School Violence: Associations With Control, Security/Enforcement, Educational/Therapeutic Approaches, and Demographic Factors
By Nickerson, Amanda B Martens, Matthew P
Abstract. This study examined the extent to which three approaches to violence prevention and response were associated with the incidence of school crime and disruption after accounting for the influence of demographic variables. Secondary data analyses were conducted with four subsets of the sample of principals who completed the National Center for Education Statistics’ School Survey on Crime and Safety. Exploratory and confirmatory factor analyses conducted with Sample 1 (n = 426) and Sample 2 (n = 459), respectively, identified four approaches, labeled educational/ therapeutic, control, security/enforcement, and crisis plans. Confirmatory factor analyses with Sample 3 (n = 382) validated the constructs of school crime and disruption. Hierarchical regression analyses with Sample 4 (n = 440) indicated that demographic characteristics (e.g., enrollment, neighborhood crime) were associated with school crime, school disruption, or both. After accounting for demographic influences, security/enforcement (e.g., law enforcement, suspensions) was significantly associated with school crime and disruption.
Preventing and responding to school violence and disruption is a national priority. Although students are safer at school than away from school and there has been a decrease in school-based incidents of homicides, thefts, assaults, and other violent crimes since the early 1990s (DeVoe, Peter, Noonan, Snyder, & Baum, 2005), any act of violence at school is cause for concern. Policy decisions have been made mandating schools to use strategies to promote safety and security; however, these decisions have often been made swiftly, without the benefit of empirical scrutiny (Pagliocca & Nickerson, 2001). Therefore, it is critical to identify approaches to school violence that can be reliably measured and assess the extent to which these strategies are associated with decreased violence.
School Approaches to Preventing and Reducing Violence
Schools use a variety of strategies to prevent and reduce violence, which may be conceptualized broadly as emphasizing physical safety and security or focusing primarily on psychological safety. A focus on physical safety is often characterized by a “get tough” approach that includes zero tolerance policies (e.g., suspending students who violate school rules), restricting autonomy through the use of punitive measures, and policing functions, such as hiring resource officers and installing metal detectors (Noguera, 1995; Pagliocca & Nickerson, 2001). Approaches concerned with psychological safety are often educational or therapeutic, with the assumption that improving school climate, involving parents, teaching conflict resolution, and counseling prevent and reduce school disruption and crime (Noguera; Pagliocca & Nickerson).
Nine out of ten principals perceive very strict policies as essential for keeping schools safe (National School Safety Center, 2001). Zero tolerance has been argued to have led to decreased school violence hi the late 1990s by discouraging students from engaging in violence to avoid harsh consequences (Axtman, 2005). Zero tolerance is also intended to provide punishment uniformly, regardless of socioeconomic status (SES; Axtman). In addition, courts tend to rule hi favor of schools that use zero tolerance policies (Stader, 2004). The trend to control crime by hiring law enforcement officers and installing security hardware is reflected hi state laws. For example, the School Safety Program (1994) legislation was established hi Arizona for placing School Resource Officers on school grounds. Despite their increasing use, there has been considerable criticism of zero tolerance and related approaches, largely because of the lack of research regarding their effectiveness (Skiba, 2000). Students who have been suspended from school are more likely to be referred for disciplinary actions in the future (Tobin & Sugai, 1996). In addition, a disproportionate number of males and children from low socioeconomic and ethnic minority backgrounds are referred for disciplinary action (Skiba, Peterson, & Williams, 1997). The empirical study of using security personnel and hardware is essentially nonexistent.
There is increasing recognition of the importance of using educational/therapeutic approaches that are universal, apply a multiple systems approach, and support educators in promoting a sense of both physical and psychological safety (Gottfredson, 1997). Examples of these approaches include school-wide positive behavior support, teacher training, parent involvement, formal violence prevention programs, and modifying the environment or schedules to increase supervision. Comprehensive approaches that include training, supervision, restructuring, and teaching expectations and skills decrease antisocial behavior and office discipline referrals (Meltzer, Biglan, Rasby, & Sprague, 2001; Sprague et al., 2001). There is also empirical support for more specific components of these approaches, such as violence prevention programs (Taub, 2002) and peer mediation interventions (Bickmore, 2002). Modifying the school climate to improve communication and positive interactions has also been found to be more effective than using coercive disciplinary practices in inner city schools (Reinke & Herman, 2002). In addition, implementing family involvement activities is related to fewer student disciplinary referrals and suspensions (Sheldon & Epstein, 2002).
Despite these positive findings, there are several criticisms of educational/therapeutic approaches. First, some widely used prevention programs have not been shown to be effective (see e.g., German, 1998). second, grouping aggressive peers together for counseling and social skills intervention may create more problems by inadvertently reinforcing deviant behaviors and increasing opportunities for criminal activity (Arnold & Hughes, 1999). Third, there are a variety of challenges to implementing interventions that focus on the social-emotional functioning of students, such as staff resistance and lack of time (Hunter, Elias, & Noms, 2001).
