Behind the Eight Ball: The Effects of Race and Number of Infractions on the Severity of Exclusionary Discipline Sanctions Issued in Secondary School

Abstract

African American and Hispanic students receive more punitive school discipline than White students even when students of color commit similar infractions as Whites. Similarly, students with a disability status are more likely to experience harsher discipline in schools compared to their counterparts without a disability label. This study examines whether these discrepancies are a result of a difference in the number of infractions students of different racial/ethnic groups and disability categories commit. Using secondary educational data from a state educational agency in the United States, we demonstrate that African American and Hispanic students and students with an emotional behavioral disorder status receive more severe sanctions than White students and students without a disability label at their first discipline encounter. This racial disparity in discipline severity continues through six sanctions and is eliminated at the 13th sanction. The disability disparity in discipline severity dissipates after 10 sanctions for students with emotional behavioral disorder and intellectual disability. Implications for school personnel and future directions are discussed.

Keywords

racial disparities disability disparities suspension exclusionary discipline students of color

Exclusionary discipline is a pressing national concern (U.S. Department of Education, 2018). Students who are suspended or expelled from school are more likely to suffer from lost instructional time, lower academic achievement, and grade retention than students who do not receive exclusionary discipline sanctions (Losen et al., 2015; Smith et al., 2020). Suspended students also show increased risk of early and continued contact with the juvenile justice system, inciting a cycle of recidivism in which these students become more likely to commit serious transgressions, receive more severe school discipline sanctions, and be arrested in the future (Horner et al., 2005; Liberman et al., 2014).

African American students are referred to the office, suspended, and expelled up to 3 times as often as White students (Bal et al., 2017). The disparate impact of school discipline for African American students is observed in both traditional K–12 school settings and alternative educational placement settings (Booker & Mitchell, 2011; Fowler, 2010). Nationally, African American male and female students received 25% and 14% of out of school suspensions (OSSs), respectively, but each accounted for 8% of total enrollment (U.S. Department of Education, 2018). In Texas, African American students comprised 14% of school-age children but accounted for 24% of students enrolled in disciplinary alternative educational placements (DAEPs) in 2005 (Reyes, 2006). These rates parallel the racial disparities observed in juvenile justice alternative educational placements (JJAEPs) and in-school arrests for African American students (Nicholson-Crotty et al., 2009).

The school discipline outlook for Hispanic students is less clear. Some studies find that Hispanic students are disciplined at rates proportional to their makeup in the general school population (Bryant, 2013). Others have found that Hispanic students are punished more harshly than their White non-Hispanic peers (Peguero & Shekarkhar, 2011) but less harshly than African American students (Shollenberger, 2015). A recent report from a large national data set demonstrates that these disparities may be better understood when gender is considered (U.S. Department of Education, 2018). Hispanic male students accounted for 15% of OSSs but 13% of the total male student enrollment, whereas Hispanic female students accounted for 6% of OSSs and 13% of the total female student enrollment. Regarding juvenile arrest rates, Hispanic youth do not differ significantly from African American or White youth (Brame et al., 2014; Piquero et al., 2014).

Several studies have demonstrated that school discipline disparities may also be related to factors outside of students’ behavior. When students of color and White students commit the same infraction, students of color receive more punitive discipline sanctions (Bal et al., 2017; Morris & Perry, 2016). For example, African American students are more likely to receive punitive disciplinary sanctions such as court referrals or police notification for misbehavior in the classroom (Morris & Perry, 2016). Prior research indicates that teachers perceive African American students as being more disruptive and dangerous than students of other races in classroom settings (Bates & Glick, 2013; Francis, 2012; Oates, 2009). Teachers’ perceptions of Hispanic students are less clear; some studies suggest they are not perceived differently than White students, and others find they are (Hughes et al., 2005; Zimmerman et al., 1995).

Studies have also indicated that disability status may be related to risk for school discipline (Bal et al., 2017; Mendoza et al., 2020; U.S. Department of Education, 2018). Students who have learning disabilities (LDs), emotional behavioral disorders (EDs), other health impairment, and intellectual disabilities (IDs) are more likely to receive exclusionary discipline than students without a disability or low-incidence disability such as autism, deaf-blindness, developmental disability, hearing impairments, orthopedic impairments, traumatic brain injury, visual impairments (Bal et al., 2017; Sullivan et al., 2014). Notably, 26% of students with a disability nationally received one or more OSSs despite these students accounting for only 12% of the general student population (U.S. Department of Education, 2018). What is more problematic is how the intersection of disability status and race affect these suspension rates, with 33.8% of African American males and 22.5% African American females with a disability receiving at least one suspension compared to 16.2% and 7.3% of White males and females with a disability, respectively (Losen et al., 2015). This disparity was also found with Hispanic students; 23.2% of Hispanic males and 12.1% of Hispanic females with disabilities at the elementary and secondary levels received at least one suspension (Losen et al., 2015). Similarly, other studies examining race and disability with exclusionary discipline have found that there is no significantly greater risk for students within these groups (Morgan et al., 2019). The conflicting findings in the literature suggest individual factors must be closely examined among students who have received disciplinary sanctions.

Although it is well-established that disparities in discipline do exist, the presence of bias in judgments about students’ behavior continues to be unclear. The behaviors observed for some disability categories (in particular, ED) may inherently lend to bias because of culturally bound expectations of misbehavior (Sullivan, 2017). Furthermore, observed misbehaviors may be affected by a variety of factors that are beyond the student’s control; for example, family/home environment, trauma, and classroom behavior management can affect the expectations a student has about acceptable behaviors (Sullivan, 2017). In addition, some studies suggest that school faculty are applying harsher punishments to students of color than to White students from the beginning of their relationships (Kunesh & Noltemeyer, 2015; Staats, 2014). Culturally deficient thinking where minority students are thought of as misbehaving because they undervalue education and live in dysfunctional situations is believed to fuel punitive discipline practices (Blake et al., 2016). Similarly, stereotypes of students of color as being loud and disruptive in classroom settings may contribute to biases against African American and Hispanic students from a very early stage in the student–teacher relationship (Blake et al., 2016; Staats, 2014). At the same time, others suggest that disparities occur over time (Okonofua & Eberhart, 2015). There is evidence to suggest that disparities in suspensions increase over time for students who have been suspended before, especially for African American students (Okonofua & Eberhart, 2015). The cycle of discipline escalation begs the question of whether discipline disparities exist at the issuance of the first discipline sanction or develop over time. We propose that bias in the severity of discipline received might be present if disparities exist at the issuance of the first consequence versus a pattern where disparity increases as the number of sanctions issued increases.

