Abstract
School crime has been a national issue for nearly 40 years and remains a concern for students, administrators, parents, and the public. Schools engage in numerous strategies aimed at curbing crime, ranging from harsh disciplinary practices to proactive strategies focused on gaining student compliance. This study examines the impact of disciplinary practices on in-school delinquency, while also considering the influence of students’ perceptions of injustice. Using student- and school-level data from the Rural Substance Abuse and Violence Project and hierarchical Poisson regression analyses, findings reveal that students’ perceptions of injustice were significantly related to in-school delinquency, while proactive and reactive discipline practices, spanning the punitiveness continuum, were not. The findings provide tentative guidance for school-based discipline management policies and practices.
Introduction
Despite evidence that school crime has remained stable or decreased each year since 1985, the perceptions that violent school crime is on the rise and that schools are unsafe are widespread (Chouhy, Madero-Hernandez, & Turanovic, 2017; DeVoe et al., 2004; Dinkes, Kemp, Baum, & Snyder, 2009; Musu-Gillette, Zhang, Wang, Zhang, & Ouderkerk, 2017). According to the 2015 Gallup Student Poll, only 31% of high school juniors strongly agreed that they felt safe in their schools (Calderon, 2016). Furthermore, nearly three in 10 (28%) parents in the United States feared for their child’s physical safety while at school (Auter, 2016; McCarthy, 2015). Trends suggest that this parental concern spikes after highly publicized mass shootings at schools (e.g., 55% after Columbine; 33% after Newtown) (Auter, 2016). Findings from opinion polls also suggest that lack of discipline in schools is considered a serious problem (Kernan-Schloss & Starr, 2015).
In response, schools across the country have implemented numerous policies and practices aimed at preventing in-school delinquency, some of which are embedded in a more punitive discipline management approach. In fact, since the early 1990s, many schools have implemented a zero-tolerance philosophy (American Psychological Association Zero Tolerance Task Force, 2008). The zero-tolerance approach requires the use of predetermined and severe sanctions for misconduct in schools, and it is based in part on the assumption that severe penalties will deter crime. This has resulted in increases in the frequency and length of exclusionary sanctions (Kupchik, 2016). For example, during the 2011-2012 academic year, 3.2 million public school students received out-of-school suspensions and 3.4 million received in-school suspensions (Musu-Gillette et al., 2017).
Research over the past decade has cast an important spotlight on the collateral consequences of harsh school discipline among those receiving it, with the notion of “school to prison pipeline” now part of the criminological lexicon (e.g., Hirschfield, 2018; Kupchik, 2016). Less is known about whether such punitive policies affect the behavior of students who remain in school. Are they more or less effective than other less punitive reactive approaches—such as the use of traditional detention, or “community service sentencing” to punish inappropriate behavior, or the use of privileges and rewards to reinforce prosocial behavior? Are they more or less effective than proactive approaches to creating a safe, well-managed space, including the provision of treatment services, use of a character-focused curriculum, and student-centered (i.e., communal) leadership?
The answers to these questions remain somewhat elusive because few studies focus on the relative effects of such a comprehensive range of discipline management practices on student delinquency, including those in proactive and reactive domains mentioned above. Instead, a good deal of what we know about delinquency prevention in schools comes from evaluation of specific programs or practices—such as the D.A.R.E. program, communal school organization, and the presence of police officers in schools (e.g., Lynam et al., 1999; Na & Gottfredson, 2013; Payne, Gottfredson, & Gottfredson, 2003). Furthermore, most previous studies of school disciplinary practices do not examine their effects within an analytic framework that explicitly distinguishes the impact of school-level practices from individual student differences, including differences in perceptions of fairness regarding school discipline that may also affect student behavior. This is an important limitation given evidence of disparities in discipline practices—in particular, evidence that racial and ethnic minority students and economically disadvantaged students are more likely to experience punitive school discipline and less likely to encounter more communal or restorative approaches (Payne & Welch, 2010, 2018)
The current study attempts to overcome these gaps in the literature by taking into consideration the potential predictors of student behavior in school at multiple levels of analysis. We explore the impact of a wide array of school proactive and reactive discipline practices, alongside student perceptions of injustice regarding school discipline, on in-school delinquency. Specifically, we examine these effects through multilevel models using three waves of data from the Rural Substance Abuse and Violence Project (RSVP), in which over 7,200 individual cases were nested within 137 unique school contexts.
Literature Review
In-school delinquency has been linked to a number of factors, including student demographic characteristics (e.g., Bailey, Flewelling, & Rosenbaum, 1997; DuRant, Getts, Candenhead, & Woods, 1997; Kann et al., 1995; Wilcox & Clayton, 2001), low self-control (see Pratt & Cullen, 2000, for a detailed review), social bonds, peer associations (e.g., Agnew, 2001), and school compositional characteristics (e.g., Astor, Behre, Fravil, & Wallace, 1997; Felson, Liska, South, & McNulty, 1994; Gottfredson & Gottfredson, 1985). The present study, accounting for these factors, focuses on more fully understanding the impact of school reactive and proactive practices, along with student perceptions of discipline management, on in-school delinquency. The literatures regarding these key areas of focus are reviewed below.
