In December 2018, the Trump administration rescinded the Department of Education’s 2014 “Dear Colleague Letter” (DCL), thus returning to local school districts and boards their traditional authority to set discipline policy. Although it was frequently described as “nonbinding guidance,” the DCL was anything but. Instead, the letter advised school districts that they could be found in violation of the Civil Rights Act if students of different groups were disciplined at different rates—even if their rules governing suspensions and expulsions were written and administered fairly. Based on that notion, the Department of Education opened investigations that compelled hundreds of school districts serving millions of students to change their school discipline policies.
The basis for this sweeping federal policy intervention was a set of claims: First, that racial disparities in school discipline—in particular, suspensions and expulsions—were not a function of differences in student behavior. Instead, these disparities were largely driven by adult bias, i.e., by discrimination. Second, that suspensions and expulsions, so-called exclusionary discipline, substantially harm students and fuel a school-to-prison pipeline. And third, that exclusionary discipline can safely be replaced by “restorative” or “positive” methods.
While the DCL is no more, the various claims that Obama administration officials made about school discipline and racial discrimination, including the suspicion cast on public school teachers, are still widely circulated and believed. This is unfortunate, because almost all these claims are based on weak or flawed empirical evidence. As school leaders revisit the rules that they maintain to ensure orderly classrooms and safe learning environments, it is essential that they understand why the federal government’s involvement with local school disciplinary policies was ill-advised—and be guided by better and more rigorous research published after the DCL.
- The most rigorous social science suggests that adult bias plays, at best, a minimal role in disciplinary “disproportionality.” Differences in discipline are driven largely by differences in student behavior, and these differences are driven largely by social and economic factors.
- Recent, robust research has substantially revised downward reasonable estimates of the negative effects that school suspensions have on students.
- There is little basis for claims that “restorative” or “positive” approaches to student misbehavior work, and there is a growing cause for concern that the recent shift away from traditional discipline is doing more harm than good.
In January 2014, then–secretary of education Arne Duncan introduced a package of reforms—spelled out in greater detail in a “Dear Colleague Letter”—that would have dramatic effects in classrooms throughout the United States. The case he made on behalf of the sweeping change in federal policy was succinct and began with an indictment. “Racial discrimination in school discipline is a real problem today,” he said, pointing to data collected by the Department of Education that showed that black students “are more than three times as likely as their white peers to be expelled or suspended.” This disparity in school discipline, he added, “is not caused by differences in children; it’s caused by differences in training, professional development, and discipline policies.” In short: “It is adult behavior that needs to change.”
Duncan said that “the overuse of suspensions and expulsions” has “taken a terrible toll on students, families, schools, and communities,” noting in particular: “Suspended students are less likely to graduate on time—and are more likely to repeat a grade, drop out of school, and become involved in the juvenile justice system.” This “school-to-prison pipeline,” he said, “must be challenged every day.”
Although characterized by discipline reform advocates and the New York Times as “nonbinding guidance,” the DCL was, in fact, the cornerstone of a campaign of federal pressure on school districts to change their disciplinary policies away from traditional discipline—with its emphasis on detentions and suspensions—to “positive” or “restorative” discipline, with its emphasis on student-mediated conflict resolution and healing circles.
The DCL put schools on notice that they could be found guilty of a civil rights violation, and therefore potentially lose federal funding, based on racial disparities in their school discipline statistics. The only way for these investigations to end was for school districts to agree to change their policies. Hence, the federal government directly coerced approximately 400 school districts serving about 10 million students to change their policies, and the threat of investigation prompted hundreds of districts serving millions more students to do the same.
This entire enterprise, in retrospect, was a mistake, beginning with the claim that racial disciplinary disparity is due to adult bias. In reality, the differences are overwhelmingly due to student behavior.
