School Security and Student Misbehavior

Abstract

Despite a nationwide trend to increase security measures in schools, their effectiveness in reducing or preventing student misbehavior remains largely unexamined. In addition, there is concern that increased security may have unintended negative side effects and is applied inequitably across students of disparate racial/ethnic backgrounds. The purpose of this study was to explore student differences between high- and low-security schools and to understand the relationship of security to student misbehavior. Data from 10,577 Grade 10 students from 504 public schools from the Education Longitudinal Study were examined. Numerous differences in students served by high- and low-security schools were noted; high-security schools were more likely to serve African American students. Security was negatively associated with student self-reported misbehavior but was unrelated to teacher ratings. Security interacted with race/ethnicity such that African American students were rated as having higher levels of disruptive and attendance-related misbehavior by teachers in schools with higher levels of security.

Keywords

education delinquency race/ethnicity African American school-based programs violence prevention program

Since 1999, many schools have significantly increased the use of security measures to reduce or prevent school violence and other forms of student misbehavior (Addington, 2009). Although the use of metal detectors and school security guards was relatively common since the 1980s in urban school districts in Los Angeles and Chicago, the 2000s saw a dramatic upsurge in the use of such measures in schools across the country, regardless of urbanicity. Increases in the use of security guards or police in schools (54% of schools in 1999 to 70% in 2010), locked school doors (39% in 1999 to 92% in 2010), and security cameras (19% in 1999 to 61% in 2010) represent the most striking increases during this time period (Robers, Kemp, & Truman, 2013). Despite its increased use, studies addressing the relationship of security to student misbehavior are sparse, tend to have focused on individual security measures rather than the overall security environment of the school, and have largely ignored the multi-level nature of the question of how school security policies might affect student behavior. This study seeks to address these issues.

Many speculate that the trend to increase school security measures was a response to the shootings that took place at Columbine High School in April of 1999 (Addington, 2009). Others suggest that this trend is a manifestation of a more generic fear of crime and violence that pervades American culture (Kupchik, 2010; Larsen, 2008). Regardless of the cause, there is widespread recognition that the effectiveness of security in reducing or preventing student misbehavior remains largely unexamined (Birkland & Lawrence, 2009; Greene, 2005). There is also concern that increased security may have the unintended side effect of increasing misbehavior by increasing student resentment due to invasions of privacy, excessive punitiveness, and betrayals of trust (Hyman & Perone, 1998).

Further concern exists as exposure to intense security environments may vary across students of disparate racial/ethnic backgrounds. In particular, evidence exists that African American students are more likely to encounter high-security school environments than their White peers (Toldson, 2011). Potentially ineffective or harmful school practices that are applied inequitably across racial/ethnic groups warrant special consideration.

The purpose of this study was to address two research questions regarding school security, student misbehavior, and student race/ethnicity: (1) What are the characteristics of students served in high-security schools relative to those served in low-security schools? (2) What is the relationship between school security and student misbehavior? How does the relationship between school security and misbehavior vary based on school characteristics and the demographic and academic characteristics of the students, with special consideration given to student race/ethnicity? A small body of theoretical and empirical work touches upon these issues.

Characteristics of High-Security Schools

Despite nationwide increases in the use of security, relatively little research has been conducted to document the characteristics of high- versus low-security schools. Studies have found that school size and location are important determinants of security. High-security schools tend to have more students, are located in urban areas and neighborhoods characterized by high/moderate crime, and are more frequent in the Southern United States (Kupchik, 2010; Nickerson & Spears, 2007; Servoss & Finn, 2014). In addition, Kupchik (2010) and Nickerson and Spears (2007) found that the proportion of students receiving free lunch, a measure of school socio-economic status (SES), was positively associated with the likelihood of certain security measures (e.g., metal detectors, locked gates). In contrast, Servoss and Finn (2014) found that having a high proportion of African American students was the strongest predictor of overall levels of security at a school; SES was not significant when the racial/ethnic composition of the school was accounted for.

Although these studies provide useful information on the characteristics of schools that tend to use specific security measures, they provide little, and at times conflicting, information on the characteristics of the students in these schools.

School Security and Student Misbehavior

When surveyed shortly after the Columbine shootings, 55% of American parents indicated that they feared for their child’s safety at school (Carroll, 2007). Noguera (1995, 2003) contended that from the early days of compulsory schooling in the United States, one of the primary functions of the public school was to control student behavior. Especially when instances of severe violence occur, the school needs to reassert their power and provide reassurance to the public that, as authority symbols, they maintain control of students. Visible displays of school security such as metal detectors and uniformed security officers serve the important symbolic function of reassuring the public that the school is a safe place for students.

However, as assertions of school control and power, high-security school environments require passivity and compliance on the part of students (Noguera, 1995, 2003). This sets the stage for conflict as most students today, particularly urban youth, are neither passive nor blindly compliant. If students feel powerless, caged, and stifled by their school environments, they may become frustrated and lose incentive to adhere to school norms for acceptable behavior.

Moreover, a high degree of dependence on police for disciplinary issues may inadvertently undermine the authority of school personnel to deal effectively with such issues and, in turn, lead to increases in misbehavior when police are not present on school grounds. Invasive procedures such as strip searches can lead to serious emotional damage and oppositional behavior for students. Punitive school consequences such as zero-tolerance policies and corporal punishment tend to engender desires for revenge, feelings of anger, and loss of faith in formerly trusted school personnel (Hyman & Perone, 1998).

Empirical investigations on the relationship between security and student misbehavior have tended to focus on the effectiveness of individual security strategies such as school resource officers (SROs) or metal detectors in relation to some measure of student violence or misbehavior. Impressions of SROs have tended to be favorable for both principals (May, Fessel, & Means, 2004) and students (McDevitt & Paniello, 2005). However, investigations into the efficacy of metal detectors to reduce or prevent any form of misbehavior have not been promising (Gastic, 2011; Hankin, Hertz, & Simon, 2011). In examining the relationships among seven individual security measures and seven forms of student misbehavior, Kupchik (2010) found only three significant relationships. Two of the three were positive, suggesting that having the security measure was associated with higher levels of student misbehavior. Taken together, these studies suggest that the connection among individual security measures and student misbehavior is mixed.