School Discipline-Related Policies and Practices
Schools utilize numerous mechanisms aimed at reducing or preventing in-school delinquency. For example, schools spend a great deal of resources on physical crime prevention technologies and personnel. In 1994, approximately 13% of schools used private security or school resource officers (Addington, 2009). By 2016, this number had risen considerably, with 54% of high schools using police, school resource officers, or security guards daily (Centers for Disease Control and Prevention, 2016). In addition in the same year, nearly 86% of high schools used security or surveillance cameras and over 6% had installed metal detectors (Centers for Disease Control and Prevention, 2016). Prior research, some utilizing the same data as the current study, finds that the policing and target hardening strategies reviewed above are largely ineffective at reducing school disorder and delinquency (see Jennings, Khey, Maskaly, & Donner, 2011; Na & Gottfredson, 2013; Schreck, Miller, & Gibson, 2003; Tillyer, Fisher, & Wilcox, 2011, for examples). In fact, a recent review of such school security measures by Jonson (2017, p. 968) pointed to the tendency of schools to initiate them without any apparent consultation of the empirical evidence regarding their effectiveness (see also Gottfredson, Cook, & Na, 2012). Other studies have suggested that certain physical crime prevention and personnel practices, such as having school police officers, can negatively impact student perceptions and foster feelings of mistrust (Na & Gottfredson, 2013). 1
Additional prevention strategies that have been widely used by schools are the “resistance” educational or training programs. Two popular programs are the Drug Abuse Resistance Education (D.A.R.E.) and the Gang Resistance Education and Training (G.R.E.A.T.). Despite their popularity, research findings suggest that these programs may not achieve the intended long-term goals. In particular, D.A.R.E. has been evaluated numerous times, with outcomes indicating it has no impact on students’ drug use (e.g., Lynam et al., 1999). In contrast, there is much more promising evidence regarding other instructional programs, especially those that focus on building self-control and social competency among at-risk youth through use of cognitive-behavioral techniques (see Gottfredson, 2001; Gottfredson et al., 2012, for reviews). A number of specific examples of such programming exist, but two of the more well-known include Promoting Alternative Thinking Strategies (PATHS) and Families and Schools Together (FAST Track). While target-hardening strategies and educational prevention programs are abundant, the focus of the present study is instead on “discipline management” practices in terms of (a) punitive versus nonpunitive reactions to student behavior, and (b) proactive nonpunitive, “communal” practices.
Reactive school practices
One group of discipline management practices that schools have traditionally utilized are reactive responses to misbehavior. These reactive strategies are based on the ideas presented in deterrence theory, which suggests that individuals participate in delinquency when the benefits outweigh the costs. In other words, deterrence is the omission or curtailment of a crime due to fear of punishment (Gibbs, 1975). When the punishment associated with a crime is certain, severe, and quickly enforced, individuals are hypothesized to be deterred from crime. Previous literature, however, indicates that there is small to negligible empirical support for deterrence theory, and this is especially true in relation to the hypothesis that the likelihood of committing crime is inversely related to the severity of punishment (see Paternoster & Bachman, 2013; Pratt, Cullen, Blevins, Daigle, & Madensen, 2006, for reviews).
Along these same lines, zero-tolerance policies in schools have been shown to be ineffective at controlling school crime (American Psychological Association Zero Tolerance Task Force, 2008). The deterrent value of harsh discipline approaches may be lacking due to the notion of perceptual deterrence—the idea that decisions about committing crime are influenced by perceived sanction as opposed to actual sanction (e.g., Paternoster & Bachman, 2013). The lack of a deterrent effect of zero-tolerance policies may also be due to the fact that school punishments are not determined by offense severity (Hirschfield, 2018). In other words, all students are punished severely no matter how minor their offense may be. Thus, these zero-tolerance practices may leave students feeling vulnerable to having their rights violated (Bracy, 2011; Kupchik, 2010; Shedd, 2015). This is similar to findings regarding intensive supervision programs, where the requirements are more intensive, demanding, and severe when compared with traditional probation (see Lowenkamp, Flores, Holsinger, Makarios, & Latessa, 2010, for review). Some research has suggested that rewards in response to good behavior, rather than punishments in response to bad behavior, may be important for regulating school crime (Gottfredson & Gottfredson, 1985; Payne, 2015).
Similarly, research has indicated that there are beneficial outcomes associated with utilizing other nonpunitive reactive responses to school delinquency. For example, restorative justice techniques to address misbehavior in school have been found to be effective (e.g., Gardella, 2015; Payne & Welch, 2015; Skiba & Rausch, 2006). Restorative justice techniques focus on reconciliation and resolution, rather than punishment, as is the case with punitive punishment strategies (Gardella, 2015; Payne & Welch, 2018). Students in schools that utilize restorative approaches are less likely to receive office referrals and participate in less misbehavior at school (Rideout, Karen, Salinitri, & Marc, 2010; Schiff, 2013). Moreover, they encourage the building of positive feelings within the school community.