Bias or Behavior: What the Research Shows
“The Color of Discipline,” a 2002 study of which Indiana University’s Russell Skiba was the lead author, has been frequently cited by academics and advocates of discipline reform to support the proposition that, in the words of NAACP official Monique Dixon, “students of color do not misbehave more than their white peers, but are often disciplined for subjective infractions, such as disrespect of authority.” This study has also been widely cited in the academic literature as evidence of similar claims.
However, Skiba’s 2002 study voiced no conclusions about whether black students misbehave more or less than their white peers. The report examined a school district where black students represented 56% of enrollment and 66% of disciplinary referrals and white students represented 42% of enrollment and 33% of disciplinary referrals. The report then examined each of the district’s 32 separate infractions to see whether student race played a statistically significant role in the variance from the mean referral rate. The authors characterized the four infractions—for which the variance from the mean referral rate was of slight positive statistical significance for black students—to be “subjective” (disrespect, excess noise, threat, and loitering) and four infractions—for which the study found the variance from the mean referral rate to be of slight positive statistical significance for white students—to be “objective” (smoking, left without permit, vandalism, and obscene language). The 24 infractions for which race was not a statistically significant factor spanned both “subjective” and “objective” infractions, e.g., “fighting,” “vandalism,” and “drugs/alcohol possession.”
These results may suggest that subjectivity factors into some disciplinary decisions. They do not, however, begin to speak to the question of underlying differences in student behavior by race, or whether teacher subjective judgment is a substantial driver of the aggregate racial disparity in school discipline. If it were, we would expect to see relatively even rates of referral for more serious infractions and pronounced racial differences for more “subjective offenses.” But this is not what the data show.
In 2018, the Education Research Alliance for New Orleans published a study using statewide disciplinary data provided by the Louisiana Department of Education. Louisiana students are split nearly evenly by race: 46% are black and 48% are white. The study showed that black students get suspended relatively more frequently than white students for the “objective” offense of fighting (467,074 vs. 125,606) rather than for the “subjective” offense of disrespecting authority (393,442 vs. 131,529).
Perhaps the most compelling evidence regarding differences by race comes from what students themselves say. According to the Youth Risk and Behavior Surveillance System, a nationally representative survey conducted under the auspices of the Centers for Disease Control and Prevention, black students are substantially more likely than whites to say that they have been in fights at school (15.5% vs. 6.5%), more likely to carry a gun (6.5% vs. 4.1%), and more likely to skip school (9% vs. 4.9%). According to the National Center for Education Statistics, black students are substantially more likely than white students to say that they arrive late to class “sometimes” or “often” (20.9% vs. 12.2%).
To be sure, none of the evidence that, in the aggregate, black students misbehave at higher rates than white students (for “objective” or “subjective” offenses) supports a conclusion that this difference is attributable to race. Rather, the differences are the result of socioeconomic factors that largely influence student behavior. For example, students who come from a single-parent household are twice as likely to get suspended—and black students are nearly three times as likely to come from single-parent households. Similarly, controlling for student poverty or school and community characteristics dramatically shrinks disciplinary disparities by race, suggesting that those factors also play a large role.
A recent study by the Brookings Institution noted that, in Michigan, black students were twice as likely as white students to be the subject of a state investigation into child maltreatment, i.e., parental abuse or neglect. Controlling for other demographic factors, researchers found “extremely large” academic gaps attributable to abuse and neglect. Although the researchers did not attempt to tie abuse and neglect to disciplinary records, it is a common finding that low academic achievement is a strong predictive factor for disciplinary problems.
Education reformers have long accepted the obvious truth that differences in family and community background affect academic achievement, and, after accepting that reality, they ask: How can schools do more to improve academic achievement? However, when it comes to school discipline, some advocates seem to dismiss any connection between student background and classroom behavior.
The existence of substantial behavioral differences should significantly inform school officials’ concerns about disciplinary “disproportionality.” Disciplinary action should be proportional to misbehavior, and if misbehavior is not evenly distributed across groups there is no reason to expect disciplinary referrals to be evenly distributed.
