race is not neutral: a national investigation of african ...equity/docs/skiba et al race is not...

Race Is Not Neutral: A National Investigation ofAfrican American and Latino Disproportionality

in School Discipline

Russell J. SkibaIndiana University

Robert H. HornerUniversity of Oregon

Choong-Geun Chung and M. Karega RauschIndiana University

Seth L. May and Tary TobinUniversity of Oregon

Abstract. Discipline practices in schools affect the social quality of each educa-tional environment, and the ability of children to achieve the academic and socialgains essential for success in a 21st century society. We review the documentedpatterns of office discipline referrals in 364 elementary and middle schools duringthe 2005–2006 academic year. Data were reported by school personnel throughdaily or weekly uploading of office discipline referrals using the Web-basedSchool-wide Information System. Descriptive and logistic regression analysesindicate that students from African American families are 2.19 (elementary)to 3.78 (middle) times as likely to be referred to the office for problem behavioras their White peers. In addition, the results indicate that students from AfricanAmerican and Latino families are more likely than their White peers to receiveexpulsion or out of school suspension as consequences for the same or similarproblem behavior. These results extend and are consistent with a long history ofsimilar findings, and argue for direct efforts in policy, practice, and research toaddress ubiquitous racial and ethnic disparities in school discipline.

The Supreme Court’s ruling in Brown v.Board of Education in 1954 set the nation ona path toward equalizing educational opportu-

nity for all children. The right not to be dis-criminated against on the basis of race, color,or national origin was explicitly guaranteed by

This research was supported in part by U.S. Department of Education Grant H326S980003. Opinionsexpressed herein do not necessarily reflect the policy of the Department of Education, and no officialendorsement by the Department should be inferred.

Correspondence regarding this article should be addressed to Russell J. Skiba, Center for Evaluation andEducation Policy, 1900 E. 10th St., Bloomington, IN 47401; e-mail: [email protected]

Copyright 2011 by the National Association of School Psychologists, ISSN 0279-6015

School Psychology Review,2011, Volume 40, No. 1, pp. 85–107

85

Title VI of the Civil Rights Act of 1964(Browne, Losen, & Wald, 2002). Those pro-tections were expanded to students with dis-abilities in the Individuals with DisabilitiesEducation Improvement Act of 2004 and toeducational outcomes for all children in theElementary and Secondary Education Act (NoChild Left Behind, 2008). Yet continuing ra-cial and ethnic disparities in education rangingfrom the achievement gap (Ladson-Billings,2006) to disproportionality in special educa-tion (Donovan & Cross, 2002) to dropout andgraduation rates (Wald & Losen, 2007) haveled some to question the extent to which thepromises of Brown have been fulfilled(Blanchett, Mumford, & Beachum, 2005). Inparticular, over 30 years of research has doc-umented racial and socioeconomic disparitiesin the use of out-of-school suspension andexpulsion. The purpose of this article is todescribe a national investigation exploring theextent of, and patterns in, racial and ethnicdisparities in school discipline at the elemen-tary and middle school level.

Consistently DemonstratedDisproportionality

For over 25 years, in national-, state-,district-, and building-level data, students ofcolor have been found to be suspended at ratestwo to three times that of other students, andsimilarly overrepresented in office referrals,corporal punishment, and school expulsion(Skiba, Michael, Nardo, & Peterson, 2002).Documentation of disciplinary overrepresenta-tion for African American students has beenhighly consistent (see e.g., Gregory, 1997;McCarthy & Hoge, 1987; McFadden, Marsh,Price, & Hwang, 1992; Raffaele Mendez &Knoff, 2003; Skiba et al., 2002; Wu, Pink,Crain, & Moles, 1982). According to datafrom the U.S. Department of Education Officefor Civil Rights, disciplinary disproportional-ity for African American students appears tohave increased from the 1970s, when AfricanAmericans appeared to be at approximatelytwice the risk of out of school suspension to2002, when African American students riskfor suspension was almost three times as great

as White students (Wald & Losen, 2003). Al-though disciplinary overrepresentation of La-tino students has been reported in some studies(Raffaele Mendez & Knoff, 2003), the findingis not universal across locations or studies (seee.g., Gordon, Della Piana, & Keleher, 2000).

Possible Causative Mechanisms

A number of possible hypotheses havebeen proposed as mechanisms to account forrates of disciplinary disparity by race/ethnic-ity, including poverty, differential rates of in-appropriate or disruptive behavior in schoolsettings, and cultural mismatch or racial ste-reotyping. The possible mechanisms are dis-cussed in the following.