The purpose of this study was to both confirm previous findings of racial and disability disparities in school discipline and examine whether the number of student discipline infractions can explain these disparities in the severity of discipline sanctions issued by administrators. We extend prior research documenting racial/ethnic and disability disparities in school discipline outcomes by investigating how discipline disparities may develop. We explore the point in time that these disparities emerge in discipline decision-making, which is important for policy and intervention considerations. For example, if students of color and students with disabilities receive numerous harsh discipline sanctions after committing more punishable infractions over the course of the year, the implications for how to best intervene are different from students who are punished more harshly at the first infraction.

To assess when racial and disability disparities in school discipline emerge, we examine the following research questions: (a) Do African American students with and without disabilities receive more severe disciplinary sanctions than Hispanic and White students with and without disabilities for their first discipline infraction? (b) Do Hispanic students with and without disabilities receive more severe discipline than White students with and without disabilities for their first discipline infraction? and (c) Do these disparities in discipline sanction severity hold even when controlling for the number of infractions students commit in a year?

Method

Participants

Data were drawn from the Texas Education Agency’s (TEA) Public Education Information Management System (PEIMS) and Texas Juvenile Probation Commission’s (TJPC; now the Texas Juvenile Justice Department) CASEWORKER system database. The PEIMS and TJPC CASEWORKER databases were merged with a match rate of 87% (Fabelo et al., 2011). All enrolled students in the 2000–2001, 2001–2002, and 2002–2003 seventh grade cohorts in Texas public schools were included in the analysis. Data from students’ sixth grade years serve as control variables. We chose to begin measurement at seventh grade because discipline rates are relatively low during elementary school, and middle school tends to be where rates of discipline peak (Losen et al., 2015; Welsh et al., 1999). These students were followed for at least 6 years—up to their grades’ expected graduation or their exit from the Texas public school system (i.e., dropout or transfer). In total, 928,940 students were included in the database. The students were 51% male, 14% African American, 39% Hispanic, and 43% White and largely representative of students in the state of Texas in the years the data represent, with one exception. In the 2015–2016 school year, the racial/ethnic composition of Texas schools was 13% African American, 52% Hispanic, and 29% White (TEA, 2016).

To allow for a more focused analysis on the relationship between race and discipline severity, we limit our analyses to African American, Hispanic, and White students (97% of the total sample). We also exclude individuals with a primary disability other than an ID, ED, or LD to allow for a more targeted examination between the linkage between disability and discipline. Because data are available on an annual basis, the student/year serves as the unit of analysis. The rate of missingness was 0.08%.

Measures

Number of discretionary discipline infractions

The Texas education code classifies behavioral infractions into two categories of disciplinary action: discretionary and mandatory. Mandatory disciplinary infractions are severe in nature (e.g., assault, arson, bringing drugs or weapons to school) and have a specific disciplinary sanction associated with them. Discretionary discipline infractions are less severe in their nature (e.g., fighting, chewing gum, dress code violations), and the sanction associated with them is chosen by school administrators. As school personnel are responsible for choosing the consequence for discretionary infractions, bias in the severity of consequences could be observed. As administrators have no discretion on what punishment to assign a student who commits a mandatory violation, students who had at least one mandatory violation in the year were excluded from the analyses. Because we focused on the student/year, students who have a mandatory offense in 1 year are included in subsequent years allowing the receipt of a mandatory offense in the prior year to serve as a predictor in the models. To minimize the effect of outliers, the number of discipline events was truncated to 25, which influences less than 1.5% of observations.

Student race/ethnicity

Student race/ethnicity was based on parent self-report collected from the TEA’s PEIMS database. Parents could identify their children in one of the following categories: Hispanic, American Indian/Alaska Native, Asian, African American or African American, Native Hawaiian/other Pacific Islander, or White. For this study, student race/ethnicity was dummy-coded into three categories representing 97% of students: White, Hispanic, and African American. Students who were identified as other races/ethnicities were excluded from analyses given their low representation in the sample. White students were the comparison group for all analyses.

Disability

The disability status of students was coded as dummy variables based on students’ receipt of special education services in their primary educational placement: ID, ED, LD, autism, traumatic brain injury, or a physical disability (Fabelo et al., 2011). Given high incidence rates of ID, ED, and LD, we focus only on students with these disabilities.

Discipline sanctions

To create the dependent variable, discipline outcomes were ordered based on severity, from least severe (in-school suspension [ISS]) to most severe (JJAEP/expulsion). ISS was chosen as the least severe discipline sanction because the student remains housed within their home school, and ISS is usually issued for shorter periods of time relative to other sanctions. OSS was chosen as the next most severe sanction. Although the student is removed from their home school, the duration of OSS is also shorter than other sanctions, limited to 3 school days in Texas (TEA, 2003). DAEP was selected as the next most severe sanction because its duration is longer than ISS or OSS (e.g., greater than 3 school days), and the student is placed in an alternative school setting external to their home campus. Finally, JJAEP was coded as the most severe discipline sentence because it is associated with the juvenile justice system and can result from being suspended from a DAEP. Expulsion was included with JJAEP in the same discipline severity category because of the small number of students that received either of these sanctions and the disruptive nature of these sanctions to a student’s life. In addition, as the decision to expel a student or place them in a JJAEP is often based on the availability of a JJAEP in the county where the student resides in Texas, only large counties in Texas are required by law to provide a JJAEP, leaving students in smaller counties to be expelled without school placement (Fabelo et al., 2011).