Differences in behavior make it foolish to attribute the totality of racially disparate discipline to adult bias, but they do not eliminate the concern that bias explains some share of the disparity. Responsible policymakers and school leaders need to understand the extent of the role that bias plays in disciplinary disparities.
Are Principals Racially Biased?
A 2011 study (Skiba was the lead author) examined school administrators’ disciplinary decisions, contingent on teacher disciplinary referral, across 344 schools. It found that black students were punished more frequently and severely than white students for the same offenses. The study was rigorous, but its authors noted two significant limitations. First, it did not control for school characteristics, leaving open the possibility that the disparity was due to differences in school policy or circumstance. Infractions like “disruption,” for example, may mean substantively different things in affluent suburban schools from what they mean in lower-income urban schools. Second, it did not control for previous student behavior, leaving open the possibility that the disparity was due to administrators reasonably differentiating response between first-time and repeat offenders. Three years later, a follow-up study (also led by Skiba) of administrators in an unnamed midwestern state controlled for both these factors and found that race ceased to be statistically significant.
In 2017, other researchers conducted a similar inquiry based on data from Arkansas and also found that race ceased to be a statistically significant factor in administrators’ disciplinary decisions after previous student behavior and school characteristics were controlled for. They concluded that the gap is “primarily driven by differences between schools, not differences within schools.” It would appear that, while different schools have different policies that affect aggregate disciplinary disparities, school administrators show no tendency to treat students who are sent to their office differently.
Are Teachers Racially Biased?
Even if principals mete out discipline without regard to race, it would still be possible that racial bias affects decisions about which students are sent by teachers to the principal’s office. This is a vastly more difficult thing to study, given that it is not possible to directly observe the totality of student–teacher interactions and draw objective conclusions about whether individual disciplinary decisions were justified. But researchers have attempted several approaches to shed light on this question.
One approach is to ask teachers about their perceptions of students or their behavior and see whether they respond differently based on student race. A 2011 study examined teacher survey data on perception of student behavior and found no statistically significant difference in perception between black teachers and white teachers.
Another approach is to examine whether there are differences in disciplinary referrals based on teacher–student racial match. Presumably, black teachers would be less racially biased toward black students than white teachers, so if white teachers send black students to the office more frequently than black teachers do, that differential would suggest that teacher bias is part of the explanation.
A 2010 study of 381 classrooms in 21 elementary schools found that “Black students were as likely as White children to receive each of the [disciplinary referral types] in classrooms with White teachers as with Black teachers” and that black male students with black male teachers had the “greatest odds of receiving any type of [referral].”
A 2017 study of disciplinary data from North Carolina similarly found that “[Black] students with greater exposure to Black teachers are more likely to receive exclusionary discipline than are their peers who take fewer classes with Black teachers, in both middle and high school.” The authors hypothesized that this differential could be due to selection bias if black students with a record of behavioral problems were more likely to be subsequently assigned a black teacher. Finding evidence to support this hypothesis and taking it into account, the authors conclude that black students with black teachers are slightly less likely to receive a suspension than black students with white teachers.
The 2017 study suggests that it would not be appropriate to categorically deny that bias plays some role in racial disparities—but not much of one. The authors estimate that a black student with all black teachers would be about 2 percentage points less likely to be suspended than a black student with all white teachers. And the authors note that an indeterminate part of this differential may also be due to students behaving differently in classrooms headed by black or white teachers.
A 2010 study compared elementary school teacher surveys regarding student behavior with their disciplinary referrals for those students. Given that disciplinary action should be proportionate to misbehavior, it would be reasonable to expect a teacher’s assessment of student behavior to fully account for racial disparity in discipline. Notably, incorporating teacher surveys on student behavior accounted for a substantial share of the disparity (it did not eliminate the disparity). The study concluded that “previous work without measures of student behavior grossly overestimated the extent to which racial disparity in school discipline is based on illegitimate factors,” i.e., on racial discrimination.