Large-Scale Studies Examining School Approaches to Violence Prevention
Most research studies on approaches to violence prevention have been conducted in individual schools. Therefore, there is an absence of research using large, nationally representative samples of schools. Given the contention that “the organization of the school environment plays a critical role as either a facilitator or inhibitor of violence and disruption” (Mayer & Leone, 1999, p. 334), it is important to study multiple schools that engage in different practices to assess the extent to which these variables relate to decreased or increased violence.
Mayer and Leone (1999) examined approaches to school violence prevention and reduction using data from the nationally representative sample of high school students who completed the 1995 School Crime Supplement to the National Crime Victimization Survey. One subscale of this measure, secure Building, which measured attempts to secure and control the physical environment (e.g., security guards, metal detectors, locker checks) had a moderate standardized path value (.54) to School Disorder (i.e., student perceptions of violence and disruption). In contrast, System of Law, a construct that measured students’ perceptions of general knowledge of school rules and awareness of consequences, had a moderate negative standardized path coefficient (-.29) to School Disorder. The authors argued that future research should examine strategies reflecting other philosophical approaches to school violence prevention (e.g., school climate, family involvement). In addition, their model did not account for the various demographic characteristics that may affect school crime and disruption.
Demographic Characteristics Associated With School Violence
When examining influences on school crime and disruption, it is important to take into account demographic factors, such as school size, poverty, and level of neighborhood crime associated with increased violence. School crime is more apparent in large schools compared to smaller schools (Khoury-Kassabri, Benbenishty, Astor, & Zeira 2004). School size may also interact with other demographic factors such as SES of students and location of the school (Ma & Willms, 2004). For instance, poverty is associated with increased school crime (Gottfredson, 1997). Youth from inner cities, compared to those from other communities, are at greater risk for violent behavior (Redding & Shalf, 2001). Poverty, population turnover, and crime in the surrounding neighborhood are among the strongest predictors of school violence; in fact, a threatening community environment is a better predictor of violent victimization in school and on the way to and from school than actual school conditions (Laub & Lauritsen, 1998).
School level (e.g., elementary, middle, and high school) and percentage of students in special education are also linked to school violence. Secondary students are 13 times more likely than elementary students to be killed from a school-related violent incident (Mercy & Rosenberg, 1998). In addition, secondary school students are more likely than those from elementary and combined elementary-middle schools to rate a variety of crimes (e.g., weapons possession) and disruptive behavior (e.g., teacher abuse, physical fights) as problems in their schools (Ma & Willms, 2004). Across grade levels, it has also been shown that students in special education make more threats of violence than those in general education (Kaplan & Cornell, 2005). Overall, community and demographic variables have been shown to contribute the most variance in student delinquency and victimization, although other school variables (e.g., teacherstudent ratio) account for additional variance (Gottfredson & Gottfredson, 1985). In addition, Nickerson and Spears (2007) found that demographic and community variables are associated with schools’ use of different approaches to violence prevention and intervention. For example, rural schools used corporal punishment and suspension without services more often than schools in other locations. Schools serving low SES students were more likely to use security, random metal detector checks, and corporal punishment, regardless of neighborhood crime levels.
The purpose of this study was to examine the extent that different approaches to school violence prevention and reduction (a) can be differentiated empirically and (b) are associated with school crime and disruption after accounting for demographic variables (e.g., SES, school size, neighborhood crime). secondary data analyses were conducted with four subsets of the sample of principals who completed the National Center for Education Statistics’ School Survey on Crime and Safety (SSOCS) for the 1999- 2000 school year. To place the study in context, data were collected after the high-profile school shooting at Columbine High School but before the No Child Left Behind Act, which increased accountability for school safety.
The four samples in this study were derived from the nationally representative sample of principals of public elementary, middle, secondary, and combined schools who completed the SSOCS in the spring and summer of 2000. The original sample of 2,270 respondents (70% response rate) was stratified by school level, location, enrollment size, percentage of minority students, and geographic region. Special education, alternative, and vocational schools were excluded, as were schools that provided instruction in only prekindergarten, kindergarten, or adult education. The sample was divided into fourths to allow for multiple analyses, including psychometric analyses on the school approaches, confirmatory factor analysis (CFA) on the school violence construct, and regression analyses to assess the relationship between the school approaches and school crime and disruption after accounting for demographic variables. All sample subsets were screened and any case with missing data was deleted. This resulted in sample sizes of 426 (Sample 1) and 459 (Sample 2) for the psychometric analyses on the school approaches, 382 (Sample 3) for the CFA on the school violence construct, and 440 (Sample 4) for the regression analyses. Sample characteristics are displayed in Table 1.
The SSOCS was developed in consultation with a technical review board consisting of experts in school safety, including the Safe and Drug-Free Schools program and the Office of Special Education Programs. The survey development was an ongoing process that included a comprehensive literature review, internal and external reviews, and pretesting.