Proactive school strategies
Schools also utilize proactive strategies to encourage compliance with rules. There is a growing literature supportive of the idea that safer schools are those with clear and consistently enforced rules, those that put a high value on actively promoting healthy social norms, and those in which students feel integral to the organization and management of the school (e.g., Gottfredson et al., 2012; Gottfredson et al., 2005). For example, in what is often viewed as the first major study of school practices and student delinquency, Gottfredson and Gottfredson (1985) found that schools with the most misconduct shared similar characteristics, including inactive administrators and poor cooperation between principals and teachers. Research since that landmark study indicates that there are numerous benefits of communally organized schools, which are those where “members know, care about, and support one another, have common goals and sense of shared purpose, and to which they actively contribute and feel personally committed” (Solomon, Battistich, Kim, & Watson, 1997, p. 236). Some of the benefits include higher levels of teacher efficacy, work enjoyment, and morale (Bryk & Driscoll, 1988); student self-esteem, empathy, academic motivation, and overall bonds to school (Battistich, Solomon, Kim, Watson, & Schaps, 1995; Payne et al., 2003; Solomon, Schaps, Watson, & Battistich, 1992); and lower levels of misconduct (see Battistich & Hom, 1997; Payne, 2008; Payne et al., 2003, for examples).
There is also research that indicates school-wide positive behavior intervention and support programs are effective at reducing misconduct. Positive behavior intervention and support programs are made up of interventions that address the problem behavior, are justified by the outcomes, and are acceptable to everyone within the community. Specifically, they focus on defining and teaching behavioral expectations, clearly defining consequences for problematic behavior, and rewarding appropriate behavior (Horner, Sugai, & Lewis, 2015). The literature suggests that schools with positive behavior intervention and support programs experience fewer office discipline referrals (e.g., Barrett, Bradshaw, & Lewis-Palmer, 2008; Bohanon et al., 2006; Farkas et al., 2012) and lower suspension rates (e.g., Barrett et al., 2008).
Student Perceptions of Justice
One explanation for why school-level reactive discipline practices, especially harsh reactive ones, do not reduce school delinquency is because they may undermine perceived fairness. Zero-tolerance practices, for example, have caused students to perceive favoritism (McNeal & Dunbar, 2010), which weakens trust in authorities and reduces perceived fairness. These zero-tolerance practices may also cause feelings of unfairness because they do not allow for exceptions or adjustments in punishment, regardless of offense severity. Exclusionary policies have also been found to undermine trust for authority among all students in the school, not just those who are sanctioned (Kirk & Matsuda, 2011). Conversely, it is also possible that some reactive policies—particularly nonpunitive ones that recognize and reward good behavior—strengthen students’ trust and therefore increase perceptions of fairness at school. Thus, reactive school policies and practices may have both direct and indirect (through perceptions of fairness vs. injustice) impacts on in-school delinquency.
Beyond the school literature, a great deal of research indicates when rules or treatment are perceived as unfair, compliance is lower (Paternoster, Brame, Bachman, & Sherman, 1997; Tyler, 1990; Tyler & Huo, 2002). Conversely, when authorities are perceived as fair and legitimate, people are more willing to comply with the law (Sunshine & Tyler, 2003; Tyler, 2006; Tyler & Jackson, 2013). Tyler and Jackson (2013), for example, conducted a national survey of U.S. citizens in 2013. Their study sought to determine whether the legitimacy of legal authorities and the law motivated compliance with every day laws, cooperation with police, and engagement. Legitimacy was captured through measures of obligation, trust and confidence, and normative alignment (Tyler & Jackson, 2013). The findings revealed that fairness of decision-making and fairness of interpersonal treatment were important factors for both compliance and cooperation (Tyler & Jackson, 2013). Notably, the findings highlighting the importance of perceptions of fairness span many different environments and rules, including workplace guidelines, police orders, and prison rules (e.g., Blader, Chang, & Tyler, 2001; Jackson, Bradford, Stanko, & Hohl, 2012; Myhill & Quinton, 2011; Sparks & Bottoms, 1995; Tyler, 2006).
The research that applies these ideas to the school context also finds support. Much work on perceived fairness and clarity of rules among school students has examined the construct as a school-level measure (e.g., measured as average perceptions of fairness among students or in multivariate analysis of covariance) and examined its correlations with school rates of misconduct. When students view disciplinary practices as fair and consistently enforced, less overall school crime is reported, as indicated by rates of misconduct, delinquency, or victimization (Arum & Way, 2003; Gottfredson et al., 2012; Gottfredson & Gottfredson, 1985; Mayer & Leone, 1999; Welsh, 2001, 2003). For example, using a nationally representative sample of over 250 secondary schools, Gottfredson, Gottfredson, Payne, and Gottfredson (2005) explored the impact of several measures, including fairness of rules and clarity of rules, on school rates of victimization and delinquency. They found that when rules were perceived as fair and clear, schools experienced lower levels of student delinquency and victimization (Gottfredson et al., 2005). Research indicates that even schools with strict rules report fewer delinquent incidents if those rules are, overall, considered fair (Arum & Way, 2003; Cornell, Shukla, & Konold, 2015).