Grade level

Student grade level was controlled because middle school students tend to receive more discipline than older high school students, but older students tend to engage in more misconduct than younger students (Welsh et al., 1999).

Free/reduced lunch status

Students’ free/reduced lunch status was used as a proxy of students’ socioeconomic status (SES) and served as a control variable given the association between low SES and the display of disruptive/delinquent behaviors worthy of discipline (Costello et al., 2001; Gregory et al., 2010).

Academic controls

Because of evidence suggesting that academic achievement can decrease the risk of delinquency and school misbehavior (Hoffman et al., 2013), students’ history of being retained and failure on state standardized achievement tests were controlled. We note any failures of any section (e.g., mathematics, science) of the state assessments in the student’s past, beginning with the sixth grade.

Behavioral history

History of misbehavior was controlled by measuring the number of discipline violations (discretionary and mandatory) students received in the previous year. Prior juvenile justice contact was controlled with measures of students’ previous record with the TJPC (dummy-coded as yes or no).

School-level variables

School-level variables have been found to predict discipline severity, so the school’s Title I status, the type of community in which schools were located (nonmetro adjacent, rural-urban served as the base category), the school’s student–teacher ratio, and county per capita income were controlled for in analyses as well (Payne & Welch, 2010; Skiba et al., 2014).

Analytic Strategy

To examine whether racial differences and differences based on student’s disability exist in the severity of discipline sanctions secondary school students receive, and if these racial and disability disparities vary based upon the number of infractions students commit in a given year, an ordered probit regression was employed. Because students who are never disciplined are likely to differ from disciplined students both behaviorally and academically, a Heckman (1979) selection model was used to remove selection bias surrounding who is initially disciplined. The Heckman selection model was carried out using Stata 13.1 and the “heckoprobit” command (StataCorp, 2013).

Heckman’s technique applied to continuous dependent variables accounts for selection bias using a two-stage procedure. We applied De Luca and Perotti’s (2011) extension of the Heckman method to ordered probit for our analyses. For our initial selection model (first stage of Heckman model), we determine the probability of being disciplined at all using a binomial probit regression. In the second stage, we model the severity of discipline a student receives while eliminating the selection bias from the first stage only for those who are disciplined using an ordered probit regression (De Luca & Perotti, 2011). As such, the number of observations in the second stage of the Heckman method is smaller than in the first stage, and the descriptive statistics differ from those in the selection model to reflect the means of those students who were disciplined in the year.

Control variables related to student academic achievement and behavioral history and school-level indicators were included in all analyses to prevent spurious findings. To account for the nested structure of the data (e.g., students nested in schools), we used robust clustered standard errors to estimate the variance related to school effects and to reduce the probability of Type I error (Kam & Franzese, 2007). We cluster standard errors on the campus and individual level.

To assess degree to which number of discipline severity varies by race and disability status, we tested two- and three-way interaction effects: Discipline Encounters × Race, Discipline Encounters × Disability, Disability × Race, and Disability × Race × Discipline Encounters. Given that large sample sizes tend to produce smaller standard errors that increase the likelihood of significant findings on substantively small coefficients, we used a more stringent p value of p < .001 to identify significant effects in our models. Using a standard of p < .001, none of the three-way interaction effects were significant. In the interest of parsimony, we reran the analyses with only two-way interaction effects: Discipline Encounters × Race and Discipline Encounters × Disability.

Results

Descriptive Statistics

Descriptive statistics are presented in Table 1. The descriptive statistics in Stage 1 includes summaries for those variables that are in the first stage of the Heckman model. Stage 1 descriptive statistics are based on variables from the full sample of 4,586,366 student/years, which are used to predict having at least one discipline event in the school year. Stage 2 descriptive statistics include the variables that are used in the second stage of the Heckman model focusing on only those who were disciplined in the year (N = 1,137,947 student/years). In a single given school year, 25% of students received at least one discipline sanction.

Table 1.

Descriptive Statistics.

Variables	Stage 1 %/M (SD)	Stage 2 %/M (SD)
Disciplined	24.81%	—
Student race: Hispanic	40.59%	46.31%
Student race: African American	14.77%	22.07%
Intellectual disability	0.87%	0.61%
Emotional behavioral disorder	1.21%	2.59%
Learning disability	9.02%	14.32%
Discipline encounters	—	3.10 (3.31)
Hispanic*# of discipline encounters	—	1.48 (2.82)
African American*# of discipline encounters	—	0.75 (2.17)
Intellectual disability*# of discipline encounters	—	0.02 (0.36)
Emotional behavioral disorder*# of discipline encounters	—	0.11 (0.91)
Learning disability*# of discipline encounters	—	0.51 (1.85)
Discretionary discipline last year (dummy)	24.02%	—
Mandatory discipline last year (dummy)	0.92%	—
Discretionary discipline last year (count)	—	2.02 (3.33)
Mandatory discipline last year (count)	—	0.04 (0.29)
Male	50.38%	62.14%
Seventh grade	18.12%	18.46%
Eighth grade	18.20%	20.10%
Ninth grade	19.58%	23.33%
Tenth grade	16.19%	16.38%
Eleventh grade	14.23%	12.18%
Retained last year	—	9.10%
Free/reduced-price lunch	44.92%	58.28%
Past juvenile justice encounter	4.75%	10.04%
Past failure of state test	37.80%	51.50%
Years behind expected grade	0.24 (0.51)	—
Title 1 school	41.11%	45.25%
Student–teacher ratio	14.92 (2.55)	14.88 (2.37)
County income per capita	34,035.80 (8,829.09)	—
Nonmetro adjacent county	10.17%	10.23%
Rural county	4.18%	3.64%

Across all races, the most severe discipline allocation in a single year was ISS for 60.84% of all disciplined students, followed by OSS for 27.93% of disciplined students, DAEP for 10.94% of disciplined students, and JJAEP/expulsion for less than 1% (0.30%) of disciplined students. ISS was the most common punishment received by students of each race. However, significant racial disparities were found, χ²(6) = 32,000, p < .001. Results from an ordered logit was consistent with chi-square analyses in that African American students received the most severe discipline (z = 148.65, p < .001), followed by Hispanic students (z = 86.76, p < .001). Most White students (70.10%) never moved to a level of severity higher than ISS, whereas just over 60.23% of Hispanic students stayed at this level and 48.83% of African American students stayed at this level. When looking at OSS, 39.08% of African American students moved up the severity scale to OSS compared to just 28.51% of Hispanic students and 19.31% of White students. The percentage of students whose most severe discipline allocation was DAEP or JJAEP/Expulsion was substantively similar across races.

