Abstract

Methods

Procedure

Questionnaires are administered during school hours. About 10 days before the administration, the students are given flyers explaining the study. In addition, their parents are sent letters explaining the study, which give them the opportunity to choose not to have their child participate, if they wish. Questionnaires are administered by local project representatives and their assistants, following standardized procedures detailed in a project instruction manual. The questionnaires are group administered in classrooms during a normal class period whenever possible; however, circumstances in some schools require the use of larger group administrations. Students' data are collected via self-administered machine-readable questionnaires. Questionnaire response rates averaged 86%. Absence on the day of data collection was the primary reason that students were missed; it is estimated that less than one percent of students refused to complete the questionnaire.

In order to simplify presentation, data are shown for only 10^th graders. The decision to focus on 10^th graders avoids the loss of most dropouts (who tend to leave in the final two years of high school), yet captures more substance use behavior than would be available from the 8^th graders. Monitoring the Future was designed to provide data representative of the nation of the whole, not data on racial and ethnic or gender differences. Accordingly, no special effort was made to over-sample students in any of the subgroups. Because a number of the racial and ethnic subgroups examined here are a relatively small proportion of the total population, their numbers in the annual samples are also relatively small. Therefore, in an effort to increase the numbers of cases for analysis, we combined data from 2001 to 2005. The combined samples include data from approximately 74,000 10^th graders, (see Table 1). For the trend analyses we combined data into three 5 year intervals: 1991-95, 1996-2000, and 2001-2005.

Table 1

Sample Distribution by Race and Ethnicity and Gender (2001-2005 data combined)^1,*

The statistical significance of the differences in the experience of exclusionary school discipline between the various ethnic groups is a function of sample size, percentage size, and design effects. Accordingly, all variance estimates presented are adjusted for the sampling design (see note to Table 1). In light of the large number of racial and ethnic and gender subgroups that we examine, however, it would be unwieldy to specify significance levels for each comparison. Further, given the relatively large numbers of cases that we use in these analyses, many of the findings may reach traditional levels of “statistical” significance (i.e., p < .05) and yet be of little substantive significance. Recognizing this possibility, we treat as significant only those differences that equal or exceed conventional standards for statistical significance (i.e., p < .01), and we focus our discussion on the differences that we judge to be both statistically and substantively important.

Results

Racial and Ethnic Differences in Zero-Tolerance Policy Violations

Table 2 presents data on racial and ethnic differences in in-school alcohol use, drug use, and gun possession – the behaviors for which school-based zero-tolerance policies were designed. The data indicate that Hispanic students are significantly (p < .01) more likely than White students to have used alcohol or other drugs at school and that Black and American Indian girls are more likely than White girls to have used alcohol. The data also indicate that Black, Hispanic, and American Indian students of both genders are more likely than White students to have carried a gun to school. Overall, however, the data suggest that racial and ethnic differences in the percentages of students who engage in these behaviors are relatively small.

Table 2

Percent of U.S. 10^th graders Committing Zero-Tolerance Related School Behaviors by Race and Ethnicity and Gender (2001-2005 data combined)

Racial and Ethnic Differences in School Discipline

Despite the fact that there are not large racial and ethnic differences in the prevalence of zero-tolerance related behaviors (see Table 2), the data presented in Table 3 reveal important racial and ethnic differences in school discipline. For example, American Indian, Black, and Hispanic students are consistently more likely than White youth to receive school discipline and Asian American youth are consistently less likely than all other groups of youth to be disciplined in school. Interestingly, racial and ethnic differences in minor disciplinary measures – being sent to the office or detained after school – are relatively small compared to the much larger differences in the harsher forms of discipline – suspension and expulsion. This pattern is particularly evident among Blacks. More specifically, although Black boys' and girls' rates of being sent to the office or detained are roughly comparable to those of other racial and ethnic groups (especially Hispanics and American Indians), they are significantly (p < .01) more likely than the other racial and ethnic groups to have been suspended or expelled. For example, approximately 56% of Black boys have been suspended or expelled compared to only 19% to 43% of boys in the other groups. Similarly, roughly 43% of Black girls have been suspended or expelled compared to only 7% to 26% of girls in the other racial and ethnic subgroups.