Perceptions regarding rule clarity or disciplinary (in)justice have also been measured at the individual level and examined in relation to several student-level outcomes. Burrow and Apel’s (2008) analysis of the National Crime Victimization Survey’s (NCVS) School Crime Supplement included an index tapping the extent to which student respondents perceived school rules to be universally understood, fair, and strictly and consistently enforced. This measure was inversely related to student experiences with assault and larceny victimization. While important, such findings have not examined perceptions of fairness on the part of students while also considering the effects of actual school-level disciplinary practices. As such, we have little evidence to date how school disciplinary practices work in relation to student perceptions.
In an important exception, Way (2011) conducted a multilevel study using data from the National Educational Longitudinal Study. She explored the impact of several measures of student perceptions and school discipline policies on the frequency of classroom disruption among 10,992 students from 1,132 schools (Way, 2011). Findings revealed that more school rules and stricter school rules increased classroom disruptions. However, individual students who perceived school authority as being legitimate were rated as less disruptive (Way, 2011).
Way’s (2011) study is an important beginning to exploring the impact of school-level practices and student perceptions simultaneously in the school context, but it focuses on a limited range of school discipline policies and in-school misbehavior. For example, Way included just two measures of school discipline policy—number of school rules and punishment severity. In addition, the “classroom disruptive behavior” dependent variable tapped behavior that rarely rises to the level of “delinquency.” In comparison, there are numerous areas beyond the classroom where more serious in-school delinquency can and does occur, including hallways, stairwells, playgrounds, parking lots, and the cafeteria. Indeed, in Astor, Meyer, and Behre’s (1999) examination of places and times in which violence occurs within high schools, all 166 reports of in-school violence occurred in locations within the school where there were few teachers (i.e., not the classroom). Thus, the present study applies a multilevel approach similar to Way, yet examines a variety of both reactive (punitive and nonpunitive) and proactive school practices in conjunction with student-level perceptions of injustice to gain a fuller understanding of student delinquency throughout the school environment.
The Present Study
The overall purpose of the current study is to determine whether and how various school-based disciplinary practices—ranging the punitiveness continuum—impact in-school delinquency. Do the actual practices matter? If so, do various reactive and proactive disciplinary practices serve to increase or decrease the likelihood of student delinquency, and do the effects hold once individual perceptions of (in)justice are considered? Or, is it the case that perceptions of (in)justice regarding discipline matter most?
Data
Such questions are addressed using data collected through the Rural Substance Abuse and Violence Project (RSVP). This project was a longitudinal study conducted in Kentucky schools from spring 2001, when the students were in seventh grade, to spring 2004 (Ousey & Wilcox, 2007). In the current study, the first three waves of the four-wave data were examined.
Students in this study were selected using a multistage sampling technique. First, using a stratified sampling procedure, 30 of the 120 counties in Kentucky were randomly selected. Only 74 public schools in these 30 counties included seventh grade, all of which were contacted. Sixty-five of the 74 principals agreed to participate. These 65 schools contained 9,488 seventh graders in Wave 1, all of whom were targeted. Parental consent was necessary before the students could participate in the study; 43% of parents gave consent. Thus, a total of 4,102 students were included in the final sample (Ousey & Wilcox, 2007).
The researchers used a mass administration method to distribute the surveys to the students. In particular, surveys were administered to students during one class period between March and May (see, e.g., Ousey, Wilcox, & Fisher, 2011; Wilcox, Augustine, & Clayton, 2006). The survey participation rate varied across the waves; completed student surveys were received from 3,692 students in Wave 1, 3,638 students in Wave 2, and 3,050 students in Wave 3 (Ousey & Wilcox, 2007). Students at each wave were treated as independent cases, resulting in 7,296 “student-wave” cases, and student-wave survey responses were used to construct individual-level variables. As students could change schools, or schools could change their disciplinary practices across waves, we treated the students’ school context for each wave as a unique school context, resulting in a total of 137 “school-year” contextual units.
Teachers and principals from the sample schools were also administered questionnaires during the first three waves of RSVP (such data were not fully collected during Wave 4). The teacher survey contained questions regarding classroom-level behavior management, school climate, and teacher characteristics, while the principal questionnaire asked about school-level behavior management and principal characteristics. These were either group-administered during a faculty meeting or handed to the teachers to complete during their planning period on the same day that student surveys were administered (Wilcox et al., 2006). Teacher and principal survey responses were each aggregated within schools (per wave) to create school-level disciplinary measures for 137 unique school contexts (school-wave cases).