Although efficacy studies of individual security measures provide specific information about particular security techniques, they assume that students respond to the security environment of their schools one security measure at a time, an assumption that seems untenable when more closely examined. Consider the following situation: Students at a given high school are required to present their student identification to an armed SRO as they pass through a metal detector on their way into school every day through a doorway monitored by a security camera. Does it make sense that the individual security measures independently affect students’ emotional experience and behavior as they enter school each day? How does one isolate the effect of the metal detector, for example, in this situation? Given the infeasibility of this piecemeal approach, an important assumption underlying the conceptualization of school security in the current study is that it is students’ experience with the entire school security environment that influences their behavior. This assumption is reflected in the way that school security was quantified for this study.

Few studies have documented the effect of the school security environment, as a whole, on student misbehavior. The studies that do exist do not reflect positively on high levels of security both in terms of its palatability to students and its effectiveness in reducing (rather than leading to) student misbehavior. Security measures can create a school environment that overwhelmingly focuses on rules and their enforcement at the expense of teaching and learning, and students (particularly students of color) may not view security procedures as relevant to their safety but rather as another means by which school personnel assert (or abuse) their power and control over the students (Bracy, 2011). In intense security environments, minor conflicts over rule infractions may escalate, leading to serious conflicts between students and school personnel, often ending with a student being escorted from the school in handcuffs, feeling treated unfairly (Bracy, 2011; Kupchik, 2010).

Other investigations suggest that the overall security level of schools is positively related to school disruption and crime (Nickerson & Martens, 2008) and several indicators of school disorder, including the presence of gangs, drugs, attacks, and thefts (Mayer & Leone, 1999). In addition, Mayer and Leone (1999) found that school disorder was, in turn, related positively to attitudinal and behavioral indicators of student fear, concluding that “creating an unwelcoming, almost jail-like, heavily scrutinized environment may foster the violence and disorder school administrators hope to avoid” (p. 349).

These studies suggest limited effectiveness of security measures in reducing violence and other forms of student misbehavior. On the contrary, the few studies of this relationship suggest that security and misbehavior are positively related. More research is needed to document the relationship of the overall security environment to student misbehavior. This study addresses the relationship between the overall security environment and multiple measures of student misbehavior, both student- and teacher-reported.

Approach and Research Questions

This study focused on the link between aspects of the school security environment and student behavior. Determining the direction of influence between security and misbehavior with cross-sectional data such as that used in this study is difficult. In line with social cognitive theory’s principle of reciprocal determinism (Bandura, 1978, 1986), the direction of influence between these constructs—environment and behavior—is likely reciprocal. That is, student misbehavior may cause rules to tighten and security measures to be imposed, but inflexible environments and the perceptions that accompany them can also promote misbehavior. Much of the research reviewed above suggests that the latter explanation is more likely than the former (in particular, see Kupchik, 2010). In addition, because school security is a malleable aspect of the environment that is directly controlled through policy decisions by administrators, the perspective taken in this study, although not the only one possible, is that security affects student misbehavior.

Quantitative studies addressing security and misbehavior tend to have been conducted either at the school level or the student level, while an investigation of school policy (i.e., security) on student outcomes (i.e., misbehavior), by its very nature, ought to include both school and student-level information and a statistical approach that accounts for the nesting of students within schools. Moreover, most research in this area has examined the relationships between individual security measures (i.e., metal detectors, police) and misbehavior rather than considering the overall security environment. This study represents one of the first multi-level investigations that considered how overall school security related to various forms of student misbehavior.

The current study utilized multi-level data from a nationally representative sample of 10th grade students to address the following specific research questions: (1) What are the characteristics of students served in high-security schools relative to those served in low-security schools? (2) How does school security relate to student misbehavior? Student self-reported misbehavior as well as two forms of teacher-rated misbehavior (i.e., disruptiveness, and attendance problems) were examined. Do the relationships between security and misbehavior vary based on student race/ethnicity?

The results of the study have the potential to inform policy decisions on the use of security measures in schools, further explicate the school security construct, and highlight how high- and low-security schools differ.

Method

Data Source

This study utilized data from the Education Longitudinal Study of 2002 (ELS: 2002; Ingels, Pratt, Rogers, Siegel, & Stutts, 2004), a multi-year, multi-level undertaking by the National Center for Education Statistics (NCES). The ELS: 2002 project was launched in 2002 with a nationally representative survey of tenth grade students, their parents, teachers, and the principal at their school. During the base year assessment, information was collected on student achievement, career aspirations, attitudes, and school experiences. In addition, teachers provided information regarding student academic performance and classroom behavior, parents shared their perceptions on many aspects of family life and their impressions of the child’s school, and the school principal provided demographic information on the school, as well as information on the school facilities and policies. Further waves of data collection occurred in 2004, when most students were high school seniors, and 2006, 2 years after high school for most students, with a final assessment in 2012.

Sampling for ELS: 2002 was conducted in two stages. Schools were sampled first, followed by students within schools. Schools were selected based on a probability proportional to size methodology. At this stage, 1,221 eligible schools were identified out of the population of nearly 27,000 schools in the United States with tenth grade students. Of the list of eligible schools, 752 chose to participate. Administrator questionnaires were completed at nearly 99% of the participating schools. Tenth grade enrollment lists were provided by each participating school and 26 students were targeted for selection from each school. The participation rate for students was approximately 87%. Hispanic and Asian students were oversampled to allow for precision in statistical analyses involving these groups of students. Both school and student weights were derived to account for differential sampling probabilities and to correct for non-response bias so that the weighted sample was representative of the national population of 10th grade students. Both weights were used in all analyses and were normalized to avoid sample size inflation. For additional information on the ELS: 2002 Survey methodology, refer to Ingels et al. (2004).