The SSOCS required respondents to report on several areas, including use of school policies and procedures; violence prevention programs; training; perceived limitations in efforts to reduce violence; violent, criminal, and disruptive incidents; availability and use of specific disciplinary actions; and student, neighborhood, and school characteristics for the 1999-2000 school year. All questions about the use of policies and procedures (e.g., using metal detectors, enforcing strict dress code), staff training, parental involvement, security personnel, and formal violence prevention programs were answered by selecting yes or no. Only respondents who responded “yes” to having formal violence prevention programs were asked follow-up questions about the types of violence prevention programs used. Principals also reported on the availability and use (i.e., Available and used; Available, but not feasible to use; Available but not used; Not available) of a variety of disciplinary actions (e.g., removal or transfer for at least one year, in-school and out-of-school suspension).
To assess school crime, respondents were asked to report the number of violent deaths (i.e., suicides and homicides) and a wide variety of other incidents that occurred during the 1999-2000 school year, including rape or attempted rape, physical attack or fight, threats of physical attack, robbery, possession of weapons, and possession or use of alcohol or drugs. To assess disorder, respondents rated how often a number of problem behaviors, including racial tensions, bullying, verbal abuse of teachers, and gang activities, occurred on a Likert-type continuous scale from 1 (Happens daily) to 5 (Never happens). Definitions were provided for many of these terms (e.g., gang, physical fight, robbery). In addition, respondents provided demographic information, including total enrollment, percentage of students eligible for reduced-cost lunch, special education, and neighborhood crime (high, moderate, low, mixed).
The use of practices, policies, and procedures regarding school crime and discipline were dummy coded (no = 0, yes = 1). To facilitate the factor analyses, the same dummy codes were used by converting the answer Available and used to yes and by collapsing all other responses (Available, but not feasible to use; Available but not used; Not available) into the no category. Although the SSOCS asks respondents to record the number of occurrences of each incident, we created ordinal categories of these variables. For low- frequency items, we coded O (no occurrence) and 1 (1 or more occurrence), and for more evenly distributed variables, we created seven levels of variables (None, 1-5, 6-10, 11-15, 16-20, 21-25, 25+ occurrences). Very low-incidence variables (e.g., homicide, rape) were excluded.
To preserve the anonymity of the schools, some of the demographic variables in the Public-Use Datafile were converted to categorical variables, such as enrollment (<300, 300-499, 500-999, >/=1000) and percentage of students eligible for free or reduced-cost lunch ( =20%, 21-50%, >/=57%). The other categorical variables of interest in this study were neighborhood crime (high, moderate, low, mixed), location (city, urban fringe, town, rural), and school level (elementary, middle, secondary, combined). To make variables more clear for interpretation, neighborhood crime levels of mixed and moderate were combined, and school level of middle and combined were combined. Variables were recoded so that higher numbers reflected higher levels (e.g., higher neighborhood crime, larger enrollment).
Identification of School Approaches
We used exploratory factor analysis (EFA) and CFA on separate samples in an effort to identify theoretically and empirically meaningful school approaches. All 66 items from the following sections of the SSOCS were considered for inclusion in the EFA: Characteristics of School Policies, Violence Prevention Programs and Practices, and Disciplinary Actions. Because only principals who indicated that their school had a formal violence prevention program answered 8 subsequent items, they were omitted from the analyses. Frequencies of yes and no responses were calculated for the remaining 58 items. Each item was then examined in terms of disproportionate endorsement (i.e., more than 95% of the sample endorsing yes or no) and relationship to the theoretical constructs of interest. Eight items that were disproportionately endorsed and rated by 4 of 5 experts with graduate training in school psychology and at least one course in crisis prevention and intervention as atheoretical or not corresponding to a particular construct of interest were omitted from the exploratory analyses, resulting in 50 remaining items.
An EFA using principal-axis extraction and promax rotation was conducted using Mplus software (Muthen & Muthen, 1998-2004). The Mplus program uses the proper estimation methods for both EFA and CFA with dichotomous variables, such as analyzing tetrachoric correlations and using a robust unweighted least squares estimator (Muthen, 1998-2004; Muthen & Muthen, 1998-2004). Because the SSOCS was not specifically designed to measure the constructs of interest in this study, EFA was used with Sample 1 to identify latent constructs that accounted for the relationships among the items. For all analyses, we used robust least squares estimation procedures. We chose an oblique (promax) rather than an orthogonal rotation method because we believed that any factors that emerged would be moderately correlated with each other. In the initial analysis, 14 items had poor loadings or cross-loaded on more than one factor. An examination of these items revealed that they were not clearly representative of either of the theoretical constructs of interest; these problematic items were eliminated and we conducted a final EFA on the remaining 36 items. Results from this analysis supported a four-factor solution, which accounted for approximately 52% of the variance in the items. There was a large gap (1.07) between the eigenvalues of the fourth and fifth factors (2.61 vs. 1.54) compared to the gaps between additional factors. Further, the standardized root mean square residual (SRMR) value of 0.07 suggested a well- fitting model (e.g., Hu & Bentler, 1999). Finally, with the exception of one item that cross-loaded onto multiple factors, which was subsequently dropped and not interpreted, all items demonstrated statistically significant and theoretically consistent loadings onto their primary factor (see Table 2). Items with weaker loadings were examined closely and decisions were made to include the items if they were clearly associated with one factor (as opposed to cross- loading) and if they were judged to be theoretically meaningful. The first factor was named Security/Enforcement, as all 13 items assessed practices related to closely supervising student behavior with security cameras or through the presence of law enforcement and enforcing rules by disciplining students. The second factor was named Crisis Plans because all S items that loaded on the factor asked principals if their school had a crisis plan that included procedures dealing with a variety of crises. The third factor was named Educational/Therapeutic, as its 12 items referred to formal violence prevention programs, teacher training, and parent involvement. The fourth factor was named Control because the 5 items that loaded on it assessed schools’ attempts to control behavior through the use of metal detectors, strict dress codes, clear book bags, and corporal punishment.