African American and Hispanic students were found to have more discipline encounters than White students. Specifically, African American students received more than twice as many discipline sanctions (M = 1.3 per year, SD = 2.7) on average than White students, M = 0.5 per year, SD = 1.6; t(2,724,796) = 2,800, p < .001. Hispanic students averaged 0.9 (SD = 2.3) discipline encounters in 1 school year more than White students, M = 0.5 per year, SD = 1.6; t(3,909,144) = 2,100, p < .001, and lower than African American students, M = 1.3 per year, SD = 2.7; t(2,538,786) = 100, p < .001.

Discipline Involvement

Although not the theoretical focus of the analysis, the Heckman ordered probit provides information on who is disciplined in addition to the severity of punishment issued. Results of the Stage 1 analyses are presented in Table 2. Unlike logit regression, binomial probit regression does not readily convert into odds ratios. As such, the interpretation presented here follows the approach suggested by Long (1997), where continuous variables are held at their mean and categorical variables held at their mode, and the change in probability of discipline associated with a one-unit change in the independent variable is reported in the column labeled “1-Unit Change.” With the modal categories set to one and the continuous variables held at their mean, we are left with a “typical” student who has a probability of 0.133 of being disciplined in a given year. In the last column, we add the percent change in probability the one-unit shift represents.

Table 2.

Predictors of School Discipline Involvement in School Year.

Variables	Coefficient	SE	1-Unit change	% Increase
Student race: Hispanic	0.139***	0.005	0.032	24.31%
Student race: African American	0.306***	0.007	0.077	58.00%
Intellectual disability	−0.226***	0.012	−0.043	−32.07%
Emotional behavioral disorder	0.368***	0.009	0.096	71.90%
Learning disability	0.213***	0.003	0.051	38.60%
Discretionary discipline last year (dummy)	0.995***	0.003	0.321	241.01%
Mandatory discipline last year (dummy)	0.286***	0.009	0.071	53.60%
Male	0.252***	0.002	0.047	35.19%
Seventh grade	0.418***	0.009	0.111	83.38%
Eighth grade	0.421***	0.009	0.112	84.14%
Ninth grade	0.323***	0.010	0.058	43.28%
Tenth grade	0.259***	0.008	0.064	48.00%
Eleventh grade	0.126***	0.006	0.029	21.81%
Free/reduced-price lunch	0.184***	0.003	0.044	32.84%
Past juvenile justice encounter	0.269***	0.007	0.067	50.13%
Past failure of state test	0.292***	0.003	0.073	54.89%
Years behind expected grade	0.080***	0.003	0.018	13.30%
Title 1 school	−0.068***	0.009	−0.014	−10.63%
Student–teacher ratio	−0.005**	0.002	−0.001	−0.71%
County income per capita	0.000**	0.000	0.000	0.00%
Nonmetro adjacent county	0.035***	0.010	0.008	5.84%
Rural county	−0.062***	0.015	−0.013	−9.76%
Constant	−1.741***	0.029	—	—
Base probability			0.133	—

Note. n = 4,586,366. Standard errors clustered on campus/year and student.

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

Consistent with other findings, African American students were more likely to be disciplined than other students (z = 46.37, p < .001), with an increase in probability of 0.077 or a 58% increase from the base probability. Hispanic students were also more likely to be disciplined than other students (z = 28.00, p < .001), with an increased probability of 0.032 or a 24% increase over the base probability.

While students with an ID see 0.043 lower probabilities for discipline (a 32% decline, z = 19.45, p < .001), students with an ED (0.096, 72%; z = 41.93, p < .001) or a LD (0.051, 39%, z = 65.34, p < .001) both have higher probabilities of discipline than do students without a disability.

Also consistent with other findings, male students were more likely to face exclusionary discipline (z = 107.69, p < .001), in this case with an increased probability of 0.047 or a 35% increase over the base probability. Not surprisingly, those students who were disciplined in the previous year were more likely to be disciplined in the current year. Those with a discretionary discipline event in the previous year saw an increase of 0.321 or 241% over the base probability (z = 312.59, p < .001).

The results of the second-stage ordered probit are found in Table 3. The significant value for Rho indicates that the Heckman model is necessary to account for selection biases, χ²(1) = 1,059.28, p < .001. Because the dependent variable severity of discipline has four possible outcomes (ISS, OSS, DAEP, and JJAEP/expulsion), interpretation of results can be complicated. To aid interpretation, we utilize two figures described below, and in Table 4, we display the effect of a one-unit change of each variable on the predicted probability for each outcome.

Table 3.

Predictors of Severity of Discipline Sanction in School Year.

Variables	Coefficient	SE
Student race: Hispanic	0.141***	0.010
Student race: African American	0.361***	0.013
Intellectual disability	0.369***	0.020
Emotional behavioral disorder	0.355***	0.013
Learning disability	0.042***	0.006
Discipline encounters	0.140***	0.002
Hispanic*# of discipline encounters	−0.022***	0.002
African American*# of discipline encounters	−0.027***	0.003
Intellectual disability*# of discipline encounters	−0.036***	0.004
Emotional behavioral disorder*# of discipline encounters	−0.036***	0.002
Learning disability*# of discipline encounters	−0.006***	0.001
Discretionary discipline last year (count)	0.015***	0.001
Mandatory discipline last year (count)	0.046***	0.005
Male	0.042***	0.004
Seventh grade	0.187***	0.017
Eighth grade	0.193***	0.017
Ninth grade	0.092***	0.018
Tenth grade	0.012	0.014
Eleventh grade	−0.016	0.011
Retained last year	0.122***	0.009
Free/reduced-price lunch	0.033***	0.005
Past juvenile justice encounter	0.298***	0.006
Past failure of state test	0.010*	0.004
Title 1 school	0.078***	0.013
Student–teacher ratio	0.021***	0.003
Nonmetro adjacent county	−0.150***	0.018
Rural county	−0.061*	0.030
OSS cut	1.238***	0.043
DAEP cut	2.326***	0.045
JJAEP/expulsion cut	4.179***	0.047
Rho	−0.127***	0.007

Note. n = 1,137,947. Standard errors clustered on campus/year and student. OSS = out of school suspension; DAEP = disciplinary alternative educational placement; JJAEP = juvenile justice alternative educational placement.