Data from 10,557 Grade 10 students from 504 public schools from the base-year assessment were used in this study. Characteristics of the schools and their students are displayed in Table 1. This group of study participants is representative of the U.S. public school 10th grade student population in 2002. Student-level data came from the student and teacher questionnaires whereas the school-level data came from the school administrator questionnaire.

Table 1.

School and Student Characteristics.

Characteristic	N (%)	Weighted N (%)
School characteristic
Urbanicity
Urban	130 (25.8)	2,340 (14.1)
Suburban	256 (50.8)	6,893 (41.5)
Rural	118 (23.4)	7,395 (44.5)
Geographic region
Northeast	82 (16.3)	2,414 (14.5)
Midwest	124 (24.6)	4,799 (28.9)
South	194 (38.5)	6,037 (36.3)
West	104 (20.6)	3,379 (20.3)
Percent students on free lunch
0-10	125 (27.6)	3,053 (20.6)
11-30	163 (35.9)	5,780 (39.1)
31-50	84 (18.5)	3,321 (22.4)
>50	82 (18.0)	2,649 (18.0)
Grade 10 student enrollment
1-199	140 (27.8)	11,249 (67.7)
200-399	170 (33.8)	3,318 (20.0)
400-549	103 (20.4)	1,238 (7.4)
>550	91 (18.0)	823 (4.9)
Student characteristic
Gender
Male	5,266 (49.9)	1,401,005 (50.5)
Female	5,291 (50.1)	1,370,762 (49.5)
Race/ethnicity
White	5,795 (55.4)	1,716,386 (61.9)
Black/African American	1,413 (13.5)	374,626 (13.5)
Hispanic/Latino	2,064 (29.8)	543,239 (19.6)
Other	1,185 (11.4)	137,517 (5.0)
Family composition
Mother and father	5,961 (56.7)	1,562,094 (56.4)
Mother only	1,927 (18.3)	514,628 (18.6)
Father only	327 (3.1)	88,986 (3.2)
Mother and male guardian	1,451 (13.8)	391,346 (14.1)
Other	891 (8.1)	214,703 (7.7)
Parents’ level of education
Did not graduate high school	769 (7.3)	138,669 (6.6)
Graduated high school	2,397 (22.8)	602,068 (21.7)
Graduated from college	3,355 (11.0)	906,396 (32.7)
Completed graduate degree	1,526 (14.6)	406,104 (14.6)
Native English speaker
Yes	8,604 (82.3)	2,398,149 (86.5)
No	1,853 (17.7)	373,618 (13.5)

Primary Measures

School security

School principals were asked to respond to a series of 21 yes/no questions with regard to whether certain security practices are used in their school (see Table 2). An overall security score was obtained by means of Rasch scaling (Rasch, 1960). Item “infit” and “outfit” mean squares and z scores indicated good fit relative to standard criteria (Bond & Fox, 2007) for all but one item (telephones in classrooms). This item was eliminated from the scale. In addition, the standard deviations of the standardized infit statistics were less than 2, indicating good overall model fit (Bode & Wright, 1999). The overall security score was unidimensional, accounting for 73.1% of the total item variability, with the largest residual component accounting for less than 3% of the total variance and no interpretable pattern of item loadings. These values indicated that the single security score was a good summary of schools’ overall security environments. Higher scores on this scale indicate that the school has higher degrees of security in general.

Table 2.

Comparing Individual Security Measures Between High- and Low-Security Schools.

Item	Overall %	Low-security %	High-security %	OR
During this school year, is it a practice of your school to do the following . . .
Control access to school buildings during school hours	73.4	64.8	92.3	6.56
Control access to school grounds during school hours	34.0	21.8	75.1	10.79
Require students to pass through metal detectors each day	2.0	1.2	9.1	8.41
Perform one or more random metal detector checks on students	8.3	0.9	38.8	70.54
Close the campus for most students during lunch	63.5	52.1	85.9	5.59
Use one or more random dog sniffs to check for drugs	52.2	36.7	79.9	6.85
Perform one or more random sweeps for contraband	28.0	18.0	56.3	5.88
Require drug testing for any students	14.2	8.8	29.1	4.27
Require students to wear uniforms	5.6	0.4	18.4	61.60
Enforce a strict dress code	48.9	35.5	83.5	9.21
Require clear book bags or ban book bags on school grounds	9.7	4.7	31.5	9.28
Require students to wear badges or picture IDs	10.6	1.0	41.0	71.42
Require faculty and staff to wear badges or picture IDs	26.5	11.5	70.8	18.58
Use one or more security cameras to monitor the school	35.4	18.6	75.1	13.20
Provide an emergency alarm or call button in most classrooms	56.0	44.5	78.2	4.46
During this school year, did your school regularly use paid law enforcement or security services at school at the following times . . .
At any time during school hours or while students were arriving or leaving	52.5	34.5	93.0	25.37
At selected school activities (e.g., athletic and social events)	75.2	61.4	99.3	83.08
When school/school activities were not occurring	16.4	7.0	40.2	8.94
Other	6.9	1.5	16.5	13.16

Note. OR = odds ratio.

To display contrasts between high- and low-security schools for Research Question 1, schools were categorized as high or low based on an extreme groups method (see Preacher, Rucker, MacCallum, & Nicewander, 2005). The top third of schools based on their total security score were considered the high-security group and the bottom third the low-security group.

Misbehavior

This study focused on commonly occurring student misbehaviors such as fighting, disruptive behaviors, and attendance-related behaviors such as absenteeism, truancy, and tardiness. As most of the misbehavior scales required forming composites from items on different response scales, the partial credit Rasch model was used to create the misbehavior composite scores. The following misbehavior scales were created, one based on student responses and the others based on teachers’ ratings of each student.