For the current study, listwise deletion was used when there were missing data on study variables or the school code. Therefore, analyses were based on 7,269 student-wave cases nested within 137 school-wave contexts. Within the 137 school-wave contexts, the average number of students was approximately 72.
Measures
Dependent variable
The dependent variable, In-School Delinquency, was measured using a delinquency variety index. The variety index is a count of the unique types of in-school delinquent behaviors reported. Eight items from the student survey were used: “In the present school year, how often have you (a) forced someone at school to give up their money or property?; (b) stolen someone’s money or property at school when they were not around?; (c) physically attacked someone at school?; (d) said unwelcome sexual remarks to someone at school?; (e) touched someone in a sexual manner without their consent or against their will at school?; (f) taken a gun to school?; (g) taken an explosive to school?; (h) taken another weapon to school?” The possible responses to these items ranged from 0 (never) to 4 (daily or almost daily). We dichotomized (0 = never; 1 = ever) then summed the recoded eight items, resulting in an index that ranged from 0 to 8. Descriptive statistics, reported for all variables in Table 1, show that students reported an average of 0.52 (SD = 1.12) types of in-school delinquent behaviors.
Descriptive Statistics of Dependent, Independent, and Control Variables.
Key independent variables
The key independent variable at Level 1, or the student level, is Perceived Injustice. This variable was created by calculating the mean of eight items from the student survey. Respondents indicated the extent to which they agreed with the following statements: (a) All students are treated fairly; (b) The school rules are fair; (c) The punishment for breaking school rules is the same for all students no matter who you are; (d) The school rules are strictly enforced; (e) If a school rule is broken, students know what kind of punishment will follow; (f) The teachers keep order in the classroom; (g) The teachers are fair; and (h) Teachers are interested in students (Cronbach’s α = 0.857). Each item was measured on a 4-point response scale: strongly agree (1), agree (2), disagree (3), and strongly disagree (4). Thus, higher average scores across the eight items indicate higher levels of perceived injustice.
The key independent variables at Level 2, or the school level, are broken up into two distinct categories: Reactive Strategies and Proactive Strategies. A principal-components factor analysis, using varimax rotation, was conducted to guide construction of variables within each of these categories. Indexes were created in instances where multiple items loaded highly on particular factors. However, there were several factors with only single highly loading items.
Reactive strategies
Six variables were included in the current study to capture reactive practices. The first two measures, Privileges and Detention, are single-item measures constructed from items on the teacher survey that did not load with other reactive discipline items. Teachers indicated the extent to which they used specific practices when dealing with student misbehaviors. Privileges is the extent to which teachers give students privileges to increase positive involvement within the teacher’s individual classroom. Detention represents the extent to which teachers give students detention. Each item was measured on a 5-point scale: never (1), seldom (2), sometimes (3), often (4), and always (5).
The final four reactive practices measures were created from items on the principal survey. These include Punitive Punishment, Community Service, Reward Practices, and Plea-Bargaining Frequency. Punitive Punishment is the average of four items where principals identified the extent to which their school used the following responses to student misconduct: (a) police or court action against student, (b) police or court action against parents, (c) expulsion from school, and (d) suspension from school (Cronbach’s α = 0.684). 2 Each item was measured on a 5-point response scale: never (1), seldom (2), sometimes (3), often (4), and always (5). Community Service was measured with a single item indicating the extent to which principals used “community service” in response to student misconduct (1 = never, 5 = always).
Reward Practices was created by taking the average of seven items from the principal survey to capture school-wide responses to student behavior. Principals indicated the extent to which their school used the following rewards for desirable student conduct: (a) formal recognition or praise, (b) informal recognition or praise, (c) activity reinforcers (e.g., free time, playground, games), (d) social rewards (parties, trips), (e) material rewards (e.g., food, toys), (f) money, and (g) redeemable tokens. Each item was measured on the same 5-point scale (1 = never, 5 = always). The Cronbach’s alpha for this measure is 0.822. 3
Finally, 14 items from the principal survey were averaged to create Plea-Bargaining Frequency. Respondents indicated the “frequency of adjustment (plea-bargaining) in the application of school-mandated punishment” for (a) possession of tobacco; (b) possession of alcohol; (c) possession of other drugs; (d) possession of a knife; (e) possession of a gun; (f) possession of other weapons; (g) physical fighting; (h) theft ($10 or less) from desk, closet, or other place at school; (i) theft (more than $10) from desk, closet, or other place at school; (j) unwelcome verbal sexual advances or propositions; (k) unwelcome physical sexual advances or propositions; (l) chronic truancy; (m) obscene remarks to teacher or administrator; and (n) vandalism (Cronbach’s α = 0.942). These items were measured on a 5-point scale (1 = never, 5 = always).