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

Table 4.

Effect of a One-Unit Change on Probabilities of Most Severe Punishment.

Variables	ISS	OSS	DAEP	JJAEP/exp.
Student race: Hispanic	−0.0541	0.0319	0.0219	0.0003
Student race: African American	−0.1413	0.0755	0.0645	0.0012
Intellectual disability	−0.1444	0.0769	0.0663	0.0013
Emotional behavioral disorder	−0.1389	0.0745	0.0633	0.0012
Learning disability	−0.0160	0.0098	0.0061	0.0001
Discipline encounters	−0.0434	0.0323	0.0110	0.0001
Hispanic*# of discipline encounters	0.0084	−0.0052	−0.0032	0.0000
African American*# of discipline encounters	0.0103	−0.0063	−0.0039	−0.0001
Intellectual disability*# of discipline encounters	0.0133	−0.0084	−0.0049	−0.0001
Emotional behavioral disorder*# of discipline encounters	0.0133	−0.0083	−0.0049	−0.0001
Learning disability*# of discipline encounters	0.0024	−0.0015	−0.0009	0.0000
Discretionary discipline last year (count)	−0.0055	0.0035	0.0020	0.0000
Mandatory discipline last year (count)	−0.0174	0.0107	0.0067	0.0001
Male	−0.0157	0.0099	0.0058	0.0001
Seventh grade	−0.0721	0.0418	0.0299	0.0005
Eighth grade	−0.0744	0.0430	0.0309	0.0005
Ninth grade	−0.0345	0.0214	0.0129	0.0002
Tenth grade	−0.0045	0.0028	0.0017	0.0000
Eleventh grade	0.0059	−0.0037	−0.0022	0.0000
Retained last year	−0.0467	0.0278	0.0187	0.0003
Free/reduced-price lunch	−0.0123	0.0077	0.0045	0.0001
Past juvenile justice encounter	−0.1160	0.0640	0.0512	0.0009
Past failure of state test	−0.0036	0.0022	0.0013	0.0000
Title 1 school	−0.0293	0.0182	0.0110	0.0001
Student–teacher ratio	−0.0078	0.0048	0.0029	0.0000
Nonmetro adjacent county	0.0546	−0.0354	−0.0189	−0.0002
Rural county	0.0228	−0.0144	−0.0083	−0.0001
Base probability	0.6358	0.2886	0.0751	0.0005

Note. Significant (p < .05) effects bolded. ISS = in-school suspension; OSS = out of school suspension; DAEP = disciplinary alternative educational placement; JJAEP = juvenile justice alternative educational placement; exp. = expulsion.

Figure 1 displays the effect of each subsequent discipline event on increasing the severity of discipline interacted with race, though not in the expected way. In Figure 1, we display the probabilities of ISS and OSS because these are the most common sanctions students received. However, it is worth noting a student with 10 discipline encounters still has a predicted probability of at least 0.62 of receiving one of these punishments. As shown in Figure 1, the probability of receiving ISS decreases and the probability of receiving OSS increases as the number of disciplinary infractions increase for all students. For the first five infractions in a school year, all students are more likely to receive ISS rather than OSS; however, Hispanic (z = 13.55, p < .001) and African American students (z = 27.65, p < .001) are less likely to receive ISS than are White students for a single discipline infraction.

Figure 1.

Probability of discipline severity by number of discipline events and race/ethnicity.

At the 6th–10th discipline events, the probability of receiving OSS is the roughly same for all students. While White students begin with a lower likelihood of getting more severe punishments, each additional discipline event contributes more to the likelihood of a severe punishment for White students than students of color as shown by the Race/Ethnicity × Number of Discipline Infractions Results (vs. African American students, z = 10.86, p < .001; vs. Hispanic students, z = 10.03, p < .001). Because of this reverse trend in the severity of discipline by race, as White students receive more discipline encounters, the predicted severity of their punishment (i.e., their likelihood of receiving the more severe OSS sanction) eventually mirrors that of the predicted severity of students of color. However, such equalization for students of color would require massive amounts of discipline—roughly six for Hispanics and 13 for African Americans (calculated by dividing the race/ethnicity dummy coefficients by their corresponding interaction coefficient for Race/Ethnicity × Number of Discipline Infractions, .141/−.022 for Hispanics and .361/−.027 for African American students).

The relationship between discipline, disability, and punishment severity are complicated. Figure 2 shows discipline severity with each discipline encounter for students with disabilities. What becomes immediately apparent in the graphic and from the main effects in the ordered probit regression is that the difference in severity for those with a LD, and students without a disability is substantively meaningless despite being statistically significant (z = 6.70, p < .001); students with an ID (z = 18.05, p < .001) and those with an ED (z = 28.40, p < .001) face much higher severity than students without disabilities. Like race, though, students with disability face lower marginal effects from each discipline encounter as shown by the Disability × Number of Discipline Infractions Results (vs. ED students, z = 15.56, p < .001; vs. ID students, z = 8.64, p < .001; vs. LD students, z = 5.17, p < .001); however, it takes approximately 10 discipline encounters for students with an ID or an ED to have lower severity punishments than nondisabled students (calculated by dividing the disability dummy coefficient by the corresponding coefficient for the interaction of Disability × Number of Discipline Infractions, .369/−.036 students with an ID and .355/−.036 for students with an ED)

Figure 2.