Student misbehavior–Self-report

This scale focused on fighting and attendance-related misbehaviors, as well as the punitive consequences of misbehavior as reported by the students themselves. This scale was comprised of student responses to how often “I got into a physical fight at school” (1 = never, 3 = more than twice) and how often “I was late for school,” “I cut or skipped classes,” “I was absent from school,” “I got in trouble for not following school rules,” “I was put on in-school suspension,” and “I was suspended or put on probation” (1 = never, 5 = 10 or more times). Scale reliability was α = .73.

Student misbehavior–Teacher-reported disruptiveness

This composite is comprised of students’ math and English teachers’ responses to five items regarding disruptiveness in the classroom. The following yes/no items, namely, “Have you communicated with the student’s parents about the student’s disruptive behavior in school?,” “Has the student fallen behind in schoolwork due to a disciplinary action?,” and “Have you spoken to a guidance counselor about the student’s disruptive behavior in school?” and the Likert-type scale items (1 = never, 5 = all the time) “How often is the student attentive in your class?” and “How often is this student disruptive in your class?” are included in this scale. Scale reliability was α = .79

Student misbehavior–Teacher-reported attendance

This composite is comprised of teachers’ responses to three items regarding student tardiness and absenteeism. The yes/no item, namely, “Have you communicated with the student’s parents about the student’s absenteeism?” and the Likert-type scale items (1 = never, 5 = all the time) “How often is the student absent from your class?” and “How often is this student tardy to your class?” are included in this scale. Scale reliability (α) was .73.

The three misbehavior measures were scaled separately. Investigation of fit statistics and a factor analyses were conducted to assure adequate item fit, local independence, and unidimensionality. All infit and outfit mean square statistics were within the acceptable range. Furthermore, factor analyses of the residuals indicated that 89.1% of the variance in the misbehavior self-report items, 75.4% of the variance in the disruptiveness items, and 80.7% of the variance in the attendance-related items were accounted for by a single construct for each measure. No significant secondary factors were present.

Other Measures

The following school and student measures were used as covariates in analyses based on their conceptual relevance to misbehavior outcomes and the school security measure.

School characteristics

School characteristics came from variables on the ELS Administrator Survey. Urbanicity was categorized as urban, suburban, or rural. The level of neighborhood crime was rated by the administrator as high, low, moderate, or mixed. School size was indicated by the school’s 10th grade enrollment. The percent of students at the school receiving free lunch was used as a measure of school socio-economic status.

Student characteristics

Additional student characteristics came from variables on the ELS Student Survey. Student gender was categorized as either male or female. Race/ethnicity was characterized as non-Hispanic Black/African American, non-Hispanic White, Hispanic, and Other (includes American-Indian, Asian/Pacific-Islander, and Multi-Racial). Student achievement was measured by NCES math and reading achievement tests.

Student academic risk, a composite constructed by NCES, was characterized based on a number of background characteristics including coming from a single-parent household, having two parents without a high school diploma, having a sibling who has dropped out of school, having changed schools 2 or more times, having repeated at least one grade, and coming from a household with an income below the federal poverty level.

Measures of sports and extracurricular activity participation were utilized as well as a five-item measure of students’ perceptions on the quality of student–teacher relationships (α= .73), a three-item measure of students’ extrinsic motivation (α = .85), a four-item measure of students’ success expectations (α = .84), and a five-item measure of students’ school-based effort and persistence (α = .89).

Analysis

Analyses were conducted in two phases. In the first phase, descriptive statistics were computed for all variables, composite scores for the security and misbehavior measures were developed using Rasch analysis, and correlations among the main variables of interest were calculated.

The second phase of analyses consisted of procedures aimed specifically at answering the research questions.

Research Question 1.

Schools were characterized as either high or low security based on an extreme groups methodology (top third vs. lowest third) on the total security score. Multi-level modeling using the HLM 6 program (Raudenbush, Bryk, & Congdon, 2004) was used to compare high- and low-security schools on the following student characteristics: race, family composition, level of parental education, socio-economic status, academic risk, and math and reading achievement scores. Each model was specified such that the student-level characteristic was the dependent variable and the school security variable was entered as a school-level predictor. Student-level intercepts were specified as random effects. Depending on the nature of the student characteristic variable (i.e., dichotomous, polytomous, ordinal, continuous), logistic, multinomial, ordinal, or linear modeling procedures were used, respectively.

Research Questions 2.

All three of the misbehavior measures were used as dependent variables in separate hierarchical linear models. The main school-level predictor of interest was the total security score for each school. Student-level predictors that were considered include student gender, race, the academic risk composite, math and reading achievement, expectations for success, extrinsic motivation, school effort, participation in sports and extracurricular activities, and perceptions of student–teacher relationships. Other school-level predictors that were considered included school size, neighborhood crime, urbanicity, and socio-economic status.

General approach to HLM analyses

The following general analytic sequence was used to address Research Question 2. (1) A null (intercept-only) model was estimated for each dependent variable to investigate the proportion of variance attributable to within-school or between-school sources. (2) A model with school security as a school-level predictor was estimated to explicitly address the relationships between school security and the misbehavior variables. (3) A student-level model was developed, excluding any student-level variables that were not statistically significant or conceptually necessary to control. (4) Additional school-level variables were entered into the model for the student-level intercept. School-level predictors that were not significant were excluded from further analyses of that dependent variable. (5) Cross-level interactions between school security and the student-race/ethnicity were considered.

Results

A list of the school security items and how commonly they occur in the sample appears in Table 2. A comparison of the proportion of schools from the top third (high security) relative to those in the bottom third (low security) for each security measure also appears in Table 2. All of the comparisons are statistically significant (p < .001) and the magnitudes of the differences are expressed as odds ratios. High-security schools showed a greater propensity for each of the security measures relative to low-security schools supporting the notion that a high-security environment is not defined by any single security measure.

Characteristics of Students Served in High-Security Schools

A series of linear models revealed that students in high-security schools have significantly higher degrees of academic risk (b = 0.265, p = .003) and significantly lower math (b = −2.41, p = .002) and reading (b = −2.52, p = .002) achievement scores relative to students in low-security schools. Students in high- and low-security schools did not differ in socio-economic status (p = .591), although the trend was in the direction of high security related to low SES.