A 2011 study reached similar conclusions. Without controlling for demographic factors and teacher perception of student behavior, black students were almost three times more likely to be suspended than white students (who were, in turn, about twice as likely to be suspended as Asian students). With controls, black students were only 30% more likely to be suspended than white students (who were about 25% more likely to be suspended than Asian students). On the one hand, the authors stated that “there may be two currents of racial stereotyping in American elementary schools, one that privileges Asian-American students and one that disadvantages African-American students.” On the other hand, they note that “it is possible that our finding of racial disparity in punishment is linked to past behavior, not cultural stereotypes.”
A 2014 study addressed that possibility by using national longitudinal survey data to incorporate prior teachers’ perceptions as well as reports by their mothers of student behavior. After controlling for past behavior, race ceased to be a statistically significant factor in suspensions. The authors concluded that “differences in rates of suspension between racial groups thus appear to be a function of differences in problem behaviors that emerge early in life and that materialize in the classroom.”
Looking at the various studies, one cannot categorically deny that teacher bias plays some role in disciplinary disparities by race. But taken together, these studies suggest that differences in behavior play a major role in disciplinary disparities and that differences between schools play a substantial role; but within schools, principals treat students fairly, and teacher bias plays, at most, a minor role. Policies that aim to achieve racial parity in school suspensions by limiting teacher discretion or subjecting them to “training” on alleged “implicit bias” are based on an incorrect diagnosis of the reasons for the racial disparity.
Do Suspensions Harm Suspended Students?
For many people, the answer to this question would seem to be obvious and incontrovertible. It is not. “The Punishment Gap: School Suspension and Racial Disparities in Achievement” noted that, “despite the growing realization of negative consequences [of school suspensions], there is surprisingly little research able to specify the direct impact of suspension on outcomes such as academic achievement.”
To be sure, the 2014 DCL cites a 2006 study in support of claims that suspensions lead to decreased academic achievement and increased risk of dropping out. That study matched suspended and non-suspended students on a handful of demographic factors and found worse academic outcomes for suspended students. But the author added a caution: “In interpreting these findings, readers should bear in mind that student behavior is a determinant of both achievement and suspensions. Students who follow instruction, focus on their academic work, and observe rules are likely to do well academically and are not likely to be suspended. Thus, suspensions cannot be assumed as the sole determinant of drop-out rates or of low achievement, and the extent to which suspensions contribute to these factors cannot be determined.” In the past few years, researchers have made substantially further headway into understanding the extent to which suspensions contribute to negative outcomes.
Short-Term Academic Effects of Suspensions
Two studies published in 2018 sought to determine the impact of a suspension by controlling for school characteristics, grade characteristics, and student characteristics, including student behavioral history—effectively comparing suspended students with themselves in semesters when they were and were not suspended. In Philadelphia, researchers found that suspensions lowered reading and math achievement standardized test scores by 0.04 and 0.05 standard deviations, respectively, a small but not academically trivial effect. In New York City, researchers found that suspended students were 3% less likely to pass math and 4% less likely to pass reading classes.
Although these studies represented a significant methodological advance over almost all the previous literature, several factors still could have biased the estimates of the effects of suspensions to be higher than they really were.
One limitation: these studies treated the receipt of suspensions as a one-or-zero phenomenon, potentially inflating the effect of “a suspension” by including results from students who received multiple suspensions in the period examined. A 2018 study of a school district in California examined the effect of suspensions in a district that administered quarterly tests to students. It found no statistically significant effect of a single suspension during the quarter in which it occurred but found that multiple suspensions in a single quarter were associated with a substantial 0.18 standard deviation decrease in standardized test scores.
Another limitation: despite their rigorous controls, the studies of New York and Philadelphia could not truly distinguish between the suspension and the behavior that led to it. As the authors of the Philadelphia study note, their research design “cannot determine whether students would experience the same decline in achievement following a misbehavior that, alternatively, was or was not followed by a suspension.”