We next conducted a CFA that tested the 35-item, four-factor model with Sample 2. We used weighted least squares estimation procedures, which are appropriate with dichotomously scored data (Muthen, 1998-2004; Muthen & Muthen, 1998-2004). For identification purposes, we estimated one regression parameter for each factor to 1. The covariance between the factors was freely estimated, and each item was only allowed to load on its hypothesized factor. To assess model fit, we used the chi^sup 2^ statistic, Tucker-Lewis index (TLI; Tucker & Lewis, 1973), comparative fit index (CFI; Bentler, 1990), root mean square error of approximation (RMSEA; Steiger & Lind, 1980), and SRMR. Higher values for the TLI/CFI (e.g., X90- .95) and lower values of the RMSEA/SRMR (e.g., <.050 -.08) indicate better model fit (e.g., Hu & Bentler, 1999).
Results from the CFA indicated that the four-factor model accounted for a small portion of the variance in 2 items (one from security/Enforcement, R^sup 2^ = .037; one from Control, R^sup 2^ = .006). We removed these 2 items and conducted a “final” CFA on the 33 remaining items, revealing a well-fitting model. Although the chi^sup 2^ test was statistically significant (and therefore not indicative of a well-fitting model, chi^sup 2^ = 374.51, p < .001), such findings are common with relatively large sample sizes. The other fit indices generally provided support for a well-fitting model: TLI = .96, CFI = .95, RMSEA = 0.06, SRMR = .11. Values for the TLI, CFI, and RMSEA are consistent with often-cited criteria for model fit, although the SRMR value exceeds most conventional criteria. Standardized parameter estimates indicated that most items loaded strongly onto their hypothesized factors. In addition, a ?2 difference test indicated that the four-factor model fit significantly better than the one-factor model, chi^sup 2^^sub diff^ = 352.72, p < .0001. Together, these findings provide support for the hypothesized four-factor model.
Validation of School Disruption and School Crime
We used CFA with Sample 3 to validate the constructs of School Disruption and School Crime. All variables were hypothesized to be unidimensional, and CFAs were conducted separately for each using the Mplus program. For the analyses involving the School Disruption factor, we used standard maximum-likelihood estimation procedures because these items were coded on a continuous Likert-type scale. For identification purposes, in each analysis we set one regression parameter for each factor to 1.
For school crime, although the chi^sup 2^ statistic was statistically significant, chi^sup 2^(9) = 54.27, p < .001, other fit indices generally indicated an adequate model fit: CFI = .93, TLI = .89, SRMR = .05. Similar findings emerged for the school disruption variable. The chi^sup 2^ statistic was again statistically significant, chi^sup 2^(41) = 139.36, p < .001, but other fit indices were generally acceptable: CFI = 0.92, TLI = 0.95, SRMR = 0.09. The individual items for each factor and their standardized loadings are presented in Table 3. Almost all of the items loaded strongly (>.40) on their hypothesized factor, suggesting that the respective hypothesized latent factor accounted for meaningful amounts of variance in the individual items.
Influence of Demographics and School Approaches on School Disruption and Crime
Linear hierarchical regression was used among participants in Sample 4 to assess the extent to which the school approaches of security/enforcement, educational/therapeutic, and control were associated with the reported incidence of school disruption and crime after controlling for demographic variables. These variables were created by summing the individual items that loaded on each of the factors. The Crisis Plans factor was not included in this analysis because we determined it to be a preparedness strategy rather than a school approach to violence prevention and response. In addition, crisis plans vary greatly and document procedures used for a wide range of incidents, including events not involving violence (e.g., natural disasters).
We used hierarchical regression analyses because we were interested in determining the extent to which school approaches added predictive value to school crime and disruption, after accounting for the influence of demographics. As shown in the correlation matrix (Table 4), all of the predictor variables were correlated significantly with at least one of the outcome variables. Therefore, all demographic variables and school approaches were entered into the regression analyses. It should also be noted that the school approaches were moderately correlated with each other (security/enforcement and educational/therapeutic r = .10; educational/therapeutic and control r = .23; security/enforcement and control r = .28), indicating that they are not mutually exclusive. In analyzing frequency of responses, we found that all schools reported using one or more security/enforcement strategies and all but four schools reported using one or more educational/ therapeutic strategy. Control strategies were the least commonly used, with 168 (38%) schools not using any control strategies.