Based on findings from a multinomial model, students in high security schools are 11.78 times more likely to be African American than White (p < .001), and 1.56 times more likely to be Hispanic/Latino than White (p < .001) relative to those students from low-security schools. A series of logistic models revealed that students from high-security schools are 1.67 times more likely to come from a home where both parents failed to complete high school (p = .012), and significantly less likely (odds ratio [OR] = 0.77, p = .011) to come from a family composed of both biological parents than students in low-security schools.

High-security schools tend to serve student populations that are otherwise at risk of negative schools outcomes (e.g., low achievement, African American, single-parent family, low levels of parental education).

School Security and Student Misbehavior

Student misbehavior self-report

For students’ self-reported misbehavior, 95.1% of the variance was attributable to within-school (student) sources whereas 4.9% was attributable to between-schools sources. See Table 3 for the final model for this and the other misbehavior outcomes. There was no significant bivariate relationship between school security and self-reported student misbehavior (b = −0.025, t(498) = −0.64, p = .519).

Table 3.

Predictors of Student Misbehavior.

	Self-report		Disruptiveness		Attendance
Parameters (fixed effects)	Coefficient	SE	Coefficient	SE	Coefficient	SE
Between-schools
Grand M	−1.236***	0.257	1.991***	0.402	−0.698*	0.022
School security	−0.134**	0.047	−0.083	0.055	−0.131	0.084
School size	0.071***	0.018	—	—	—	—
Within-schools
Female gender	−0.126^†	0.070	−0.791***	0.089	0.134	0.087
Race/ethnicity
Black–White	0.130	0.099	0.373**	0.140	0.320*	0.148
Hispanic–White	0.196^†	0.103	0.127	0.158	0.291*	0.144
Other–White	−0.174	0.202	−0.401^†	0.231	−0.350^†	0.212
Academic risk	0.189***	0.025	0.074^†	0.041	0.207***	0.039
Math achievement	−0.016**	0.006	−0.052***	0.008	−0.040***	0.007
Reading achievement	−0.008^†	0.005	−0.029***	0.006	−0.013*	0.006
Extracurriculars	−0.091**	0.026	−0.088**	0.030	−0.091**	0.031
Number of sports	—	—	0.076*	0.034	—	—
Student–teacher relations	−0.313***	0.033	−0.222***	0.037	−0.173***	0.045
Control expectations	0.134**	0.046	—	—	—	—
School effort	−0.299***	0.041	−0.286***	0.045	−0.137***	0.046
Interactions
Security × Black	—	—	0.252*	0.103	0.277*	0.141
Security × Hispanic	—	—	0.137^†	0.074	0.141	0.124
Security × Other	—	—	−0.034	0.183	0.111	0.116
Random effects	Variance	χ²(492)	Variance	χ²(496)	Variance	χ²(496)
Conditional error variance components:
Between-schools	0.059	625.93	0.446	1,324.64	0.704	1,730.42
Within-schools	1.874		2.766		2.982
Total	1.933		3.212		3.686
Percentage reduction in conditional error variance at school level^a	51.2		27.6		21.1
Percentage reduction in total conditional error variance^a	21.9		26.5		18.6

Percentage reductions in error variance relative to the null model.

†

p < .10. *p < .05. **p < .01. ***p < .001.

However, when adjusting for student gender, race/ethnicity, academic risk, math and reading achievement, extracurricular activity participation, student–teacher relationships, expectations for success, and school effort and school size, security was negatively related to self-reported misbehavior (b = −0.134, p = .047). This result suggests that, all other student and school characteristics being equal, students engage in less misbehavior in higher security schools.

A number of other student characteristics were negatively related to self-reported misbehavior. These include math achievement test scores (b = −0.016, p = .004), being of Other race relative to White (b = −0.374, p = .033), increased participation in extracurricular activities (b = −0.091, p = .001), more positive student–teacher relationships (b = −0.313, p < .001), and positive attitudes toward effort and persistence in school (b = −0.299, p < .001). Having a greater number of academic risk factors (b = 0.189, p < .001) and higher levels of control expectations (b = 0.134, p = .004) were positively related to self-reported misbehavior. There was no significant difference between African American and White students or between Hispanic/Latino students and White students in self-reported misbehavior.

At the school-level, controlling for the aforementioned student characteristics, students reported more misbehavior in larger schools (b = 0.141, p = .003). Levels of neighborhood crime, urbanicity, and percent free lunch were not significantly related to student self-reported misbehavior.

Teacher-reported disruptiveness

For teacher-reported disruptive misbehavior, 85.9% of the variance was attributable to within-school (student) sources with 14.1% attributable to between-schools sources. There was no significant bivariate relationship between school security and student disruptive misbehavior (b = −0.025, t(498) = −0.48, p = .634).

When student characteristics were introduced to the model, there was no significant main effect of security on disruptiveness. There was, however, a significant interaction between security and race. Without considering security, there was no significant difference between African American and White students. However, when school security was entered, teacher ratings of disruptive behavior for African American students were significantly greater compared to their White peers (b = 0.373, p = .008). Moreover, there was a significant interaction between race/ethnicity and school security such that African American students were rated as even more disruptive by their teachers relative to White students in schools with higher levels of security (b = 0.252, p = .015). Solving the simple slopes for the African American versus White regression coefficient in low- versus high-security schools gave a value of −0.069 in low-security schools versus 0.457 in high-security schools.

At the student-level, females (b = −0.779, p < .001), those students higher in math (b = −0.053, p < .001) and reading achievement (b = −0.029, p < .001), with more participation in extracurricular activities (b = −0.088, p = .003), more positive relationships with their teachers (b = −0.222, p < .001), and more positive attitudes toward effort in school (b = −0.286, p < .001) were rated as significantly less disruptive. Those who participated on a greater number of athletic teams were rated as more disruptive by teachers (b = 0.076, p = .043).