Poverty

Race and socioeconomic status (SES)are unfortunately highly connected in Ameri-can society (McLoyd, 1998), raising the pos-sibility that any finding of racial disparities inschool discipline can be accounted for by dis-proportionality associated with SES. Low SEShas been consistently found to be a risk factorfor school suspension (Brantlinger, 1991; Wuet al., 1982). Yet when the relationship of SESto disproportionality in discipline has beenexplored directly, race continues to make asignificant contribution to disproportionatedisciplinary outcomes independent of SES(Skiba et al., 2002; Wallace, Goodkind, Wal-lace, & Bachman, 2008; Wu et al., 1982).

Higher Rates of Disruption AmongStudents of Color

A related hypothesis might be that stu-dents of color, perhaps because they have beensubjected to a variety of stressors associatedwith poverty (see e.g., Donovan & Cross,2002), may learn and exhibit behavioral stylesso discrepant from mainstream expectations inschool settings as to put them at risk for in-creased disciplinary contact. Investigations ofstudent behavior, race, and discipline haveconsistently failed, however, to find evidenceof differences in either the frequency or inten-sity of African American students’ school be-

School Psychology Review, 2011, Volume 40, No. 1

86

havior sufficient to account for differences inrates of school discipline. Some studies havefound no significant differences in behaviorbetween African American and White students(McCarthy & Hoge, 1987; Wu et al., 1982),while others have reported that African Amer-ican students receive harsher levels of punish-ment for less serious behavior than other stu-dents (McFadden et al., 1992; Shaw &Braden, 1990). Skiba et al. (2002) comparedthe types of infractions for which AfricanAmerican and White middle school students ina large urban district were referred to the of-fice, and found no obvious differences in se-verity of behavior, but that African Americanstudents tended to be referred to the officemore often for offenses that required a higherdegree of subjectivity, such as disrespect orloitering.

Cultural Mismatch or RacialStereotyping

With a teaching force in most schooldistricts in this nation that is predominantlyWhite and female (Zumwalt & Craig, 2005),the possibility of cultural mismatch or racialstereotyping as a contributing factor in dispro-portionate office referral cannot be discounted.Townsend (2000) suggested that the unfamil-iarity of White teachers with the interactionalpatterns that characterize many African Amer-ican males may cause these teachers to inter-pret impassioned or emotive interactions ascombative or argumentative. In an ethno-graphic study of disciplinary practices at anurban elementary school, Ferguson (2001)documented the seemingly unconscious pro-cess whereby racial stereotypes may contrib-ute to higher rates of school punishment foryoung African American males.

There is some indication that teachersdo make differential judgments about achieve-ment and behavior based on racially condi-tioned characteristics. Neal, McCray, Webb-Johnson, and Bridgest (2003) found that stu-dents who engaged in a “stroll” style ofwalking more often associated with AfricanAmerican movement style were more likely tobe judged by teachers as being more aggres-

sive or lower achieving academically, whetherthe student was African American or White. Inan extensive study of teacher ratings, Zimmer-man, Khoury, Vega, Gil, and Warheit (1995)found evidence that African American stu-dents were more likely to be rated as havingmore extensive behavior problems by bothHispanic and non-Hispanic White teachers. Inaddition, teachers were more likely than par-ents to rate African American students as moreproblematic and less likely than parents to rateWhite students’ behavior as more problematic.In a more restricted sample set in a high-poverty inner-city setting, Pigott and Cowen(2000) found no evidence of a child–teacherrace interaction in teacher ratings of their stu-dents, but found that all teacher groups re-ported a higher incidence of race-related ste-reotypes for African American students.

There is some classroom observationaldata consistent with either a cultural mismatchor racial stereotyping explanation. Vavrus andCole (2002) analyzed videotaped interactionsamong students and teachers, and found thatmany ODRs were less the result of seriousdisruption than what the authors described as“violations of … unspoken and unwritten rulesof linguistic conduct” (p. 91), and that studentssingled out in this way were disproportion-ately students of color. In a study of officereferral practices in an urban high school,Gregory and Weinstein (2008) found that,among a sample of African American studentswith ODRs, differences in classroom manage-ment style significantly contributed to bothstudent attitudes toward classroom manage-ment and actual disciplinary outcomes. Fur-ther, even among students with multiple refer-rals to the office, only certain student–teachercombinations resulted in higher rates of officereferral.

Summary

A number of hypotheses might be ap-plied to explain the ubiquitous overrepresen-tation of African American students in a rangeof school disciplinary consequences. It seemslikely that, in the face of multiple hypotheses,the disproportionate representation of students

Disciplinary Disproportionality

87

of color in school discipline is complex andmultiply determined.