Table 3

Percent of U.S. 10^th graders Experiencing School Discipline by Race and Ethnicity and Gender (2001-2005 data combined)

Within racial and ethnic subgroups, boys are consistently more likely than girls of the same racial or ethnic group to have experienced school discipline. Looking at race and gender simultaneously, however, the data reveal important race and ethnicity by gender differences. Specifically, all groups of boys, with the exception of Asian American boys, are sent to the office or detained for misbehavior at higher rates than all groups of girls, across grade levels. For suspensions and expulsions, however, rates are highest among Black boys, followed first by American Indian and Hispanic boys, and then by Black girls. White boys, whose rates of suspension and expulsion are similar to those of Hispanic and American Indian girls, have rates well below those of Black girls. Finally, Asian American boys have the lowest rates of any group of boys, followed by White girls and then Asian American girls.

Racial and Ethnic Differences in School Discipline Trends, 1991-2005

In light of the findings of substantial racial and ethnic differences in the experience of school discipline, the next question to which we turn our attention is whether over time these patterns have increased, decreased, or remained largely unchanged. Figure 1 shows racial and ethnic differences in the percent of students who have experienced school discipline from 1991 to 2005, separately by gender. The data in the first panel of the figure show that Black, Hispanic, and American Indian students have been more likely than White and Asian American youth to be sent to the office or detained after school, and that this pattern has existed from at least the early 1990s to the present. The data further suggest that although there was a slight decline in the percentage of students who were sent to the office between 1996 and 2005 among most racial and ethnic groups, Blacks rates' remained constant. The figure further highlights the fact that boys, irrespective of their racial and ethnic identification, are more likely than girls to be sent to the office or detained after school. The only exception to this general conclusion is Asian American boys, whose frequencies of being sent to the office are generally comparable to those of Black, Hispanic, and American Indian girls (but well above the frequencies for Asian American girls).

Figure 1

Trends over Time in School Discipline

The second panel of Figure 1 shows trends in suspension and expulsion. Consistent with the prevalence data presented in Table 3, these data indicate that suspensions and expulsions have consistently been highest among Blacks, at an intermediate level among American Indian and Hispanic students, lower among White students, and lowest for Asian American students. Perhaps the most striking finding is the fact that the prevalence of suspension and expulsion among Blacks increased between 1991 and 2005 while for the other racial and ethnic groups suspension and expulsion rates increased slightly between 1991 and 2000 and then decreased slightly between 2000 and 2005. Looking across gender, suspension and expulsion rates are clearly highest among Black boys (i.e., more than 50%), followed by American Indian and Hispanic boys and Black girls. The other racial and ethnic and gender groups all have suspension and expulsion rates that are less than 30%.

Racial and Ethnic Differences in School Discipline Controlling for Socio-demographic Factors

Table 4 shows the results of a series of logistic regression analyses that examine the likelihood that Black, Hispanic, Asian American, and American Indian students have experienced school discipline relative to their White counterparts. The table shows the coefficients for race and ethnicity alone (i.e., Model 1) and for race and ethnicity adjusted for key socio-demographic factors that past research suggests relate to both race and ethnicity and school disciplinary practices (i.e., Model 2). The demographic factors controlled in Model 2 include family structure, parental education (as a proxy for socio-economic status), and the region and urbanicity of the community in which students live. The regression coefficients indicate whether non-White students are more likely (i.e., value > 1), less likely (i.e., value < 1), or equally likely (value = 1), as White students to have experienced school discipline. If racial and ethnic differences in school discipline are the result of racial and ethnic differences in socio-demographic factors, the coefficients for the non-White students should approach a value of 1 when the socio-demographic variables are controlled. Given that White youth comprise the vast majority of the sample, the coefficients in Model 2 reveal whether non-White students' frequency of experiencing school discipline would be more comparable to White students' if the non-White students were as likely as White youth to live in two parent families, to have highly educated parents, to live in non-urban areas, and so forth. The table also shows the strength of the relationship between race and ethnicity and the school discipline measures (i.e., the gamma coefficient), alone (Model 1) and controlling for the socio-demographic variables (Model 2).