At the school level, school size, urbanicity, levels of neighborhood crime, and percent free lunch were not significantly related to disruptiveness.

Teacher-reported attendance-related misbehavior

For attendance-related misbehavior, 80.3% of the variance was attributable to within-school (student) sources and 19.7% to between-schools sources. There was no significant bivariate relationship between school security and student attendance-related misbehavior (b = −0.026, t(498) = −0.48, p = .737).

Similar to the results for teacher-rated disruptiveness, there were no racial/ethnic differences in attendance-related behavior until the school security variable was entered into the school-level model. When security was considered, teacher ratings of African American students’ attendance problems were significantly higher than their White peers (b = 0.320, p = .03) as were Hispanic/Latino students relative to White students (b = 0.291, p = .042). Moreover, there was a significant interaction between race/ethnicity and school security such that African American students were rated as having even more attendance-related misbehavior by their teachers in schools with higher levels of security (b = 0.277, p = .049). Solving the simple slopes for the African American versus White regression coefficient in low versus high-security schools (defined as above) gave a value of −0.166 in low-security schools versus 0.332 in high-security schools.

Background variables

At the student-level, students higher in math (b = −0.040, p < .001) and reading (b = −0.013, p = .026) achievement, with more participation in extracurricular activities (b = −0.091, p = .004), more positive relationships with their teachers (b = −0.173, p < .001), and more positive attitudes toward effort in school (b = −0.176, p < .001) were rated as having significantly fewer attendance-related misbehaviors. However, those high in academic risk factors had more attendance-related difficulties (b = 0.207, p < .001).

At the school level, school size, urbanicity, levels of neighborhood crime, and percent free lunch were not significantly related to attendance problems.

Discussion

Results of this study support the following conclusions: (1) Student misbehavior varies more as a result of individual student characteristics than school characteristics; (2) High-security schools tend to serve student populations that are otherwise at risk for negative school outcomes, including misbehavior; (3) The relationship between school security and student misbehavior depends on whether one asks teachers or students. Higher levels of security were related to lower levels of self-reported student misbehavior. Security was not related to teacher ratings of student misbehavior. (4) Racial/ethnic differences in teacher ratings of misbehavior depend on school security level. African American students display more disruptive and attendance-related misbehavior (or at least are rated as such by their teachers) relative to White students in high-security schools.

Analyses revealed that relatively little of the variability (5%-20%) in student misbehavior is attributable to differences in the school environment. Misbehavior seems to rely very strongly upon differences among the students themselves. This finding raises the question as to how useful school-wide security policies might hope to be in reducing student misbehavior relative to intervention with individual students. Unfortunately, school personnel are often unable to directly change a number of the student characteristics that are related to misbehavior and other negative school outcomes, and so they rely upon manipulating school policies to affect change in their students. School-level policies and programs do exist that improve student behavior (Gottfredson, Gottfredson, & Hybl, 1993; Sugai & Horner, 2002); beefing up school security does not seem to be one of them.

High security level was related to a number of student characteristics: low math and reading achievement, high academic risk, low levels of parent education, and African American race/ethnicity. This last finding is consistent with Toldson (2011), who found that African American students were about 6 times more likely to enter school through a metal detector relative to their White peers. It appears that high-security environments are largely reserved for those students who are otherwise already at the highest risk of school-related difficulties, and one questions the wisdom of using a punitive, prison-like approach as opposed to a more positive approach (Muscott et al., 2004) with these students.

School security was negatively related to self-reported student misbehavior. This effect was not corroborated by findings on either of the teacher-rated misbehavior measures, however. This inconsistency is not entirely surprising given research that documents the weak relationship between teacher ratings of problem behavior and student self-reports (see, for example, Achenbach, McConaughy, & Howell, 1987), including ratings of disruptive behavior (Henry, 2006) and threats of interpersonal violence (Liau, Flannery, & Quinn-Leering, 2004). However, it is possible that students actually do misbehave less in high-security environments, but teachers fail to recognize this improvement. As teachers tend to be the arbiters of the misbehavior/punishment system in schools, student-reported behavioral improvements that fail to be noticed by teachers are essentially non-improvements.

Perhaps the most important finding with regard to the relationship between school security and student misbehavior involves the race/ethnicity effects. It is only when school security was considered that race/ethnicity became a significant predictor of teacher ratings of misbehavior. When only considering student-level characteristics, there were no racial/ethnic differences in teacher-rated misbehaviors. However, when school security was added to the model, differences between African American and White students emerged. Moreover, there were significant interactions between security and race/ethnicity such that African American students were rated as more disruptive and as having more attendance-related misbehaviors in high-security schools relative to their White peers.

One potential explanation for this interaction is that teachers exhibit more bias in their perceptions of African American students’ behavior in higher security school contexts. African American students are often perceived by teachers to be more disruptive, confrontational, and less engaged in the classroom (Ainsworth-Darnell & Downey, 1998). These teacher biases are generally not attributed to explicit racism, but rather to misinterpretations of Black students’ cultural style (Allen & Boykin, 1992), differences in status (Alexander, Entwisle, & Thompson, 1987), or racial/ethnic mismatch between teacher and student (Downey & Pribesh, 2004). The aspects of higher security environments that might exacerbate teachers’ racially motivated appraisals of student behavior remain unclear but seem an important avenue for future research.

In addition, there are plausible reasons why African American students may have more behavior problems in high-security schools compared to schools lower in security. As suggested by Rios (2011) regarding the experiences of urban youth of color in school settings, “school personnel, police officers, and other adults in the community, had created an environment that made these young people feel criminalized from a young age” (p. 74). Resistance to this criminalization through acting out, perhaps particularly under the increased scrutiny of a high-security school environment, may be a form of coping, “an escape from their punitive reality” (Rios, 2011, p. 80). In addition, Bracy (2011) found that students perceived many of the security rules and procedures to be opportunities for teachers to abuse their power and to treat students in an unfair manner. Despite their awareness of the disciplinary consequences of involvement in conflict with teachers, students’—particularly African American males—desires to defend themselves from injustice, support their friends, maintain their self-respect, and fight for fairness often supersede their desire to avoid punishment (Sheets, 1996; Sheets & Gay, 1996). This dynamic could explain the interactions between school security and race/ethnicity found in the current study. However, further research is needed to better understand the associations among school security, student perceptions of fairness, and misbehavior.