Probability of discipline severity by number of discipline events and disability.

Table 4 places the results of the ordered probit in context, depicting the effect of a one-unit change among the variables on the various disciplinary outcomes using Long’s (1997) method which was described above. Although some of the effects appear small, it is important to consider the results in context with the last row, base probabilities; many results have small absolute changes, but when compared to the small base values, they become more substantive. For example, the relationship with being African American on the probability of receiving placement in a DAEP (0.06455 increase) appears small; however, this represents an 86% increase on the base probability of 0.0751 for receiving DAEP placement.

Discussion

We investigated whether the number of infractions a student commits in a given year accounts for racial disparities in the issuing of school discipline sanctions. Our findings contribute to research-supported assertions of systemic discrimination in school discipline (Skiba et al., 2002, 2011, 2014). Skiba and colleagues (2002, 2011, 2014) have demonstrated that when the severity of infraction does not differ between White and non-White students, racially/ethnically diverse students are disciplined more severely. Our findings support this argument; African American and Hispanic students are punished more harshly than White students at the same number of infractions in a school year even after controlling for a variety of student and school-level characteristics. The severity of the first punishment in a school year is twice as high for African American students as for White students, and it requires more than 13 discipline events for this inequality to balance. A balance in the severity of sanctions occurs at the sixth discipline event for Hispanic and White students, at the 13th discipline event for African American and White students, and at the 10th disciplinary event for students with an ID or an ED.

Although a balance eventually occurs, it is problematic that more than six discipline events for Hispanic students and 13 discipline events for African American students are issued before racial disparities in sanction severity subside. First, only 1.32% of students in our study had 10 or more discretionary discipline events in a single school year, so few students ever reach a threshold for “equality” in discipline sanctions. Second, it is likely that when students accumulate 13 discipline events in 1 school year, they are approaching risk of receiving more severe levels of school discipline (DAEP, JJAEP, or expulsion). Students with this rate of discipline are likely facing a slew of academic problems, such as obtaining reputations as troublemakers, missing an excessive number of school days, sensing discrimination, and gaining a distaste for their school and for education in general (Losen & Skiba, 2010).

These findings make sense in light of racially based reputation biases that are found in classrooms (Ferguson, 2003; Steele, 1997) and judicial settings (Bodenhausen, 1988; Leiber et al., 2007),where by African American students and offenders are perceived as more dangerous and menacing regardless of the nature of their offense. Recently, Kunesh and Noltemeyer (2015) found direct evidence that teachers-in-training perceived African American students’ actions as more serious and more stable than those of White students. To our knowledge, no similar studies have been conducted with Hispanic youth. However, students of color’s increased receipt of school discipline may be a result of educator reputational bias such that educators are more likely to severely discipline students perceived as troublemakers irrespective of the offense committed. This bias may be especially salient for African American students, whose teachers perceive them as demonstrating worse behavior than White or Hispanic students (Wright, 2015).

Because stereotypes of African American and Hispanic students are often negative, these students are not given the additional allowances to correct their misdeeds; as discipline events accumulate, their behaviors are responded to more punitively (Okonofua & Eberhart, 2015). The findings of the present study support this notion. If implicit racial bias influences administrators to perceive students of color as not deserving the “benefit of the doubt,” then they may react more harshly to the infractions of these students and punish them more harshly initially. Teachers and administrators may not understand these students, especially African American students (Wright, 2015). As a result, they are more likely to rely on stereotypes to interpret their behavior (Blake et al., 2016).

In contrast, White students are less likely to be perceived by teachers as troublemakers and therefore afforded more opportunities for behavioral remediation in the absence of school discipline. As the results of this study show, White students receive less harsh sanctions at their first discipline event. However, this racial disparity disappears after 13 discipline events. We suspect that as students engage in more behaviors that warrant discipline, and as administrators have more time to learn about students individually, racial stereotypes may be less salient.

This study supports the extant research that suggests students with an ED or LD in particular have higher risk of disciplinary sanctions than their nondisabled counterparts. This is concerning for academic and behavioral outcomes for these students, as these students already have difficulty with behavior and their academics. Exclusionary discipline can likely exacerbate the existing difficulties associated with their disability statuses by removing these students from their learning environments altogether (Sullivan et al., 2014).

Limitations and Directions for Future Research

Although this study adds to the literature on racial and disability disparities in school discipline, it is not without limitations. First, this study considers students who have at least one recorded discretionary infraction in a single school year. Students may have unrecorded discipline events prior to their first recorded event that influenced decisions about their punishment. The inability to examine differences in low-level sanctions is a limitation of this study in itself, but low-level sanctions might additionally contribute to differences in recorded punishments.

It is possible that some students had a mandatory punishment discipline event prior to their first discretionary discipline event in a given year, which could bias results about discretionary offenses. Because we are concerned with variations in punishment, it was crucial to remove those whose punishment is a matter of law. However, it is important to understand how committing more severe offenses prior to discretionary offenses influences the choice of discretionary punishments. Future studies should explore differences in the application of mandatory actions (e.g., a fight may be considered a code of conduct violation [discretionary] or as assault [mandatory], and this choice may be related to a student’s race).

Another limitation results from the definition of discretionary discipline infractions. Discretionary discipline infractions can vary in severity (e.g., a dress code violation is much less severe than participating in a fight at school). Unfortunately, with the current data, we are unable to control for such differences in the severity of the actual infraction. Results should be interpreted with the caveat that students of color and students with disabilities may be receiving harsher disciplinary actions sooner because the types of discretionary infractions they commit are more severe. However, as Skiba and his colleagues (2002, 2011, 2014) note, the severity of misbehavior between African American and White students is largely similar.

As mentioned above, the discipline severity disparity seemed to dissipate past the OSS level of discipline. We suspected that this is because there is less room for systemic discrimination in the allocation of the most severe discipline categories due to stricter guidelines and rules for placement in DAEP and JJAEP settings in Texas. Future research should examine why rates for DAEP, JJAEP, and expulsion are similar across races by examining school codes of conduct and legal procedures for issuing these types of discipline sanctions (Fenning & Rose, 2007).