Table 4

Logistic Regression Analyses (Odds Ratios for School Discipline, Unadjusted (Model 1) and Adjusted Family Structure, Parental Education, Urbanicity and Region (Model 2), by Race and ethnicity and Gender (2001-2005 data combined)

Looking at Model 1 for being sent to the office or detained after school, the data reveal that Black, Hispanic, and American Indian 10^th graders are significantly (p. < .01) more likely than White 10^th graders to have been sent to the office or detained after school and that Asian American students are significantly (p. < .01) less likely than White students to have been sent to the office or detained. The magnitudes of these differences vary by race and ethnicity and by gender, with the differences generally being larger for American Indians and for girls. For example, the data suggest that American Indian boys are 1.7 times as likely as White boys and American Indian girls are 2.0 times as likely as White girls to have been sent to the office or detained after school. Overall, the relationship between race and ethnicity and being sent to the office or being detained is relatively weak (i.e., gamma coefficients .14 to .26), although it should be kept in mind that the large discrepancies in subgroup size places limits on the possible size of gamma. The fact that the gamma coefficients are, on average, larger among the girls than the boys is consistent with the differences in the magnitudes of the regression coefficients, where the differences appear somewhat larger among girls.

Model 2 shows the relationship between racial and ethnic group membership and being sent to the office or detained, controlling for family structure, parental education, urbanicity, and region. Although the absolute magnitudes of most of the logistic regression coefficients are reduced slightly from Model 1 to Model 2, all of the relationships remain statistically significant, with Black, Hispanic and American Indian 10^th graders being between 10% and 70% more likely than White 10^th graders to be sent to the office or detained and Asian American 10^th graders being approximately 40% less likely than White 10^th graders to be sent to the office or detained.

The data for the suspension and expulsion measure also demonstrate that there are significant racial and ethnic differences. Specifically, Black, Hispanic, and American Indian 10^th graders are between two to five times more likely than White 10^th graders to have been suspended or expelled compared to Asian American 10^th graders who are only half as likely (Model 1). The differences are particularly large for Blacks (3.3 times for boys and 5.4 times for girls). The magnitudes of the gamma coefficients suggest that the relationship between race and ethnicity and being suspended or expelled are considerably stronger than those for being sent to the office or detained after school. Consistent with the results for the office referral and detention measure, the gammas for the suspension and expulsion measure are, on average, larger among girls than boys. This finding suggests that although the prevalence of suspension and expulsion are higher among boys, the racial and ethnic differences in the probability of being suspended or expelled are actually larger for girls than for boys. Controlling for socio-demographic factors reduces the magnitudes of the racial and ethnic differences only modestly, and all of the subgroups remain significantly different from their White counterparts.

Discussion

The purpose of this study was to examine the problem of school discipline among youth in the United States, with a particular focus on racial and ethnic differences. Specifically, we sought to document the magnitude of racial and ethnic differences in the prevalence and disproportionality of the problem, to investigate the extent to which it has or has not changed over time, and to determine whether controlling for potential socio-demographic confounders might help to explain why non-White youth experience higher rates of school discipline than their White counterparts.

The results of the study reveal that both minor and more severe disciplinary practices are used widely, but distributed unequally, in U.S. schools. Consistent with the findings of previous research, we found that Black youth were significantly more likely than White youth to experience school discipline. We add to the literature by demonstrating that Hispanic and American Indian youth are also more likely – and Asian American youth less likely – than White youth to experience school discipline. Also consistent with past research, we found that race differences in the most punitive disciplinary practices (i.e., suspension and expulsion) are particularly large for Black youth. For example, Black boys are 30 percent more likely than White boys to be sent to the office or detained but they are 330 percent (3.3 times) more likely than White boys to be suspended or expelled. Among girls, the race gap in discipline is even larger. Black girls are approximately twice as likely as White girls to be sent to the office or detained but they are more than five times more likely than White girls to be suspended or expelled. Although the magnitudes of the differences are not as large as those for Black students, Hispanic and American Indian students are also significantly more likely than White youth to experience school disciplinary actions, even after socio-demographic differences are controlled.