Proactive strategies
The proactive strategies include the following variables: Prevention Programming, Character Development, Student-Centered Leadership, and School Efficacy. Five items from the teacher survey were averaged to create Prevention Programming. Respondents were asked to indicate the extent of prevention programs utilized in their school, including (a) counseling, social work, psychological or therapeutic activity; (b) involving members of the community in helping the school; (c) services or programs for families or family members; (d) provision of information about the harmfulness of violence, drug use, risky sexual behavior, or about the availability of services directed at students, parents, educators, or community members; and (e) treatment or prevention services for administrators, faculty, or staff (Cronbach’s α = 0.757). 4 Potential responses were have not heard about it (1), does not happen (2), not a major activity (3), an important activity (4), or a very important activity (5).
Character Development, a measure of classroom emphasis, was created by taking the mean of four items from the teacher survey. Teachers identified the extent to which they emphasized the following in their classroom: (a) character development in students, (b) classroom control/management, (c) student interest in learning, and (d) respect for other races and nationalities (Cronbach’s α = 0.801). 5 Possible responses for each of the four items were not at all (1), a little (2), some (3), and very much (4).
Student-Centered Leadership was created by averaging four items from the principal survey. Respondents indicated the priority given to each of the following: (a) know students by name; (b) praise students for a job well-done in academics, sports, and so forth; (c) attend/support student after-school functions/activities; and (d) reprimand students for misconduct (Cronbach’s α = 0.779). 6 This was measured on a scale of very low (1) to very high (5).
The final measure capturing proactive strategies, School Efficacy, was created from 12 items from the teacher survey. Respondents indicated the extent to which they agreed with the following statements: (a) administrators and teachers collaborate toward making the school run effectively; (b) there is little administrator–teacher tension in this school; (c) our principal is a good representative of our school before the superintendent and the board; (d) the principal encourages experimentation in teaching; (e) teacher evaluation is used in improving teacher performance; (f) the principal is aware of and lets staff members and students know when they have done something particularly well; (g) teachers and students can arrange to deviate from the prescribed program of this school; (h) teachers feel free to communicate with the principal; (i) the administration is supportive of teachers; (j) teachers have a say about how this school is run; (k) the principal of this school shares decision-making; (l) teachers and administrators get along well at this school (Cronbach’s α = 0.933). Each item was measured on a 5-point response scale ranging from strongly disagree (1) to strongly agree (5).
Several control variables were also controlled for in the multivariate analyses. At the individual level, Sex was coded as female (0) or male (1) and Race was coded as White (0) or non-White (1). Low Self-Control was calculated by taking the mean of 11 items from the student survey measuring frustration, temper, restlessness, and attention span (Cronbach’s α = 0.91). Parental Attachment was created by taking the mean of 24 survey items asking about the students’ relationships with parents (Cronbach’s α = 0.93). School Attachment was created by taking the average of six survey items asking the respondents to indicate their feelings toward teachers, school, and education (Cronbach’s α = 0.70). Peer Attachment is the mean of six survey items that asked about relationships with close friends (Cronbach’s α = 0.91). Delinquent Peers was created by taking the average of 17 items measuring friends’ involvement in delinquency (Cronbach’s α = 0.91). Access to Illicit Goods was measured by taking the mean of six items asking about the ease of obtaining cigarettes, drugs, alcohol, and weapons at school (Cronbach’s α = 0.88). Finally, the Wave of the study was controlled for in the multivariate analyses. Several school-level characteristics were also controlled for, including Percent Male, Percent Non-White, and Percent Receiving Free or Reduced Price Lunches.
Multivariate Analysis and Results
The distribution of the dependent variable, In-School Delinquency, is highly skewed and violates the assumption of normality associated with the linear model. Thus, Poisson models with overdispersion were estimated using hierarchical linear modeling (HLM 7.0) to account for the error associated with event counts (Osgood, 2000). 7 Again, because the same school might have policies and practices that varied across waves, student-wave cases were nested within school-wave contexts for hierarchical modeling purposes.
To test the research questions, four models were estimated. 8 The results for the multilevel regression analyses are presented in Tables 2 and 3. The intraclass correlation, which was estimated prior to predictor variables being added to the model, was 0.06. This suggests that 6% of the total variance in in-school delinquency occurs between schools. Thus, the remaining 94% of the total variance occurs between students within schools.
School-Level Reactive Strategies.
p ⩽ .05. **p ⩽ .01. ***p ⩽ .001.
School-Level Proactive Strategies.
p ⩽ .05. **p ⩽ .01. ***p ⩽ .001.
Table 2 presents the models that examine school-level reactive policies, which are the traditionally utilized school crime practices. Model 1 presents the results of the school-level reactive policies on in-school delinquency. Only one measure, community service, had a significant relationship with in-school delinquency (b = 0.10, p ⩽ .05). This relationship was positive, indicating in-school delinquency increased as the frequency of the use of community service as a disciplinary tool increased.
Model 2 in Table 2 displays the full reactive school practices model, which added in the individual-level variables. At the school-level, community service remained significant (b = 0.07, p ⩽ .05). The control variable, percent receiving free or reduced price lunches, became significant, although the effect was small (b = −0.00, p ⩽ .05). At the student-level, all of the variables, except Wave 2, were significantly related to in-school delinquency. Of most interest, perceived injustice had a significant positive relationship with in-school delinquency (b = 0.31, p ⩽ .001). Male, non-White, low self-control, delinquent peers, and access to illicit goods had significant positive relationships with in-school delinquency. Parental attachment, school attachment, peer attachment, and Wave 3 all had significant negative effects.