Two 2017 studies based on statewide data in Arkansas partly remedy this deficiency by examining differences in academic outcomes based on whether students were assigned in-school or out-of-school suspensions. One study finds, perhaps counterintuitively for advocates of discipline reform, that receiving an out-of-school versus an in-school suspension is associated with a statistically significant increase in standardized test scores in reading and math one year after the suspension.
The increases are so small (a 0.003 standard deviation increase in math and a 0.002 in reading) that it would hardly be appropriate to conclude that out-of-school suspensions benefit students. Furthermore, another study based on disciplinary data from Arkansas found that students who received an out-of-school, rather than an in-school, suspension were 4.4% more likely to be held back the following year, a result that could account for the apparent increase in standardized test scores. Taken together, these two studies suggest that some concern is warranted about the short-term effect of suspensions, but they fail to substantiate the concern that suspensions have dramatic negative effects.
The Long-Term Effects of Suspensions
A 2013 study examined Australian student responses to that nation’s Youth in Focus survey. Controlling for the answers to 25 questions, it first found a persistent negative association between student-reported suspensions and academic outcomes such as high school graduation and college entry. But after employing a sensitivity analysis, to try to control for the effect of unobserved variables, their findings became statistically insignificant. They concluded that the relationship between suspensions and long-term academic outcomes was “unlikely to be causal but rather stems from differences in the characteristics of those suspended compared to those not suspended,” such as family welfare history.
A 2018 study looked at the effects of suspensions on U.S. students based on responses to the National Longitudinal Study of Adolescent and Adult Health. The author compared students who had never been suspended with students who reported having not been suspended in Year A but who were then suspended in Year B. After matching those groups based on the 60 variables, the study found that suspended students were less likely than matched non-suspended students to earn a high school diploma or a bachelor’s degree and more likely to have been arrested.
Both studies take their data on suspensions from student reports, not from official school records. That is a problem because, as Joseph Robinson-Cimpian has demonstrated, a small percentage of “mischievous responders,” who lie on surveys, can “lead to inaccurate conclusions that substantively affect research, policy, and public discourse regarding a variety of disparities.”
Nevertheless, based on research to date, school leaders have reasonable cause to be somewhat concerned about the effects of suspensions on misbehaving students. But discipline policy should be established and implemented with the interests of all students, those who misbehave as well as those who behave properly, in mind.
Suspensions and Non-Suspended Students
The Obama administration’s Department of Education bluntly declared: “Suspensions don’t work—for schools, teachers, or students.” In doing so, it cited a 2012 study by Russell Skiba that noted that “there is no evidence that exclusionary school discipline has a beneficial effect on student behavior or school climate.” This was true, but in 2012, there were no academic studies that remotely approached a proper evaluation of this issue. Today, there still aren’t.
Does Discipline Reform Work?
The key question facing policymakers and school leaders today is not whether suspensions “work” but whether the alternative—reducing suspensions and moving toward the “restorative justice” method—helps or hurts students. In the past few years, evidence regarding the effects of discipline reforms has started to mount, and it strongly suggests that discipline reform, on net, has done more harm than good.
The empirical examinations of the effects of discipline reform with the least troubling findings were conducted by the University of Chicago Consortium on School Research on Chicago Public Schools. A 2015 study examined the effects of requiring central office approval for suspensions of longer than five days and eliminating mandatory 10-day suspensions for the most severe offenses. It found no effects on academics, a slight positive effect on attendance, and negative effects on student perception of order and teacher perception of violence. A 2018 study of Chicago Public Schools examined the effects of a modest decrease in the frequency with which severe offenses resulted in out-of-school suspensions, from 93% to 84% (those students often received in-school suspensions instead). It found a very small but statistically significant positive effect on academics, a positive effect on attendance, and no overall effect on school climate. Taken together, school officials should consider the findings from Chicago as evidence that modest and incremental efforts to decrease out-of-school suspensions—and replace them with in-school suspensions—may be pursued without substantial cost or benefit. But the discipline reforms adopted by many school districts in recent years have been far more aggressive.