Although the conventional wisdom suggests that there has been an increase in schools' use of disciplinary practices over time, this conclusion has been based on assumptions, largely without the kind of data necessary to confirm or disconfirm its validity. In the present study – one of the first studies to examine explicitly trends in school discipline, we found that there was an increase in discipline rates for most racial and ethnic subgroups during the 1990s that declined thereafter. Unlike most racial and ethnic groups however, Blacks' school discipline rates did not decline after 2000; rather, they continued to increase. Overall, we found that the general pattern of racial and ethnic differences in receiving school discipline has existed since at least since the early 1990s, with rates being highest among Black and American Indian students, somewhat lower among Hispanic students, lower still among White students, and lowest among Asian American students.

Given that school discipline has been found to relate to a number of key socio-demographic factors that are correlated, if not confounded, with race and ethnicity, we ran a series of logistic regression models to ascertain the extent to which the observed racial and ethnic differences in school discipline might result from racial and ethnic differences in variables like family structure, parental education, or urbanicity of residence. The results of these analyses failed to account for all (or most) of the differences, suggesting that other factors are responsible for the persistent finding that minority youth are more likely than White youth to experience school discipline.

In light of our findings that non-White youth are more likely than White youth to experience school discipline, an obvious question is, “why?” Past research suggests that there are at least four possible explanations for racial and ethnic disparities in school discipline (Skiba et al., 2002) The first, and perhaps most obvious, potential explanation for racial and ethnic disproportionality in school discipline would be that non-White youth are more likely than White youth to engage in behaviors that warrant discipline. The data presented in Table 2 address this question, at least for the kinds of behaviors that zero tolerance disciplinary policies were designed to address. The data suggest that there are relatively small but statistically significant racial and ethnic differences in these behaviors, with Hispanic youth being more likely than White youth to use substances and Black, Hispanic, and American Indian youth being more likely than White youth to have carried a weapon to school. That said, however, the magnitude of the racial and ethnic differences in school discipline far exceed the magnitude of the differences in substance use and weapons possession, suggesting that racial and ethnic differences in these particular behaviors are insufficient to account for the relatively large racial and ethnic differences in school discipline. That, of course, leaves open the possibility that other misbehaviors, such as acting up in class, account for or contribute to the differences. But if that were so, one would expect to see larger differences for “Sent to the Office or Detained,” given the large differences for “Suspended or Expelled” (see Table 3).

A second potential explanation for racial and ethnic disparities in school discipline that past research has identified is that the differences are a statistical artifact of the way that the data are reported (Skiba et al., 2002). For example, some research presents discipline data in terms of the percent of students within a particular group who have experienced school disciplinary actions (see Table 3) while other research presents racial and ethnic differences in disproportionality. In an effort to address this explanation, in the present study we examined the data both ways. These analyses provided consistent findings that many non-White youth, and Black youth in particular, were significantly more likely than their White counterparts to experience school discipline.

A third potential explanation for the finding of racial and ethnic differences in school discipline is, “the possibility that any finding of disporportionality due to race is a by-product of disproportionality associated with SES [socioeconomic status]” (Skiba et al., 2002, p. 321). In an effort to test this hypothesis, we conducted a series of logistic regressions in which we attempted to control for key indicators of socioeconomic status (e.g., parental education, family structure). Although our indicators of socioeconomic status are not comprehensive, the results of these analyses suggest that differences in socioeconomic status have relatively little impact on racial and ethnic differences in school discipline.