Models in Table 3 explore the impact of school-level proactive policies. Model 1 examines the influence of school-level proactive policies on in-school delinquency, while controlling for the school-level control variables. The results reveal that none of the school-level proactive strategies were significantly related to in-school delinquency. Model 2 presents the full model, which also includes the student-level measure of perceived injustice and the student-level control variables. Again, none of the school-level practices were significantly related to in-school delinquency. At the individual-level, perceived injustice had a significant positive relationship with in-school delinquency (b = 0.31, p ⩽ .001). Being male and non-White, along with having low self-control, delinquent peers, and access to illicit goods were significantly and positively related to in-school delinquency. Parental attachment, school attachment, peer attachment, and Wave 3 all had significant negative relationships with in-school delinquency. 9
The results presented thus far suggest that school discipline practices are unrelated to in-school delinquency. Given that these null effects exist even in models that do not control for individual-level perceived injustice (i.e., Model 1 in Tables 2 and 3), we have little reason to believe that school discipline practices work through individual perceived injustice. In short, results thus far indicate that there are no effects of school practices to be mediated by individual perceptions of justice. Nonetheless, we conducted supplemental analyses to determine whether the school-level policies impact perceptions of injustice. In a model examining the effects of reactive practices and school-level control variables, along with individual-level controls, findings (not shown) indicated that detention was significantly related to perceived injustice (b = −0.02, p ⩽ .05). In other words, schools with teachers that utilized detention as a disciplinary practice more frequently had students with lower mean levels of perceived injustice. A second model examining the effects of proactive practices and school-level control variables, while also controlling for individual-level variables, on perceived injustice indicated that prevention programming (b = − 0.05, p ⩽ .01), and student-centered leadership (b = −0.02, p ⩽ .05), had significant negative relationships with perceived injustice. This suggests that schools that emphasize prevention programming and those that prioritize student-centered leadership had students with lower mean levels of perceived injustice. 10 In short, while three of the 10 discipline practices were significantly related to perceptions of injustice, it does not appear that perceived injustice is mediating the relationships between school practices and in-school delinquency. In fact, our findings reveal that there is little to mediate, with the effects of reactive and proactive practices on student delinquency largely null, even when individual perceptions of injustice and other individual differences are unaccounted for in the models.
Discussion
Schools across the country have spent a great deal of resources on developing strategies to reduce school crime. Thus, it is vital to determine whether the resources are being allocated in the most effective manner. The current study applied a multilevel approach to explore the impact of school-level reactive and proactive discipline management strategies, in addition to student perceptions of injustice, on in-school delinquency. Several unexpected findings emerged from the analyses. We found no evidence that school-level reactive practices and proactive strategies directly impact in-school delinquency. Despite the widespread public support of zero-tolerance policies and ideas from deterrence theory, our results provide no empirical support for the use of harsh punishment in schools. Regarding the nonpunitive or restorative-like reactive strategies, community service was significantly related to in-school delinquency, but this relationship was not in the expected direction, with the use of community service positively associated with in-school delinquency. The proactive strategies were also nonsignificant, which is unexpected given previous research demonstrating that communally organized schools have been found to have less disorder (e.g., Payne, 2008; Payne et al., 2003). There is, however, support across all of our models for the hypothesis that perceived injustice impacts in-school delinquency.
The largely null effects of school reactive and proactive practices were observed in models with and without perceived injustice, suggesting that effects of actual discipline practices do not work through perceptions regarding fairness thereof. Furthermore, it is unclear that the measured school practices do anything to address the other individual-level correlates of in-school delinquency. Our findings indicate that low self-control, associating with delinquent peers, access to illicit goods, and weak attachments to parents, school, and peers increase risk for in-school delinquency. Perhaps the most commonly implemented school discipline practices are ineffective because they do little to systematically address these risk factors.
Of particular interest is the fact that discipline practices seem to have little relation to perceived injustice in the current study. One concern about harsh disciplinary practices is that they may have unintended consequences, such as enhancing students’ perceptions of injustice in school. Yet our findings do not support this claim, as none of the examined disciplinary practices were associated with increased perceptions of injustice among students. While detention, prevention programming, and student-centered leadership were all negatively associated with perceived injustice, these practices were not significantly related to in-school delinquency.
On the contrary, parental attachment, school attachment, peer attachment, delinquent peers, low self-control, and access to illicit goods were all significantly related to perceived injustice. Specifically, students who were more attached to parents, school, and peers reported lower levels of perceived injustice, while students with more delinquent peers, lower self-control, and greater access to illicit goods had higher levels of perceived injustice. These findings suggest that individual differences play a large role in shaping perceptions of injustice and raise questions about the extent to which actual school practices can alter such perceptions for the purposes of reducing in-school delinquency.