The first hierarchical regression was used to assess the association between school approaches and school crime after controlling for demographic characteristics. Seven demographic variables (enrollment, percentage low SES, percentage special education students, neighborhood crime, percentage minorities, grade level, and urbanicity) were entered into the first block of the regression equation, followed by the three school approaches (educational/therapeutic, security/enforcement, and control). The demographic variables accounted for 36% of the variance in the model, Adj R^sup 2^ = .36, p < .001. The addition of the school approaches added significantly to the model, DeltaR^sup 2^ = .02 for Step 2 (p < .01). Overall, the predictor variables accounted for 38% of the variance in school crime, Adj R2 = .38, F(10, 430) = 27.02, p < .001. An examination of the beta weights in Table 5 reveals that schools with larger enrollments, greater percentages of students in special education, and those that served older students were more likely to report school crime. After accounting for demographic influences on school crime, security/enforcement was the only school approach that was significantly associated with school crime in a positive direction, indicating that principals who reported use of more security and enforcement procedures (e.g., use of school security personnel, suspension, transfer) were also more likely to report more incidents of school crime.
The second hierarchical regression was the same as the first, except the outcome variable was school disruption. The demographic variables accounted for 12% of the variance in the model, Adj R^sup 2^ = .12, p < .001. The addition of school approaches added significantly to the model, DeltaR^sup 2^ = .02, p < .05. Overall, the predictor variables accounted for 14% of the variance in school disruption, Adj R^sup 2^ = .14, F(10, 430) = 7.87, p < .001. Similar to the first regression model, enrollment and percentage of students in special education were significantly related to disruption (see Table 5). In addition, level of crime in the neighborhood was a significant predictor of school disruption. Consistent with the first regression, educational/therapeutic and control were not associated with disruption, but security/enforcement was associated with a greater incidence of disruption, even after controlling for demographics.
Overall, results of this study indicated that (a) demographic variables account for substantial variance in disruption and crime, and (b) security/enforcement, or strategies used to secure the environment and enforce rules (e.g., security guards, suspension), was associated with more incidents of school crime and disruption. These findings are consistent with findings from over two decades ago revealing that community and demographic variables contribute the most variance to student delinquency (Gottfredson & Gottfredson, 1985). The finding that larger schools and those with a greater percentage of students receiving special education services reported more school crime and disruption is consistent with past research (Kaplan & Cornell, 2005; Khoury-Kassabri et al., 2004). Interestingly, location in urban areas was associated with school crime but not disruption, and neighborhood crime was related to school disruption but not crime. That SES did not contribute to school crime and disruption is also puzzling. It is possible that the percentage of children receiving free and reduced-cost lunch was not an adequate indicator of SES. Alternatively, Wright, Caspi, Moffitt, Miech, and Silva (1999) found that both low SES and high SES were related to delinquency (low SES promoted delinquency through increased alienation, aggression, decreased educational aspirations; high SES promoted delinquency via increased risk taking, social power, decreased conventional values), which may explain the lack of correlation in the current study. It is also possible that a third factor not measured in this study, such as parental management practices, mediated the relationship between SES and school crime and disruption. It is noteworthy that the security/ enforcement approach had near zero-order correlations with demographic variables, whereas control, and, to a lesser extent, educational/ therapeutic approaches were related to some of these variables. It is possible that there were differential demographic effects on specific practices within each approach that were not detected in our analyses. For example, Nickerson and Spears (2007) found that city school principals, as opposed to rural school principals, were more likely to report using security, but rural schools were more likely than other schools to suspend students without services. It is also possible that some of these practices (e.g., suspension, detention, use of law enforcement) are so widespread among schools that specific demographics are not related to these strategies.
The significant relationship between security/enforcement and school crime and disruption is consistent with Mayer’s and Leone’s (1999) findings that student reports of a secure building were also positively associated with increased disorder in schools. In addition, past research has indicated that suspension leads to increased discipline problems in the future (Tobin & Sugai, 1996) and punitive procedures are used disproportionately with males and children from ethnic minority backgrounds (Skiba et al., 1997). Given these findings, administrators and policy makers should carefully assess the use of, need for, and outcomes of these practices.
In contrast to the significant positive relationship between security/enforcement and school crime and disruption, both control and educational/therapeutic approaches were not significantly associated with the incidence of school crime and disruption. As noted previously, the analyses did not determine the extent to which this approach was used exclusively or in combination with other approaches, which may have affected results. For instance, it is likely that securing the physical safety of students, which may be accomplished through control or security strategies, may be necessary before attempting to ensure psychological safety through education and therapeutic intervention. It should also be noted that principals reported only whether the schools used these strategies, as opposed to the extent to which empirically validated strategies were used. Given the growing body of research supporting the effectiveness of some of these interventions, particularly those related to improving discipline and formal violence prevention programs (Meltzer et al., 2001; Sprague et al., 2001), it is possible that simply aggregating reported use of educational and therapeutic strategies does not capture the effects of these interventions. These results may also indicate that counseling and other therapeutic interventions are not potent enough to reduce violence (Gottfredson, 1997).
