	(.229)	.262	(.230)	.256	(.236)
School enrollment	.001***	(.000)	.001***	(.000)	.001***	(.000)
Student-teacher ratio	.001	(.002)	.001	(.002)	.000	(.002)
Community crime	.487**	(.162)	.419*	(.162)	.415*	(.162)
% Disadvanatged	.002	(.003)	.000	(.003)	.000	(.003)
% Special education	.004	(.009)	.004	(.009)	.006	(.009)
% Male	.000	(.007)	.000	(.007)	.000	(.007)
% Underachievement	.000	(.004)	.000	(.004)	.001	(.004)
% Daily attendance	.002	(.009)	.003	(.009)	.004	(.009)
Urban	.142	(.154)	.097	(.155)	.079	(.155)
Town	.256	(.186)	.250	(.187)	.195	(.188)
Rural	.194	(.182)	.177	(.184)	.171	(.186)
Northeast	.210	(.195)	.216	(.190)	.265	(.195)
Midwest	.690***	(.160)	.644***	(.160)	.685***	(.163)
West	.455**	(.174)	.493**	(.177)	.534**	(.179)
Threshold 1	−1.016	(1.044)	−.710	(1.109)	−.823	(1.143)
Threshold 2	2.075	(1.043)	2.418	(1.108)	2.326	(1.142)
Threshold 3	3.272	(1.042)	3.639	(1.106)	3.558	(1.138)
Threshold 4	5.002	(1.055)	5.406	(1.120)	5.340	(1.152)

Data Source: School Survey on Crime and Safety, 2015–16, N = 2,060.

Note: Variable reference categories include majority White school, combined school, suburban locale, and South region. Models present weighted estimates.

*p < .05, **p < .01, ***p < .001 (two-tailed).

Segregation and Cyberbullying

Table 4 presents estimates from ordinal logistic regression models predicting the frequency of cyberbullying at school. This study began by focusing on the role of school racial/ethnic context in Model 1. We find that majority-Latina/o/x schools have lower odds of reporting more frequent cyberbullying than majority-White schools (b =  −.476, p < .05) net school controls. The coefficients for majority-Black schools and diverse schools were both smaller in magnitude and not significant, meaning no differences in the frequency of cyberbullying relative to majority-White schools. Other model covariates also emerge as significant predictors of cyberbullying in school. Cyberbullying was less likely to occur more often in elementary than in combined schools but more likely to occur more often in middle schools. Schools with higher student enrollment are more likely to have more frequent occurrences of cyberbullying, as are schools in communities with either moderate or high levels of crime (vs. a low level of crime). Cyberbullying was also more likely to occur with greater frequency in schools located in the Midwest and West compared to those located in the South.

Variables measuring the total number of surveillance measures, SROs, and disciplinary actions taken were added in Model 2. Here, we find positive associations between both surveillance measures and disciplinary actions and how often cyberbullying occurred in the school. However, we did not find any significant association between the number of SROs in the schools and cyberbullying. Specifically, each additional surveillance measure added was associated with a 9 percent increase in the odds of having more frequent cyberbullying in the school. Likewise, each additional disciplinary action taken increased the odds of having more frequent cyberbullying among students by just over 1 percent. After adding these measures to the model, it rendered the coefficient for majority-Latina/o/x schools as not significant. Otherwise, these variables appear to have a little impact on the relationships identified in Model 1.

To address the question of moderation, we turn to the results in Model 3. There is evidence of moderator by school racial/ethnic contexts across the three key independent variables. The main effect for surveillance measures and interaction effects for majority-Black and majority-Latina/o/x schools were all not significant. Rather, it appears that the effect of surveillance measures in Model 2 was entirely driven by the significant relationship between surveillance measures and cyberbullying in diverse schools (b = .223, p < .001). The main effect for SROs and interaction effects for majority-Black and majority-Latina/o/x schools were also not significant. However, a significant relationship between SROs and cyberbullying was present in diverse schools (b =  − .050, p < .05). In contrast to surveillance measures and SROs, the main effect for disciplinary actions reveal a positive association between cyberbullying in majority-White schools (b = .027, p < .001). Significant and negative interactions for majority-Black schools (b =  − .015, p < .05), majority-Latina/o/x (b =  − .018, p < .05), and diverse schools (b =  − .020, p < .05) reduce the main effect of disciplinary actions on cyberbullying.