In the 2012–13 school year, the School District of Philadelphia issued a ban on suspensions for “conduct” offenses, such as profanity and refusal to follow school rules, which was not fully implemented across all district schools. But a 2018 study found alarming results: after three years, discipline reform had reduced academic achievement by 3% in math and nearly 7% in reading. In a perverse irony, African-American students spent more days out of school on suspension because, even as suspensions for “conduct” offenses declined, suspensions for serious misbehavior increased. And truancy, which had been falling before the suspension ban, rose from about 25% to more than 40%.
The Los Angeles Unified School District also implemented a full suspension ban for “willful defiance” infractions but, unlike Philadelphia, with full fidelity across all schools. Two 2018 studies examined the effects. One controlled for student characteristics and found no effect on reading but a huge harm to math achievement: students at the 50th percentile before a ban would be at the 32nd percentile three years later. Another examined school-level growth in other California districts and found substantial negative effects for academic growth in Oakland, San Francisco, and Los Angeles.
Alternatives to “Exclusionary” Discipline
Discipline reformers frequently advance Restorative Justice (RJ) and Positive Behavioral Interventions & Supports (PBIS) as alternatives to traditional discipline. RJ emphasizes nonpunitive conflict resolution, such as student-mediated discussions intended to repair harm caused by student misbehavior and attempts to get to the “root cause” of the behavior.
In 2018, the Rand Corporation published a randomized control trial of RJ in Pittsburgh Public Schools for school years 2015–16 and 2016–17. With suspensions falling across the school district, 22 schools were chosen to pilot RJ and were compared with 22 schools that did not. The results were mixed. Teachers reported that schools were safer, that the professional culture and morale among the staff improved, and that their classroom management abilities were strengthened. Students disagreed. They said that their teachers’ ability to manage the classroom deteriorated, and they reported deteriorated relations between their peers in classrooms. Most troubling, math achievement deteriorated by a substantial 0.15 standard deviation decline for black students. Students in middle school saw a 0.105 standard deviation decline.
While RJ focuses on student-mediated conflict resolution dialogues and healing circles, PBIS tries to preempt conflict and misbehavior by “teaching behavioral expectations in the same manner as any core curriculum subject.” Teachers receive training in emphasizing “preferred behaviors” and then teach them to students. Educators create posters regarding these preferred behaviors—for example, explaining that part of “Respect Property” is to “flush the toilet after use” and “use equipment as it was designed.” Frequently, PBIS involves making classroom behavior into a game, with students receiving points for good behavior, losing points for bad behavior, and eventually cashing in those points for prizes.
A 2012 paper published in Pediatrics on the effects of PBIS in 37 elementary schools found that teacher ratings of students’ ability to control their emotions, prosocial behavior, concentration, and disruptive behaviors all improved. However, PBIS had no effect on school suspensions. The study, a randomized control trial, provides reasonable grounds for optimism for PBIS in elementary schools. It does not provide grounds for the belief that PBIS could replace suspensions (or that it could work in middle or high schools).
The 2012 study examined PBIS as a stand-alone intervention in an era before PBIS was coupled with a pressure to lower suspensions. In 2017, the Wisconsin Institute for Law and Liberty (WILL) used seven years of data from more than 2,000 Wisconsin schools to determine the effects of PBIS. Its research found that reading and math proficiency scores were lower in Wisconsin schools that implemented PBIS than in schools that did not, with the negative effect strongest in suburban and rural schools. WILL’s research was not a randomized control trial, so it does not allow for a clear conclusion that PBIS caused these negative effects. However, the results should still give policymakers pause: while the 2012 study shows that PBIS can work in elementary schools, WILL’s study suggests that policymakers should not expect it to work.
What Do Teachers Think?