The security by race interaction may also partially explain the discipline gap. A number of studies have indicated that African American students are about 2 to 5 times more likely to receive out-of-school suspensions (Eitle & Eitle, 2004; Losen & Skiba, 2010; Mendez & Knoff, 2003) than White students. If African American students are rated as more disruptive or truant in higher security environments by their teachers, and a lot of African American students attend schools with high security environments (62% of African American students attended schools in the highest third in terms of overall security in this study), increased suspensions for African American students may follow, particularly as most suspensions are the result of, or escalate from, relatively minor infractions like disrupting class, being truant, or perceived as disrespectful (Mendez & Knoff, 2011).

A number of additional findings emerged from the current study. Females were rated as less disruptive and as having fewer academic problems than their male counterparts. Low math and/or reading achievement scores were related to all types of misbehavior. This finding is concordant with a sizable body of research suggesting that students with low academic performance are approximately twice as likely to engage in school misbehavior relative to students with high academic performance (Maguin & Loeber, 1996).

Students who participate in a higher number of sports, extracurricular activities, and have positive attitudes toward school are also less likely to misbehave. These findings are consistent with research by Stewart (2003) showing that students who are more attached and committed to their schools tend to misbehave less.

Students’ ratings of the quality of the relationships they have with their teachers were also negatively related to misbehavior. This finding is consistent with work by Miller, Ferguson, and Byrne (2000), who examined the causal attributions students make regarding their own misbehavior. Students rated teacher unfairness, playing favorites, inconsistent discipline, and rudeness as the most important predictors. A focus on improving student–teacher relationships may prove to be a superior investment to the substantial costs, financial (see, for example, DeAngelis, Brent, & Ianni, 2011) and otherwise, of school security.

Limitations

Given the cross-sectional nature of the data, determining the direction of causality for the observed relationships was not possible. Steps were taken in the analyses to control for covariates that make the assumption on the direction of causality more plausible, however. Longitudinal research that allows the examination of changes in security relative to changes in misbehavior is warranted.

The measures of misbehavior available in the ELS datasets are perhaps not those that are most targeted by the implementation of school security, so the analyses may have been predisposed to finding null results. If more items on school-related violence had been included on the survey, perhaps the benefits of security could have been documented. As a counterpoint, the forms of misbehavior that were included are those that are commonly dealt with by teachers and administrators and represent important outcomes.

Footnotes

Acknowledgements

The author would like to express his gratitude to Jeremy Finn, Amanda Nickerson, and Michele Shanahan for their guidance in this project. Special thanks also go to Mikael Akerfeldt and Steven Wilson for their assistance during manuscript development.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biography

Timothy J. Servoss is an associate professor in the Department of Psychology at Canisius College in Buffalo, NY. His research focuses broadly on the effects of school policies on students, with particular focus on the unintended consequences of school security, the school-to-prison pipeline, and racial/ethnic disparities in exclusionary discipline.

References

Achenbach

T. M.

McConaughy

S. H.

Howell

C. T.

(1987). Child/adolescent behavior and emotional problems: Implications of cross-informant correlations for situational specificity. Psychological Bulletin, 101, 213-232.

Addington

L. A.

(2009). Cops and cameras: Public school security as a policy response to Columbine. American Behavioral Scientist, 52, 1426-1445.

Ainsworth-Darnell

J. W.

Downey

D. B.

(1998). Assessing the oppositional culture explanation for racial/ethnic differences in school performance. American Sociological Review, 63, 536-553.

Alexander

K. L.

Entwisle

D. R.

Thompson

M. S.

(1987). School performance, status relations, and the structure of sentiment: Bringing the teacher back in. American Sociological Review, 52, 665-682.

Allen

B. A.

Boykin

A. W.

(1992). African-American children and educational process: Alleviating cultural discontinuity through prescriptive pedagogy. School Psychology Review, 21, 586-596.

Bandura

(1978). The self system in reciprocal determinism. American Psychologist, 33, 344-358.

Bandura

(1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Birkland

T. A.

Lawrence

R. G.

(2009). Media framing and policy change after Columbine. American Behavioral Scientist, 52, 1405-1425.

Bode

R. K.

Wright

B. D.

(1999). Rasch measurement in higher education. In Smart

J. C.

Tierney

W. G.

(Eds.), Higher education: Handbook of theory and research (Vol. XIV., pp. 287-316). New York, NY: Agathon Press.

10.

Bond

T. G.

Fox

C. M.

(2007). Applying the Rasch model: Fundamental measurement in the human sciences (2nd ed.). New York, NY: Routledge.

11.

Bracy

N. L.

(2011). Student perceptions of high-security school environments. Youth & Society, 43, 365-395.

12.

Carroll

(2007, September). The divide between public school parents and private school parents. Gallup News Service. Available from http://www.gallup.com

13.

DeAngelis

K. J.

Brent

B. O.

Ianni

(2011). The hidden cost of school security. Journal of Education Finance, 36, 312-337.

14.

Downey

D. B.

Pribesh

(2004). When race matters: Teachers’ evaluations of students’ classroom behavior. Sociology of Education, 77, 267-282.

15.

Eitle

T. M.

Eitle

D. J.

(2004). Inequality, segregation, and the overrepresentation of African-American in school suspensions. Sociological Perspectives, 47, 269-287.

16.

Gastic

(2011). Metal detectors and feeling safe at school. Education and Urban Society, 43, 486-498.

17.

Gottfredson

D. C.

Gottfredson

G. D.

Hybl

L. G.

(1993). Managing adolescent behavior: A multiyear, multischool study. American Educational Research Journal, 30, 179-215.