Table 5 presents marginal effects for the key independent variables within majority-White, majority-Black, majority-Latina/o/x, and diverse schools (full models are available in Appendix Table 5). Average marginal effects reveal positive associations between surveillance measures and cyberbullying in diverse schools, as well as majority-Latina/o/x schools. Specifically, each additional surveillance measure decreases the likelihood of cyberbullying never occurring by around 4 percent and increases the likelihood of cyberbullying occurring either monthly or weekly by 1 to 2 percent, and daily by .1 to .3 percent. Within-group results also reveal a negative association between the number of SROs and cyberbullying in diverse schools. Each additional SRO reduces the likelihood of cyberbullying occurring daily, weekly, or monthly by .2 percent each and increases the likelihood of cyberbullying never occurring by .8 percent. In majority-White, majority-Black, and majority-Latina/o/x schools, each additional disciplinary action taken increases the likelihood of having cyberbullying occur monthly, weekly, or daily, and decreases the likelihood of having it never occur. However, changes in average marginal effects are larger in magnitude in majority-White schools than in majority-Black or majority-Latina/o/x schools. Disciplinary actions do not appear to predict the frequency of cyberbullying in diverse schools.

OPEN IN VIEWER

Table 5.

Marginal Effects Predicting Cyberbullying Among Students By School Racial Context.

	Frequency	Majority White	Majority Black	Majority Latina/o/x	Diverse
Surveilance measures	Never	−.003	−.002	−.037	−.042**
	Occassionally	.000	−.002	.003	.001
	Monthly	.001	.001	.017	.011
	Weekly	.001	.002	.013	.014*
	Daily	.001	.001	.003	.001*
SROs	Never	.000	.018	−.007	.008*
	Occassionally	.000	.012	.001	−.002
	Monthly	.000	−.009	.004	−.002*
	Weekly	.000	−.016	.003	−.002*
	Daily	.000	−.006	.001	−.002*
Disciplinary actions	Never	−.003**	−.001	−.002*	−.001
	Occassionally	.000	−.001	.000	.000
	Monthly	.001*	.001	.001*	.000
	Weekly	.001**	.001*	.001*	.000
	Daily	.001*	.000	.000	.000

Data Source: School Survey on Crime and Safety, 2015–16, N = 2,060.

Note: Average marginal effects are calculated from ordinal logistic regression analyses predicting likelihood of bullying among students. Models also control for school level, total enrollment, student-teacher ratio, community crime, % free/reduced lunch, % special education, % male, % underachivement, % daily attendance, locale, and region.

*p < .05, **p < .01, ***p < .001 (two-tailed).

Discussion

The current study aims to contribute to the literature on segregation, criminalization, and bullying by investigating the associations between policing, surveillance, punishment, and bullying in majority-Black, majority-Latina/o/x, and majority-White schools. Based on our results, we highlight three key findings. First, majority-Black schools have increased police presence and surveillance as well as punishment rates in comparison to majority-White schools as shown in previous research. Majority-Latina/o/x schools also have increased surveillance and punishment rates relative to majority-White schools. However, the criminalization, surveillance, and punishment in majority-Black schools is more significant. Second, policing, surveillance, and punishment within majority-Black and majority-Latina/o/x schools seem to have a limited role in the rate of bullying, which begs the question about the rationale for sustaining increased policing, surveillance, and punishment practices within these schools. Third, the descriptive findings highlight the educational inequities between majority-Black, majority-Latina/o/x, and majority-White schools. Scholars have shown that punitive discipline and control exist in racially/ethnically segregated minority schools in comparison to majority-White schools (Morris, 2016; Owens & McLanahan, 2020; Rios, 2011, 2017; Shedd, 2015). Because racially/ethnically segregated minority schools are generally situated in low-income areas that are characterized by crime and disorder, this gives the impression that these schools “need” more control and punishment than majority-White schools. Looking at police presence, surveillance, and punishment, our study finds that majority-Black and majority-Latina/o/x schools do in fact experience hyper-criminalization in U.S. schools in comparison to majority-White schools.