Surveys of classroom teachers should give policymakers and school officials even greater pause. Teachers’ unions have surveyed their members in 11 school districts on the effects of their district’s efforts to lower suspensions and implement PBIS and/or RJ. In Denver, only 23% of teachers believe the new approach to be effective; in Madison, Wisconsin, only 13% believe that it improved behavior; in Charleston, South Carolina, about 13% of teachers believe that it works in their classrooms; in Oklahoma City, only 11% of teachers believe that a more thorough implementation of PBIS would make them more effective teachers. On the other hand, in Philadelphia more than 80% of teachers said that suspensions are essential to send a message to parents about the seriousness of their child’s behavior, to ensure a safe school, and to encourage other students to follow the rules.
Classroom Disorder: News from the Front Lines
There are very good reasons for public policy not being made exclusively on the basis of opinion surveys, academic journals, or anecdotal evidence. Nevertheless, a growing body of reportage on the negative effects of discipline reform in major districts across the country cannot be ignored. Prior to implementing discipline reform, crime had been falling in the Charlotte-Mecklenburg Schools (CMS) in North Carolina. But after implementing reform, crime in CMS started steadily increasing. Similarly, in Durham, North Carolina, reported crime rose in the wake of discipline reform, and assaults on teachers increased by 56% from the 2015–16 to 2016–17 school year. In St. Paul, Minnesota, the local district attorney described the rise in school violence and assaults on teachers as a “public health crisis.”
A district attorney ordered the school district to reverse course on discipline reform in Syracuse, New York, following an outbreak of school violence. In Buffalo, 81% of teachers believe that their school administrators are underreporting disciplinary problems.
In Broward County, Florida, the local news reported that the district’s discipline reforms created a “culture of leniency” that allows students like the Parkland shooter to commit multiple crimes in school with minimal consequences. After DeKalb County, Georgia, implemented discipline reform, school climate steadily declined.
In Fresno, California, teachers voiced concerns that their district’s discipline reforms had thrown their classrooms into disorder. In Fresno’s McLane High School, 70 of 85 teachers signed a petition protesting the climate created by these policies. One teacher, Michael Clark, told the Fresno Bee: “A student can say ‘f--- you’ and we’re told that’s just his personality... How many times do you get kicked until you say, ‘OK, I’m not going to do this anymore’?”58 Clark and his colleagues were successful in their effort to restore traditional discipline, and, according to Clark, the school snapped back into order and stability.
The School-to-Prison Pipeline Revisited
The “school-to-prison pipeline”—the causal linking of exclusionary school discipline to incarceration—has been discussed and denounced for decades and is now part of the conventional wisdom. Yet, as a 2009 study observed: “Despite scholarly consideration of the factors that may facilitate entry into the ‘pipeline’ for minority youth, there is surprisingly little empirical work verifying that it exists. Often, the relationship between school behavior and justice system contact is simply assumed due to similarity in factors that predict racial disparity in both institutions. For example, [Johanna] Wald and [Daniel J.] Losen flatly state that ‘the racial disparities within the two systems are so similar—and so glaring—that it becomes impossible not to connect them.’”
Correlation is not causation. Student race is correlated with school suspensions, but that correlation weakens significantly when variables such as poverty or single-parent families are taken into account. School suspensions, in turn, are correlated (as are other behaviors, such as truancy) with increased chances of dropping out, which are, in turn, correlated with an increased risk of incarceration. Yet the omitted—and difficult to observe—variable hovering behind these correlations is student behavior. And increasingly sophisticated empirical efforts to control for student behavior suggest that the diagnoses animating the discipline reform movement—that adult bias accounts for a large share of the disciplinary disparities and that suspensions harm students—are largely mistaken.
The Trump administration withdrew the 2014 Dear Colleague Letter in December 2018. The letter announcing the rescission declared: “States and local school districts play a primary role in establishing educational policy, including how to handle specific instances of student misconduct and discipline, and in ensuring that classroom teachers have the support they need to implement appropriate discipline policies.”