Implications for Practice
An important finding is that demographic characteristics accounted for much larger proportions of the variance in school crime and disruption than did any of the violence prevention or intervention approaches used by schools. This highlights the fact that school-based interventions may only affect violence in a limited way, underscoring the need for school-community partnerships to tackle this complex and multifaceted issue (see, e.g., Sheridan, Napolitano, & Swearer, 2002). For students having ongoing problems with violence, approaches that are most effective involve work with families (e.g., parent management training; Kazdin, 2003) or with the multiple systemic relationships in the student’s life, such as the family, peers, and school (e.g., multisystemic therapy; Henggeler, Schoenwald, Rowland, & Cunningham, 2002).
Findings about the demographics related to school crime and disruption also suggest that large schools and schools with large percentages of students with special education classifications are more likely to experience greater school crime and disruption. Therefore, districts may consider identifying schools with these characteristics to examine ways that the climate could be changed to reduce these incidents. For example, creating “houses” within schools or increasing supervision may be helpful, as well as ensuring that the students with special needs are provided with individualized educational and mental health supports. The lack of influence of several demographic characteristics (e.g., low SES, percentage of students from ethnic minority backgrounds) on school crime and disruption suggest that these variables are not necessarily associated with more behavior problems in schools. These findings underscore the importance of tailoring interventions to meet the specific needs of the school and its students, families, and surrounding communities.
School psychologists and other mental health professionals, as well as administrators and policy makers, may consider using results of this study to critically examine current practices related to violence prevention and intervention. Findings regarding the significant positive association between security/enforcement and school crime and disruption suggest that advocates of “get tough” approaches should critically examine the effects of these practices. Although these data do not allow for causal inferences to be made, they do suggest the need for further study to examine the possibility that these practices may have a detrimental effect. No firm conclusions can be drawn about control and educational/ therapeutic approaches, because of the lack of significant relationships between the use of these practices and school crime and disruption. It is well recognized that “one size does not fit all” when it comes to preventing and reducing the complex problem of violence, and results of this study highlight the difficulty of pinpointing specific approaches that work well across schools.
Because principals’ endorsement of using multiple educational/ therapeutic approaches did not have a significant relationship with school crime and disruption, yet the systematic, comprehensive implementation of these approaches has been shown to be effective (Meltzer et al., 2001; Sprague et al., 2001), it seems logical that quality is more important than quantity. Indeed, Sugai and Horner (1999) assert that the implementation of school-wide positive behavior support must be one of the top three goals for school improvement so that the necessary resources (e.g., materials, time, and staff) can be allocated to create an effective program.
One of the major limitations of this study was that school principals both completed the survey and reported on incidents of school crime and disruption. This resulted in a lack of independence between the independent and dependent variables as compared to using external criterion measures of disruption and school crime (e.g., office discipline referrals, suspensions, school vandalism). Rates of school disruption and crime may have been underrepoited in light of findings that school administrators report far less school violence than do students (Coggeshall & Kingery, 2001).
Another limitation is that the nature of the data collection process did not allow for the assessment of the fidelity with which the approaches were implemented. The SSOCS provided limited response options (e.g., yes or no) about the school practices used. Although this was not problematic for practices where it is clear that schools either have it or do not (e.g., out-of-school suspension, metal detectors, security personnel), it becomes more of an issue for items such as formal violence prevention programs, crisis plans, and providing technical assistance to parents, where there can be a great deal of variation in the type, quality, fidelity, and frequency of these practices. To illustrate the problem of yes or no responses, Sandoval, London, and Rey (1994) conducted follow-up interviews with school districts that endorsed having suicide prevention programs and found that only 58% actually had a program of at least 1 hr duration. Schools also used a variety of approaches to prevent and reduce violence, making it difficult to pinpoint specific strategies associated with school crime and disruption.
Although the purpose of the study was to examine the influence of school-level variables on school crime and disruption, it should be noted that important individual-level variables were excluded. It has been demonstrated that 6-9% of children account for more than half of the total discipline referrals and nearly all of the serious offenses in a school (Sprague & Walker, 2000). Therefore, it is likely that several factors specific to that individual (e.g., biological, parent relationships, and so on) are important in the overall incidence of disruption and crime in a school. Because this was a large-scale study, the level of study was the school rather than the individual, which did not allow for the study of these important issues.
Last, it is important to acknowledge the correlational nature of the data, which do not allow causal inferences to be made about the relationship between school approaches and incidence of violence and disruption. Although the regression analyses allowed us to test predictions about the extent to which violence prevention and intervention approaches contribute to the variance in school crime and disruption after accounting for demographics, the study was nonexperimental. Therefore, although security/enforcement is associated with increased crime and disruption, it is impossible to know whether the school practices led to more crime or disruption, or schools that had more crime and disruption then adopted more security and enforcement. Future Directions
Because of the relative lack of empirical studies on the relationship of approaches to violence prevention on actual incidents of school crime and violence, there are several directions for future research. Studies of this nature are best conducted with large, nationally representative samples, yet reliable and valid measures are needed. As stated previously, the SSOCS could be further improved by acquiring more detailed information about the educational/therapeutic approaches used. For instance, assessing who implements the intervention, how frequently it is used, and if it is based on theoretically and empirically sound practice would allow researchers to conduct more fine-grained analyses to attempt to determine which variables may have an influence on school crime and disruption.