The findings from this study have several important theoretical implications. First, our study adds to a growing body of sociological literature on how segregation leads to significant racial and ethnic disparities in social outcomes by exploring criminalization and victimization in U.S. public schools (Reardon & Owens, 2014). Our findings support the argument made by numerous scholars that students in majority-Black and majority-Latina/o/x schools continue to be criminalized as they would have been treated in the past to achieve racial control in the U.S. It is clear that minority students are still perceived as disruptive and potentially criminal, as they have in the past, and therefore, administrators carry the impression that majority-Black and majority-Latina/o/x schools need more control and punishment than majority-White schools (Alexander, 2010). Second, our findings shed light on how the hyper-criminalization in majority-Black and majority-Latina/o/x schools reinforces racial/ethnic inequalities that, along with other factors such as a lack of resources, facilitates the school-to-prison pipeline. Finally, our study also extends the line of research on bullying. Despite increased crime control efforts and punishment in majority-Black and majority-Latina/o/x schools, we find that these school practices do not seem to impact rates of bullying. Research on school criminalization and bullying suggests that school administrators perceive this type of violence as a large part of the everyday lives of Black and Latina/o/x students and therefore, pay less attention and concern on addressing these issues in comparison to White students (Bradshaw et al., 2013; Peguero & Hong, 2020: Peguero & Williams, 2013).

Findings from this study also highlight major implications for school discipline and security measures. First, policymakers should consider addressing the racial/ethnic gap in the application of school crime control and punishment. While all U.S. public schools have experienced a rise in crime control and punishment, our study indicates that majority-minority school systems have a greater police presence, surveillance, and punishment than majority-White schools. This differential treatment is found to be counterproductive to school safety outcomes and a disruption in the educational progress and success of minority students (Bracy, 2011; Rios, 2011, 2017; Shedd, 2015). Second, crime control and punitive school measures do not impact bullying victimization in majority-minority schools. On the contrary, these crime control measures place minority students on the path of the school-to-prison pipeline. To prevent this from happening, school administrators should focus on developing school safety policies and practices that meet the needs of the students, parents, and the local community. Finally, schools should reconsider whether to adopt policies and programs such as zero-tolerance policies and SRO programs. These policies and programs are found to be ineffective, problematic, and discriminatory (Bracy, 2011). School administrators of majority-minority schools should provide additional resources and support to teachers to create a healthy and safe environment, which can improve academic outcomes (Koth et al., 2008).

This study is not without limitations. First, the study excludes predominately indigenous schools in the analysis. Criminalization as a tool of race-making also applies to Indigenous students with a force that mirrors that of Black and Latina/o/x students. Indigenous students make up less than 1 percent of the total enrollment, but 3 percent of those who are expelled from U.S. public schools (U.S. Department of Education, 2014). Indigenous students are referred to law enforcement or have school-related arrests at two to three times their enrollment rate (U.S. Department of Education, 2014). Although research on Indigenous youth and bullying is sparse, Campbell and Smalling (2013) found that Indigenous youth had the highest dropout rates and experienced greater levels of verbal and physical violence than all other racial/ethnic groups according to data from Minnesota school surveys. Nationally, the pushout rate for American Indian and Alaska Natives is almost 2.5 times what it is for their White peers (U.S. Department of Education, 2014). Due to the segregation of schools on reservations, predominantly Indigenous schools continue to act as a source of educational inequality that contributes to the marginalization and criminalization of Indigenous students (Chin et al., 2019; Hunt et al., 2020).

Second, it is important to examine the impact of criminalization in majority-minority schools and its impact on segregation, bullying, and inequality over time. The data in our study are cross-sectional and therefore, we cannot discern whether the composition of schools is related causally to increased police presence, surveillance, and/or punishment in schools. Although previous research on punishment trends supports the argument by racialized crime scholars that increased minority presence explains police strength, future studies should explore whether these patterns exist in majority-Black and majority-Latina/o/x schools (Kent & Jacobs, 2004). As urban, suburban, and rural communities grow more diverse, future research should also examine how the temporal relationship in schools varies across these different places.