Although this study was not a direct test of the effect of a racialized reputation bias on racially disparate discipline outcomes, our results are consistent with such a theory. Future research should examine the effect of teacher and administrator bias directly on individual discipline outcomes. We were also unable to control for the actual behavior that prompted the discipline sanction. Many variables used in analysis were dichotomous and measured with a single item, so results should be interpreted cautiously. In addition, JJAEP and expulsion were combined into one discipline category and thus may be confounded. Finally, all participants included in the study were from Texas public schools. Although Texas is racially and socioeconomically diverse, caution should be taken when generalizing these findings to other states.

Overall, this study demonstrates that African American, and to a lesser extent Hispanic, students are punished more harshly that their White peers at their first behavioral infraction and that students with ED and ID are at greater risk of being assigned punitive disciplinary sanctions than their nondisabled peers and peers in other special education disability categories. Although biases were not directly assessed in the current study, this is consistent with the theory that bias may be at play in creating disparate rates of school discipline for students with disabilities and students of color. Educators have an important role to play in ensuring that all students, regardless of their race/ethnicity, and disability are treated in a fair and just way by their school systems. These findings suggest that school systems might be most effective in meeting this goal by working to reduce disparities in the initial sanction.

Footnotes

Acknowledgements

The authors gratefully acknowledge the use of these data. The opinions, findings, views, and conclusions or recommendations expressed in this publication are those of the authors and should not be attributed to the Texas Education Research Center, the funders, or supporting organizations mentioned herein, including The University of Texas, the State of Texas, or Office of Juvenile Justice and Delinquency Prevention .

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 project was supported by Grant 2012-JF-FX-4064 awarded by the Office of Juvenile Justice and Delinquency Prevention, Office of Justice Programs, U.S. Department of Justice. The research presented here utilizes confidential data from the State of Texas supplied by the Texas Education Research Center at The University of Texas at Austin.

References

Bal

Betters-Bubon

Fish

R. E.

(2017). A multilevel analysis of statewide disproportionality in exclusionary discipline and the identification of emotional behavior disorder. Education and Urban Society, 51, 1–22.

Bates

L. A.

Glick

(2013). Does it matter if teachers and schools match the student? Racial and ethnic disparities in problem behaviors. Social Science Research, 42, 1180–1190.

Blake

J. J.

Smith

D. A.

Marchbanks

M. P.

Wood

Seibert

(2016). Does student-teacher racial/ethnic match impact Black students’ discipline risk? In Skiba

Mediratta

Karega

(Eds.), Inequality in school discipline: Research and practice to reduce disparities (pp. 79–98). Palgrave.

Bodenhausen

G. V.

(1988). Stereotypic biases in social decision making and memory: Testing process models of stereotype use. Journal of Personality and Social Psychology, 55, 726–737.

Booker

Mitchell

(2011). Patterns in recidivism and discretionary placement in disciplinary alternative education: The impact of gender, ethnicity, age, and special education status. Education and Treatment of Children, 34, 193–208.

Brame

Bushway

S. D.

Paternoster

Turner

M. G.

(2014). Demographic patterns of cumulative arrest prevalence by ages 18 and 23. Crime & Delinquency, 60(3), 471–486.

Bryant

(2013). Addressing the problem of unfair school discipline for boys of color. Center for Law and Social Policy.

Costello

E. J.

Keeler

Angold

(2001). Poverty, race/ethnicity, and psychiatric disorder: A study of rural children. American Journal of Public Health, 91, 1494–1498.

De Luca

Perotti

. (2011). Estimation of ordered response models with sample selection. The Stata Journal, 11(2), 213–239.

10.

Fabelo

Thompson

M. D.

Plotkin

Carmichael

Marchbanks

M. P.

Booth

E. A.

(2011). Breaking schools’ rules: A statewide study of how school discipline relates to students’ success and juvenile justice involvement. Council of State Governments Justice Center.

11.

Fenning

Rose

(2007). Overrepresentation of African American students in exclusionary discipline: The role of school policy. Urban Education, 42(6), 536–559.

12.

Ferguson

R. F.

(2003). Teachers’ perceptions and expectations and the Black-White test score gap. Urban Education, 38(4), 460–507.

13.

Fowler

(2010). Texas’ school-to-prison pipeline: School expulsion, the path from lockout to dropout. Texas Appleseed.

14.

Francis

D. V.

(2012). Sugar and spice and everything nice? Teacher perceptions of Black girls in the classroom. The Review of Black Political Economy, 39, 311–320.

15.

Gregory

Skiba

Noguera

(2010). The achievement gap and the discipline gap two sides of the same coin? Educational Researcher, 39, 59–68.

16.

Heckman

J. J.

(1979). Sample selection bias as a specification error. Econometrica, 47(1), 153–161.

17.

Hoffman

J. P.

Erickson

Spence

(2013). Modeling the association between academic achievement and delinquency: An application of interactional theory. Criminology, 51, 629–660.

18.

Horner

R. H.

Sugai

Todd

Lewis-Palmer

(2005). School-wide positive behavior support. In Bambara

L. M.

Kern

(Eds.), Individualized supports for students with problem behaviors: Designing positive behavior plans (pp. 359–390). Guildford.

19.

Hughes

J. N.

Gleason

Zhang

(2005). Relationship influences on teachers’ perceptions of academic competence in academically at-risk minority and majority first grade students. Journal of School Psychology, 43, 303–320.

20.

Kam

C. D.

Franzese

R. J.

(2007). Modeling and interpreting interactive hypotheses in regression analysis. University of Michigan Press.

21.

Kunesh

C. E.

Noltemeyer

(2015). Understanding disciplinary disproportionality: Stereotypes shape pre-service teachers’ beliefs about Black boys’ behavior. Urban Education, 54(4), 471–498. https://doi.org/10.1177/0042085915623337

22.