Given that school administrators report far less school violence than do students (Coggeshall & Kingery, 2001), it is likely that perceptions about school approaches and their relation with school crime and disruption would vary depending on the informant. The “gold standard” for research in this area is to obtain independent verification of school crime and disruption. Therefore, future studies should compare the relative influence of different school practices on incidents of school crime and disruption as reported by multiple informants, such as administrators, teachers, and students. The outcome variables of school crime and disruption could be determined after examining the convergence of ratings across teachers, school psychologists, administrators, and students to enhance the validity of findings.
Although the Crisis Plans factor was not included in the regression analyses because of their wide variability and the fact that they are more of a preparedness strategy, there is a need to assess the effectiveness of these plans. In particular, it would be important to study the extent to which these are updated and exercised in schools. It might also be helpful to survey schools that have experienced a crisis to assess the extent to which their plans were useful in responding. From there, essential elements for crisis plans could be identified.
Arnold, M. E., & Hughes, J. N. (1999). First do no harm: Adverse effects of grouping deviant youth for skills training. Journal of School Psychology, 37, 99-115.
Axtman, K. (2005). Why tolerance is fading for zero tolerance in schools. Christian Science Monitor, 97, 1-2.
Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.
Bickmore, K. (2002). Peer mediation training and program implementation in elementary schools. Conflict Resolution Quarterly, 20, 137-160.
Coggeshall, M. B., & Kingery, P. M. (2001). Crosssurvey analysis of school violence and disorder. Psychology in the Schools, 38, 107- 116.
DeVoe, J. F., Peter, K., Noonan, M., Snyder, T. D., & Baum, K. (2005). Indicators of school crime and safety: 2005 (NCES 2006-001/ NCJ 210697). U.S. Departments of Education and Justice. Washington, DC: U.S. Government Printing Office.
Gorman, D. M. (1998). The irrelevance of evidence in the development of school-based drug prevention policy, 1986-1996. Evaluation Review, 22, 118-146.
Gottfredson, D. (1997). School-based crime prevention. In L. W. Sherman, D. Gottfredson, D., MacKenzie, J. Eck, P. Reuter, & S. Bushway (Eds.), Preventing crime: What works, what doesn’t, what’s promising: A report to the United States Congress. Washington, DC: Department of Justice, Office of Justice Programs.
Gottfredson, G. D., & Gottfredson, D. C. (1985). Victimization in schools. New York: Plenum Press.
Henggeler, S. W., Schoenwald, S. K., Rowland, M. D., & Cunningham, P. B. (2002). Serious emotional disturbance in children and adolescents: Multisystemic therapy. New York: Guilford Press.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.
Hunter, L., Elias, M. J., & Noms, J. (2001). School-based violence prevention: Challenges and lessons learned from an action research project. Journal of School Psychology, 39, 161-175.
Kaplan, S. G., & Cornell, D. G. (2005). Threats of violence by students in special education. Behavioral Disorders, 31, 107-119.
Kazdin, A. E. (2003). Problem-solving skills training and parent management training for conduct disorder. In A. E. Kazdin & J. R. Weisz (Eds.), Evidence-based psychotherapy for children and adolescents (pp. 241-262). New York: Guilford Press.
Khoury-Kassabri, M., Benbenishty, R., Astor, R. A., & Zeira, A. (2004). The contributions of community, family, and school variables to student victimization. American Journal of Community Psychology, 34, 187-204.
Laub, J. H., & Lauritsen, J. L. (1998). The interdependence of school violence with neighborhood and family conditions. In D. S. Elliot, B. A. Hamburg, & K. R. Williams (Eds.), Violence in American schools (pp. 127-155). New York: Cambridge University Press.
Ma, X., & Willms, J. D. (2004). School disciplinary climate: Characteristics and effects on eighth grade achievement. The Alberta Journal of Educational Research, 50, 169-188.
Mayer, M. J., & Leone, P. E. (1999). A structural analysis of school violence and disruption: Implications for creating safer schools. Education and Treatment of Children, 22, 333-356.
Meltzer, C. W., Biglan, A., Rasby, J. C., & Sprague, J. R. (2001). Evaluation of a comprehensive behavior management program to improve school-wide positive behavior support. Education and Treatment of Children, 24, 448-479.
Mercy, J. A., & Rosenberg, M. L. (1998). Preventing firearm violence in and around schools. In D. S. Elliot, B. A. Hamburg, & K. R. Williams (Eds.), Violence in American schools (pp. 159-187). New York: Cambridge University Press.
Muthen, B. O. (1998-2004). Mplus technical appendices. Los Angeles: Muthen & Muthen.