Third, this study focuses on surveillance, criminalization, and bullying in racially/ethnically segregated schools, but does not include immigration status. Previous research demonstrates that immigration matters significantly in school violence and safety concerns. A variety of factors, such as state and federal policy, legal status, nationality, religion, gender, and English language proficiency can contribute to the marginalization of children of immigrants in U.S. schools (Bondy, 2015; Bondy, 2016; Gonzales, 2016; Peguero, 2009; Peguero & Bondy, 2011, 2015, 2020, 2021; Peguero & Hong, 2020). Further, public and political discourse on the “immigrant criminal” myth has seeped into schools, creating startling levels of fear for the children of immigrants and intensifying racial and ethnic tensions in classrooms (Costello, 2016a, 2016b; Rogers et al., 2017). The children of immigrants are the fastest-growing student population in U.S. public schools. Currently, 25 percent of all youth in U.S. public schools have at least one immigrant parent and this is expected to increase to 33 percent by 2040 (U.S. Census Bureau, 2019). With the rapid growth of immigrant students in the U.S. educational system, any effort to address school violence and implementation of school safety policies must consider barriers associated with immigrant students’ schooling and educational experiences.

Finally, data shortcomings foreclose the role of intersectionality and community environments. Regarding criminalization, segregation, bullying, and racial/ethnic inequality in schools, for instance, research demonstrates that Black and Latina/o girls and boys are disproportionately impacted by criminalization, zero-tolerance policies, and over-policing in schools in relation to their White peers (Bettie, 2014; Crenshaw et al., 2015; Kupchik, 2010, 2016; Morris, 2016; Rios, 2011, 2017). Additionally, Black and Latina girls are more likely to be sanctioned for disciplinary infractions, such as dress code violations, be labeled “loud” or “defiant,” and be suspended or expelled from school than their White peers (Bettie, 2014; Crenshaw et al., 2015; Morris, 2016). While Black and Latina girls outperform their male counterparts academically, disproportionate punishment rates still place them at an educational and social disadvantage, potentially slowing their educational progress. It is imperative to address gender and its intersections with race/ethnicity in future research on school safety and in the development of school safety measures. Research has also demonstrated that the lack of collective efficacy in communities makes individuals more vulnerable to cyberbullying (Hsieh et al., 2021). Future research should consider examining the role of other community-level characteristics such as collective efficacy to determine whether it acts as a protective factor against school bullying and cyberbullying for racial and ethnic minority youth.

In summary, our study makes an important contribution to the research on school segregation, securitization, bullying, and safety and suggests future lines of research about school safety, segregation, and inequality. One such line is to pursue a more comprehensive understanding of the relationship between segregation, school violence or safety, and racial/ethnic inequalities that accounts for historic and contemporary relationships between U.S. schools, society, politics, culture, and race/ethnic relations at the local, state, and national levels. Many scholars have suggested that the veil of pursuing safety with zero-tolerance social control, policing, security, and punishment policies and practices were fueled by mass shootings and public fear of panic over school bullying (Muschert et al., 2013; Muschert & Peguero, 2010; Schildkraut & Muschert, 2019). In general, our findings highlight the significance of developing school safety measures or amending the existing ones for racially/ethnically segregated schools in the U.S. public school system. Many school-based violence prevention programs are found to be less effective in racially/ethnically diverse school settings (Evans et al., 2014). Students who attend majority-Black and majority-Latina/o/x schools are confronted with numerous challenges that include criminalization, over-policing, hyper-surveillance, and stringent and disproportionate punishment, which are known to impede educational progress, success, and attainment as well as facilitate the school-to-prison pipeline and arguably, mass incarceration (Alexander, 2010; Irwin et al., 2013; Rios, 2011, 2017; Shedd, 2015). By developing and implementing effective and culturally relevant school safety programs and measures, U.S. public school systems can take the first necessary step towards enhancing healthy learning environments and promoting a more equitable school climate for majority-Black and majority-Latina/o/x schools, which are marginalized, under-resourced, and underserved.