18.

Greene

M. B.

(2005). Reducing violence and aggression in schools. Trauma, Violence, & Abuse, 6, 236-253.

19.

Hankin

Hertz

Simon

(2011). Impacts of metal detector use in schools: Insights from 15 years of research. Journal of School Health, 81, 100-106.

20.

Henry

D. B.

(2006). Associations between peer nominations, teacher ratings, self-reports, and observations of malicious and disruptive behavior. Assessment, 13, 241-252.

21.

Hyman

I. A.

Perone

D. C.

(1998). The other side of school violence: Educator policies and practices that may contribute to student misbehavior. Journal of School Psychology, 36, 7-27.

22.

Ingels

S. J.

Pratt

D. J.

Rogers

Siegel

P. H.

Stutts

(2004). ELS: 2002 base year data file user’s manual (NCES 2004-405). Washington, DC: U.S. Department of Education, National Center for Education Statistics.

23.

Kupchik

(2010). Homeroom security: School discipline in an age of fear. New York: New York University Press.

24.

Larsen

M. A.

(2008). North American insecurities, fears and anxieties: Educational implications. Comparative Education, 44, 265-278.

25.

Liau

A. K.

Flannery

D. J.

Quinn-Leering

(2004). A comparison of teacher-rated and self-reported threats of interpersonal violence. Journal of Early Adolescence, 24, 231-249.

26.

Losen

D. L.

Skiba

R. J.

(2010). Suspended education: Urban middle schools in crisis. Los Angeles, CA: The Civil Rights Project at UCLA. Retrieved from http://civilrightsproject.ucla.edu/research/k-12-education/school-discipline/suspended-education-urban-middle-schools-in-crisis/Suspended-Education_FINAL-2.pdf

27.

Maguin

Loeber

(1996). Academic performance and delinquency. In Tonry

(Ed.), Crime and justice: A review of research (Vol., 20, pp. 145-264). Chicago, IL: University of Chicago Press.

28.

May

D. C.

Fessel

S. D.

Means

(2004). Predictors of principals’ perceptions of school resource officer effectiveness in Kentucky. American Journal of Criminal Justice, 29, 75-93.

29.

Mayer

M. J.

Leone

P. E.

(1999). A structural analysis of school violence and disruption: Implications for creating safer schools. Education & Treatment of Children, 22, 333-356.

30.

McDevitt

Paniello

(2005). National assessment of the school resource office programs: Survey of students in three large new SRO programs (National Institute of Justice final report). Retrieved from http://www.eric.ed.gov/PDFS/ED486271.pdf

31.

Mendez

L. M.

Knoff

H. M.

(2003). Who gets suspended from school and why: A demographic analysis of schools and disciplinary infractions in a large school district. Education & Treatment of Children, 26(1), 30-51.

32.

Miller

Ferguson

Byrne

(2000). Pupils’ causal attributions for difficult classroom behavior. British Journal of Educational Psychology, 70, 85-96.

33.

Muscott

H. S.

Mann

Benjamin

T. B.

Gately

Bell

K. E.

Muscott

(2004). Positive behavioral interventions and supports in new Hampshire: Preliminary results of a statewide system for implementing schoolwide discipline practices. Education & Treatment of Children, 27, 453-475.

34.

Nickerson

A. B.

Martens

M. P.

(2008). School violence: Associations with control, security/enforcement, educational/therapeutic approaches, and demographic factors. School Psychology Review, 37, 228-243.

35.

Nickerson

A. B.

Spears

W. H.

(2007). Influences on authoritarian and educational/therapeutic approaches to school violence prevention. Journal of School Violence, 6(4), 3-31.

36.

Noguera

P. A.

(1995). Preventing and producing violence: A critical analysis of responses to school violence. Harvard Educational Review, 65, 189-212.

37.

Noguera

P. A.

(2003). Schools, prisons, and social implications of punishment: Rethinking disciplinary practices. Theory Into Practice, 42, 341-350.

38.

Preacher

K. J.

Rucker

D. D.

MacCallum

R. C.

Nicewander

W. A.

(2005). Use of the extreme groups approach: A critical reexamination and new recommendations. Psychological Methods, 10, 178-192.

39.

Rasch

(1960). Probabilistic models for some intelligence and attainment tests. Copenhagen: Danmarks Paedagogiske Institut.

40.

Raudenbush

S. W.

Bryk

A. S.

Congdon

(2004). HLM 6 for windows. Lincolnwood, IL: Scientific Software International.

41.

Rios

V. M.

(2011). Punished: Policing the lives of Black and Latino boys. New York: New York University Press.

42.

Robers

Kemp

Truman

(2013). Indicators of school crime and safety: 2012 (NCES 2013-036/NCJ 241446). Washington, DC: National Center for Education Statistics, U.S. Department of Education, and Bureau of Justice Statistics, Office of Justice Programs, U.S. Department of Justice.

43.

Servoss

T. J.

Finn

J. D.

(2014). School security: For whom and with what results? Leadership and Policy in Schools, 13(1), 61-92.

44.

Sheets

R. H.

(1996). Urban classroom conflict: Student-teacher perception: Ethnic integrity, solidarity and resistance. The Urban Review, 28, 165-183.

45.

Sheets

R. H.

Gay

(1996). Student perceptions of disciplinary conflict in ethnically diverse classrooms. NASSP Bulletin, 80(580), 84-93.

46.

Stewart

E. A.

(2003). School social bonds, school climate, and school misbehavior: A multilevel analysis. Justice Quarterly, 20, 575-604.

47.

Sugai

Horner

(2002). The evolution of discipline practices: School-wide positive behavior supports. Child & Family Behavior Therapy, 24(1-2), 23-50.

48.

Toldson

I. A.

(2011). Breaking barriers 2: Plotting the path away from juvenile detention and toward academic success for school-age African American males. Washington, DC: Congressional Black Caucus Foundation.