Local education leaders now must decide what to do with the authority that Washington has restored to them. Several recommendations can be drawn from the research literature, many of which are at odds with contemporary conventional wisdom on school discipline:
Do not target disciplinary disparities. If adult bias played a substantial role in disciplinary disparities, district leaders could feel comfortable promoting policies that limited teacher discretion. But given that the research now suggests that bias plays, at most, a minor role, such policies are almost certain to backfire by preventing teachers from exercising the judgments they need to make, day in and day out. Given aggregate differences in student behavior, discipline policy should not aim to achieve racial parity in school discipline. Rather, disciplinary policy should focus on promoting safe classrooms and respectful school environments through rules that are fairly and consistently applied to all students.
Focus on improving, not decreasing, suspensions. Most of the utility of suspensions likely stems from their role as a consequence of misbehavior within a consistent system of rules and consequences. When parents could be expected to provide moral reinforcement at home during a suspension, they may have been intrinsically useful. But in communities and neighborhoods where out-of-school suspensions leave troublemaking students to their own devices, they may lead toward further misbehavior. Some school districts, such as New York City Public Schools and Miami-Dade Public Schools, have off-site centers so that students can be both off-campus for the duration of their suspension but also in an academic setting rather than on the streets.
Respect the rights of students who want to learn. The entirety of the discipline reform conversation has focused on students who break the rules. Students who are well behaved are, at best, an afterthought to which some passing lip service is paid or are left out of the conversation entirely. Discipline policy must balance the rights of all students lest, as appears to have happened in districts that have been rigorously evaluated, discipline reform provides no benefit to misbehaving students while harming the educational prospects of those who want to learn.
Do not expect restorative justice to work. There is no shortage of reformers who argue that RJ is a more effective alternative to exclusionary discipline—provided, of course, that the school district provides proper training and teachers buy in to the idea. The Pittsburgh study suggests that even with training, and even with enthusiastic teachers, RJ can harm students’ academic performance and their perceptions of school climate. In school districts where such training or buy-in is unlikely to materialize, the odds of success seem slimmer still. While teachers may adopt RJ methods if they so choose, school boards and superintendents should not impose it on schools and expect it to work.
Reject the false choice between discipline reform and “zero tolerance.” Discipline reform advocates frequently frame their policies as an effort to end “zero tolerance,” i.e., policies that automatically dole out certain penalties for certain offenses. But discipline reform is less a solution to zero tolerance than it is the flip side of the same coin. Zero tolerance was adopted by school districts largely as a reactive measure to pressure from federal and state legislators and bureaucrats, and it diminished teacher discretion by ordering certain consequences. Discipline reform was adopted by school districts largely as a reactive measure to pressure from federal and state legislators and bureaucrats, and it diminished teacher discretion by inhibiting certain consequences. Rather than have the pendulum swing back yet again, policymakers should craft policies that allow teachers discretion to exercise their best judgment on order and safety.
Listen to teachers and students. The overwhelming majority of school districts that have implemented discipline reform in the past five years have not stopped to ask their students or teachers whether it is working; in most school districts that have asked, teachers say that it is not working. District leaders should use surveys and open-ended anonymous-response questionnaires to gauge whether students feel safe and supported and whether teachers believe that the reforms have helped or hurt.
Roll back failed policies. If teachers say that these reforms have failed, they should be rolled back. When faced with evidence of failure, advocates of school discipline reform frequently say that the policy would have worked had it been implemented properly. There is no evidence to support this claim. School leaders should value the perspective of teachers, grounded in the daily classroom realities.
Give students priority over statistics. Rolling back discipline reforms where they have been implemented would almost certainly lead to an increase in suspensions. This would be followed, predictably, by news stories filled with inflammatory expressions from angry activists accusing school district leaders of rolling back civil rights, even returning to the era of Jim Crow. However difficult it would be, school leaders should stand up for safe schools and respectful classrooms where students can learn.