Leiber

M. J.

Johnson

Fox

Lacks

(2007). Differentiating among racial/ethnic groups and its implications for understanding juvenile justice decision making. Journal of Criminal Justice, 35(5), 471–484.

23.

Liberman

A. M.

Kirk

D. S.

Kim

(2014). Labeling effects of first juvenile arrests: Secondary deviance and secondary sanctioning. Criminology, 52(3), 345–370. https://doi.org/10.1111/1745-9125.12039

24.

Long

J. S.

(1997). Regression models for categorical and limited dependent variables. Sage.

25.

Losen

D. J.

Hodson

Keith

I. I.

Michael

Morrison

Belway

(2015). Are we closing the school discipline gap? K-12 racial disparities in school discipline. University of California, Los Angeles.

26.

Losen

D. J.

Skiba

(2010). Suspended education: Urban middle schools in crisis. Civil Rights Project/Proyecto Derechos Civiles.

27.

Mendoza

Blake

J. J.

Marchbanks

Ragan

(2020). Race, gender, disability and risk for juvenile justice contact. The Journal of Special Education, 53(4), 226–235.

28.

Morgan

P. L.

Farkas

Hillemeir

M. M.

Wang

Mandel

DeJarnett

Macquga

(2019). Are students with disabilities suspended more frequently than otherwise similar students without disabilities? Journal of School Psychology, 72, 1–13. https://doi.org/10.1016/j.jsp.2018.11.001

29.

Morris

E. W.

Perry

B. L.

(2016). The punishment gap: School suspension and racial disparities in achievement. Social Problems, 63, 68–86. https://doi.org/10.1093/socpro/spv026

30.

Nicholson-Crotty

Birchmeier

Valentine

(2009). Exploring the impact of school discipline on racial disproportion in the juvenile justice system. Social Science Quarterly, 90, 1003–1018.

31.

Oates

G. L.

(2009). An empirical test of five prominent explanations for the Black–White academic performance gap. Social Psychology of Education, 12, 415–441.

32.

Okonofua

Eberhardt

(2015). Two strikes: Race and the disciplining of young students. Psychological Science, 26(5), 617–624.

33.

Payne

A. A.

Welch

(2010). Modeling the effects of racial threat on punitive and restorative school discipline practices. Criminology, 48, 1019–1062.

34.

Peguero

A. A.

Shekarkhar

(2011). Latino/a student misbehavior and school punishment. Hispanic Journal of Behavioral Sciences, 33(1), 54–70. https://doi.org/10.1177/0739986310388021

35.

Piquero

A. R.

Schubert

C. A.

Brame

(2014). Comparing official and self-report records of offending across gender and race/ethnicity in a longitudinal study of serious youthful offenders. Journal of Research in Crime and Delinquency, 51(4), 526–556. https://doi.org/10.1177/0022427813520445

36.

Reyes

(2006). The criminalization of student discipline programs and adolescent behavior. Journal of Civil Rights and Economic Development, 21, 73–110.

37.

Shollenberger

T. L.

(2015). Racial disparities in school suspension and subsequent outcomes: Evidence from the National Longitudinal Survey of Youth 1997. In Losen

(Ed.), Closing the school discipline gap: Research to practice (pp. 31–43). Teachers’ Press.

38.

Skiba

R. J.

Chung

Trachok

Baker

Sheya

Hughes

(2014). Parsing disciplinary disproportionality contributions of infraction, student, and school characteristics to out-of-school suspension and expulsion. American Educational Research Journal, 51, 640–670.

39.

Skiba

R. J.

Horner

Chung

Rausch

M. K.

May

Tobin

(2011). Race is not neutral: A national investigation of African American and Latino disproportionality in school discipline. School Psychology Review, 40, 85–107.

40.

Skiba

R. J.

Michael

Nardo

Peterson

(2002). The color of discipline: Sources of racial and gender disproportionality in school punishment. The Urban Review, 34, 317–342.

41.

Smith

Ortiz

Blake

J. J.

Marchbanks

M. P.

Unni

Peguero

(2020). Tipping point: Effect of the number of in-school suspensions on academic failure. Contemporary School Psychology, 44, 379–389.

42.

Staats

(2014). Implicit racial bias and school discipline disparities. Kirwan Institute for the Study of Race and Ethnicity.

43.

StataCorp. (2013). Stata statistical software: Release 13.

44.

Steele

C. M.

(1997). A threat in the air: How stereotypes shape intellectual identity and performance. American Psychologist, 52(6), 613–629.

45.

Sullivan

A. L.

(2017). Wading through quicksand: Making sense of minority disproportionality in identification of emotional behavior disorder. Behavioral Disorders, 43(1), 244–252. https://doi.org/10.1177/0198742917732360

46.

Sullivan

A. L.

Van Norman

E. R.

Klingbeil

D. A.

(2014). Exclusionary discipline of students with disabilities: Student and school characteristics predicting suspension. Remedial and Special Education, 35(4), 199–210. https://doi.org/10.1177/0741932513519825

47.

Texas Education Agency. (2003). School districts, subchapter CC, commissioner’s rules concerning school facilities. http://ritter.tea.state.tx.us/rules/tac/chapter061/ch61cc.html

48.

Texas Education Agency. (2016). 2015-2016 student enrollment reports. https://rptsvr1.tea.texas.gov/adhocrpt/adste.html

49.

U.S. Department of Education. (2018). 2015-2016 Civil rights data collection: School climate and safety. https://www2.ed.gov/about/offices/list/ocr/docs/school-climate-and-safety.pdf

50.

Welsh

W. N.

Greene

J. R.

Jenkins

(1999). School disorder: The influence of individual, institutional, and community factors. Criminology, 37, 73–116.

51.

Wright

A. C.

(2015). Teachers’ perceptions of students’ disruptive behavior: The effect of racial congruence and consequences for school suspension [Unpublished manuscript, Department of Economics, University of California].

52.

Zimmerman

R. S.

Khoury

Vega

Gil

Warheit

(1995). Teacher and parent perceptions of behavior problems among a sample of African American, Hispanic, and non-Hispanic White students. American Journal of Community Psychology, 23, 181–197.