The final possible explanation that has been given to account for racial and ethnic differences in school discipline is teacher and administrator bias, or discrimination, in the ways in which they perceive and respond to non-White versus White youth (see Skiba et al., 2002; Townsend, 2000). Unfortunately, our data did not include measures of discrimination, bias, or differential treatment, and thus we were unable to investigate the tenability of this hypothesis. That said, however, there are data, both qualitative and quantitative, that suggest that there are language, cultural, and other differences between many educators and Black youth that may help to account for race differences in both disciplinary and academic outcomes (e.g., Gregory & Mosely, 2004; Ruck & Wortley, 2002; Skiba et al., 2002; Vavrus & Cole, 2002). For example, anecdotally, a White female school administrator recently shared with one of the co-authors of the present study the series of events that led up to a disciplinary hearing in which an Black male high school student was referred to the office to be suspended. A White male teacher was running late for class. Upon his arrival, the Black male student met the teacher at the door and said, “Man, I was just fixin' to bounce on you.” To the student's bewilderment, the teacher wrote him up to be suspended. The teacher (mis)interpreted the phrase, “fixin' to bounce on you,” as a threat of physical violence, when from the student's perspective he was noting the teacher's tardiness and jokingly saying that he was just about to leave the classroom (i.e., “bounce”). While this is a simple example, it highlights the kind of language and cultural misunderstandings that can result in disproportionality in school discipline.

Limitations

Unlike most previous investigations of racial and ethnic differences in school discipline, the present study used students' self-reports versus school records or other “official” statistics. To the extent that young people might be inclined to under-report their experiences of school discipline, the data we present may not provide fully accurate estimates of the magnitude of the problem. To our knowledge the extant research has examined neither the validity of adolescents' self-reports of school discipline nor whether there are racial and ethnic differences in reporting behavior. Interestingly, however, there has been some assessment of the validity of schools' self-reporting of violence and related disciplinary actions (see Snell, 2005). The Federal No Child Left Behind Act of 2001 (20 U.S.C. § 7912) permits parents to transfer their students to other schools, if the school that their children attend is determined by state-level definitions to be “persistently dangerous” (Sec. 9532). In light of the potential loss of funds and the political cost of being identified as a dangerous school, many schools may underreport their violence and discipline statistics. Accordingly, student self-reports of school discipline may be as valid, if not more valid, than the data that individual schools, districts, and perhaps even states are willing to report.

Directions for Future Research

Past research and the present study reveal that race and ethnicity, gender, and their interaction are important predictors of who is likely to experience school disciplinary actions. In light of recent qualitative and quantitative research efforts that document language, class, and cultural, mismatches between students, teachers, and other school officials, future research should seek to better understand and ameliorate the discontinuities that exist between non-White young people and the White and non-White teachers, administrators, and other school personnel with whom students come in contact. Future research should also examine explicitly the extent to which racial and ethnic differences in misbehavior versus teacher biases in perceptions and attitudes about non-White students account for racial and ethnic differences in school discipline. Future research might also benefit greatly by looking at the role that social contexts play in causing subgroup differences in discipline. For example, is discipline higher or lower in schools in which the vast majority of students, teachers and administrators are non-White? Are school discipline rates higher in integrated versus segregated schools? To what extent might a community's economic status influence suspension rates of its children in their neighborhood schools? These and related questions must be addressed if America is going to maximize the talents of all of its students and reduce the days, weeks, months, or years of learning that are lost as students are excluded from learning environments across the nation.

Appendix A: Zero Tolerance Behavior Questions

The alcohol use measure asks students, “During the last 12 months, how often (if ever) have you used alcohol in each of the following places…At school during the day?” We recoded the variable 1 = 1 or more times, 0 = not at all.

The drug use measure asks students, “During the last 12 months, how often (if ever) have you used marijuana or any other drugs (like cocaine amphetamines, etc.) in each of the following places…At school during the day? We recoded the variable 1 = 1 or more times, 0 = not at all.

The gun possession question asks students, “During the last four weeks, on how many days (if any) did you carry a gun to school?)” We recoded the variable 1 = 1 or more days, 0 = not at all.

Contributor Information

John M. Wallace, Jr., University of Pittsburgh, School of Social Work and Center on Race and Social Problems.

Sara Goodkind, University of Pittsburgh, School of Social Work and Center on Race and Social Problems.

Cynthia M. Wallace, University of Pittsburgh, Learning and Research Development Center.

Jerald G. Bachman, University of Michigan, Institute for Social Research.

References