Muthen, L. K., & Muthen, B. O. (1998-2004). Mplus user’s guide (3rd ed.). Los Angeles: Muthen & Muthen.
National School Safety Center. (2001). Review of school safety surveys: School crime and violence statistics. Retrieved July 6, 2005, from http://www.nsscl.org/ studies/ statistic%20resourcespdf.pdf
Nickerson, A. B., & Spears, W. H. (2007). Influences on authoritarian and educational/therapeutic approaches to school violence prevention. Journal of School Violence, 6,3-31.
Noguera, P. A. (1995). Preventing and producing violence: A critical analysis of responses to school violence. Harvard Educational Review, 65, 189-212.
Pagliocca, P. M., & Nickerson, A. B. (2001). Legislating school crisis response: Good policy or just good politics? Law and Policy, 23, 373-407.
Redding, R. E., & Shalf, S. M. (2001). The legal context of school violence: The effectiveness of federal, state, and local law enforcement efforts to reduce gun violence in schools. Law and Policy, 23, 297-343.
Reinke, W. M., & Herman, K. C. (2002). Creating school environments that deter antisocial behavior in youth. Psychology in the Schools, 39, 549-559.
Sandoval, J., London, M. D., & Rey, T. (1994). Status of suicide prevention programs in California schools. Death Studies, 18, 595- 608.
School Safety Program, A.R.S. [section] 15-154 (1994).
Sheldon, S. B., & Epstein, J. L. (2002). Improving student behavior and school discipline with family and community involvement. Education and Urban Society, 35, 4-26.
Sheridan, S. M., Napolitano, S. A., & Swearer, S. M. (2002). Best practices in school-community partnerships. In A. Thomas & J. Grimes (Eds.), Best practices in school psychology IV (pp. 321-336). Bethesda, MD: National Association of School Psychologists.
Skiba, R. J. (2000). Zero tolerance, zero evidence: An analysis of school disciplinary practice (Policy Research Report No. SRS2). Bloomington: Indiana University, Indiana Education Policy Center.
Skiba, R. J., Peterson, R. L., & Williams, T. (1997). Office referrals and suspension: Disciplinary intervention in middle schools. Education and Treatment of Children, 20, 295-315.
Sprague, J., & Walker, H. (2000). Early identification and intervention for youth with antisocial and violent behavior. Exceptional Children, 66, 367-379.
Sprague, J., Walker, H., Golly, A., White, K., Myers, D. R., & Shannon, T. (2001). Translating research into effective practice: The effects of a universal staff and student intervention on indicators of discipline and school safety. Education and Treatment of Children, 24, 495-511.
Stader, D. L. (2004). Zero tolerance as public policy: The good, the bad and the ugly. The Clearing House, 78, 62-66.
Steiger, J. H., & Lind, J. C. (1980, May). Statistically based tests for the number of common factors. Paper presented at the annual meeting of the Psychometric Society, Iowa City, IA.
Sugai, G., & Horner, R. (1999). Discipline and behavioral support: Practices, pitfalls, and promises. Effective School Practices, 17, 10-22.
Taub, J. (2002). Evaluation of the second Step Violence Prevention Program at a rural elementary school. School Psychology Review, 31, 186-201.
Tobin, T., & Sugai, G. (1996). Patterns in middle school discipline records. Journal of Emotional and Behavioral Disorders, 4, 82-95.
Tucker, L. R., & Lewis, C. (1973). A reliability coefficient for maximum likelihood factor analysis. Psychometrika, 38, 1-10.
Wright, B. R., Caspi, A., Moffitt, T. E., Miech, R. A., & Silva, P. A. (1999). Reconsidering the relationship between SES and delinquency: Causation but not correlation. Criminology, 37, 175- 194. Date Received: September 19, 2006
Date Accepted: November 5, 2007
Action Editor: Susan Swearer
Amanda B. Nickerson
University at Albany, State University of New York
Matthew P. Martens
University of Memphis
This research was supported by a grant from the American Educational Research Association, which receives funds for its ABRA Grants Program from the National Science Foundation and the National Center for Education Statistics of the Institute of Education Sciences (U.S. Department of Education) under National Science Foundation grant REC-0310268. Opinions reflect those of the authors and do not necessarily reflect those of the granting agencies.
Correspondence regarding this article should be addressed to Amanda B. Nickerson, Division of School Psychology, University at Albany, State University of New York, 1400 Washington Avenue, ED 232, Albany, NY 12222; E-mail: email@example.com
Amanda B. Nickerson, PhD, is Assistant Professor of school psychology at the University at Albany, State University of New York. Her research interests include school violence prevention and intervention, assessment and intervention of children with emotional and behavioral disorders, and parent and peer relationships.
Matthew P. Martens, PhD, is Associate Professor of counseling psychology at the University of Memphis. His research interests include addictive behaviors, psychological assessment, and statistics-evaluation.
Copyright National Association of School Psychologists Jun 2008
(c) 2008 School Psychology Review. Provided by ProQuest Information and Learning. All rights Reserved.