While we do not suggest that schools abandon their school-level reactive and proactive strategies, our findings do indicate more attention needs to be spent on understanding and improving student perceptions of injustice. This will require additional research and data sources to more fully unpack how individual student perceptions are formed and how those perceptions can change over time with new experiences. For example, while the current study did control for individual differences in several established correlates of delinquency, we did not have access to information on individual student experiences with the school discipline practices. Therefore, while we can conclude that school-level practices measured in the current study on the whole do not appear to influence perceptions of injustice, it remains unknown whether the quality of a student’s specific interactions with school personnel and personal experiences with school disciplinary practices shape perceptions of injustice.
Nearly 35 years ago, Gottfredson and Gottfredson (1985) suggested that school discipline must be both consistent and predictable to be considered fair. Subsequent research examining the factors that impact student perceptions of school climate suggests that both school-level factors (e.g., school enrollment, faculty turnover, student-teacher ratio, zero-tolerance practices) (e.g., Griffith, 2000; Hirschfield, 2018; Koth, Bradshaw, & Leaf, 2008; Mitchell, Bradshaw, & Leaf, 2010) and student-level factors (e.g., race, gender) (e.g., Koth et al., 2008; McNeely, Nonnemaker, & Blum, 2002) influence perceptions of school climate. As fairness of school rules is often part of school climate, future research can build on this existing literature to determine what factors impact students’ perceptions of injustice and what can be done to reduce it.
The nature and structure of the RSVP data offer several strengths that allowed us to answer the current study’s research questions. First, the nesting of student-level data within schools allowed us to observe the influence of both student- and school-level variables simultaneously. This method is largely absent from the literature examining school-level practices and student perceptions. In addition, the dataset provided direct measures of numerous reactive and proactive strategies, including a range of strategies that varied in punitiveness. This allowed us to examine simultaneously the relative effects of a broad range of school disciplinary approaches.
Despite the strengths of the current study, there are some limitations that must be noted. First, data collected for RSVP began in the spring of 2001. It is likely that high-profile school shootings in recent years, and the widespread social media attention focused on these events, impact students’ perceptions. Thus, student perceptions in today’s society may be influenced by different factors than they were 17 years ago. Second, only schools from Kentucky were sampled, which is a predominantly rural state. Thus, results may not be generalizable to other environmental settings. It is important to note, however, that the sample does include schools in communities that span the urban–rural continuum and that previous studies using these data have found results that are consistent with nationally representative samples (Ousey et al., 2011). While only 43% of parents gave consent to their children to participate, this response rate is consistent with studies that require parental consent (Ellickson & Hawes, 1989; Esbensen et al., 1996; Ousey & Wilcox, 2007).
As school context can change (e.g., a student could change schools or a school could modify their disciplinary practices across waves), the only way to accurately link a student to the correct school environment was to treat the students’ school context for each wave as a unique context. Therefore, the relationships were examined cross-sectionally and we thus cannot fully establish temporal order. In addition, because a primary goal of this study was to examine the relative effects of a range of existing school discipline practices on in-school delinquency, we were unable to randomly assign the school practices. Thus, we cannot be certain that the observed relationships are causal, though the analyses do include a broad range of relevant school- and student-level variables to minimize threats to internal validity. It is also important to note that the measures of reactive and proactive strategies may not accurately represent school policies. To explain, teachers and principals may provide inaccurate reports for various reasons, including lack of knowledge of school policies, biased views of the policies, or intentionally misrepresenting policies to meet social desirability standards. Relatedly, even if teacher and administrator reports of proactive and reactive strategies are accurate, they are potentially different from students’ perceptions of such practices, which could be another reason for the largely null effects of school practices on student delinquency. Finally, it is important to note that while we have found that, generally, school policies and practices do not impact students’ perceived injustice, it is possible that individual student experiences with school discipline could influence perceptions of injustice, as noted above. Feelings of injustice could develop if a student was punished unfairly for their misbehavior or if they were victimized and the school did not take the complaint seriously or resolve the issue. These possible relationships, however, were beyond the scope of our study.
In sum, our findings demonstrate that school-level practices are less influential on in-school delinquency than many may expect, while student perceptions of injustice do negatively influence delinquency. Simply having authority over students does not appear to impact delinquency; rather students’ perceptions of how schools use that authority appear to be most important. What remains unclear is how schools can strengthen students’ perceptions of justice to enhance their legitimacy while also reducing in-school delinquency. Future research should continue to explore the impact of school-level policies and perceived injustice on various types of school delinquency and examine the factors that impact students’ perceptions of injustice.
Footnotes
Acknowledgements
The authors thank Richard R. Clayton, Scott A. Hunt, Michelle Campbell Augustine, Shayne Jones, Kimberly Reeder, Staci Roberts Smith, and Jon Paul Bryan for their contributions to the Rural Substance Abuse and Violence Project, which provides the data analyzed here.
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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the National Institute on Drug Abuse (Grant DA-11317, 1999).