Risks of DisproportionateRepresentation in School Exclusion

Overrepresentation in out-of-school sus-pension and expulsion appears to place Afri-can American students at risk for a number ofnegative outcomes that have been found to beassociated with those consequences. First,given documented positive relationships be-tween the amount and quality of engaged timein academic learning and student achievement(Brophy, 1988; Greenwood, Horton, & Utley,2002), and conversely between school alien-ation/school bonding and subsequent delin-quency (Hawkins, Doueck, & Lishner, 1988),procedures like out-of-school suspension andexpulsion that remove students from the op-portunity to learn and potentially weaken theschool bond must be viewed as potentiallyrisky interventions. Second, a substantial da-tabase has raised serious concerns about theefficacy of school suspension and expulsion asa behavioral intervention in terms of eitherreductions in individual student behavior oroverall improvement in the school learningclimate (see e.g., American Psychological As-sociation, 2008). Finally, by removing stu-dents from the beneficial aspects of academicengagement and schooling, suspension andexpulsion may constitute a risk factor for fur-ther negative outcomes, including poor aca-demic performance (Skiba & Rausch, 2006),school dropout (Ekstrom, Goertz, Pollack, &Rock, 1986), and involvement in the juvenilejustice system (Wald & Losen, 2003). Thus,the overrepresentation of African Americanstudents in such high-risk procedures must beconsidered highly serious.

Gaps in Knowledge

There are substantial gaps in the re-search literature exploring racial and ethnicdisparities in school discipline, some extend-ing to basic descriptive information. Data con-cerning the representation of Hispanic/Latinostudents in school discipline are limited andhighly inconsistent. Few studies of school dis-

cipline have focused on school level as a vari-able (elementary vs. middle vs. high school)and fewer still have examined disproportion-ality across school levels (Skiba & Rausch,2006). Third, although the disciplinary processhas been recognized as a complex, multilevelprocess proceeding from office referral to ad-ministrative disposition (see e.g., Morrison,Anthony, Storino, Cheng, Furlong, & Morri-son, 2001), little attention has been paid to therelative contribution of office referrals andadministrative consequences to racial and eth-nic disparities in school discipline. Finally,few investigations have been both comprehen-sive and detailed. That is, empirical investiga-tions of school disciplinary processes appeareither to rely on national or state databases(e.g., U.S. Department of Education Office forCivil Rights data) that provide a comprehen-sive perspective on suspension or expulsion,but little detail concerning the initial offensethat led to referral; or to analyze local schoolor district databases of ODRs that provide aricher picture of student infractions, but mayor may not be generalizable to other locations.

Purpose and Assumptions

The purpose of this investigation was toexplore racial and ethnic disparities in officereferrals and administrative discipline deci-sions in a nationally representative sample.The schools in the sample had been involvedin efforts to reform their school disciplinarypractices using School-wide Positive BehaviorSupports (SWPBS) for at least one year.SWPBS is a whole-school approach to preven-tion of problem behavior that focuses on de-fining, teaching, and rewarding behavioral ex-pectations; establishing a consistent contin-uum of consequences for problem behavior;implementing a multitiered system of behaviorsupports; and the active use of data for deci-sion making (Sugai & Horner, 2006). A coreelement of the SWPBS implementation pro-cess is systematic data collection on occur-rence of problem behaviors that result in officereferrals and the discipline decisions associ-ated with those referrals.


88

Although the data were drawn from asubsample of schools implementing SWPBS,the purpose of this investigation was not inany way to explore the effects or effectivenessof SWPBS as an intervention for reducingdisciplinary referrals or disproportionality inreferrals. Rather, to our knowledge, these dataprovide the most comprehensive and nation-ally representative sample for addressing someof the gaps in research knowledge regardingracial and ethnic disproportionality in schooldisciplinary procedures. We used descriptiveand logistic regression analyses to explore pat-terns of disproportionality in office referralrates, patterns of disciplinary decisions acrossdifferent racial/ethnic groups (African Ameri-can, Hispanic, White), and school level (ele-mentary vs. middle school).

The analyses make two assumptionsabout effective disciplinary practices andhence about the types of data that would pro-vide evidence of an effective and equitabledisciplinary system. First, we presume that themost effective disciplinary systems are gradu-ated discipline systems (American Psycholog-ical Association, 2008) in which minor infrac-tions produce less severe administrative con-sequences than more severe infractions. Thephilosophy and practice of zero tolerance hastended to emphasize an alternate model, inwhich both minor and major infractions aremet with more severe consequences, but asubstantial database has failed to support theefficacy of practices based on that perspective(American Psychological Association, 2008).Second, given that there is no evidence sup-porting a distribution of infractions that variesin severity by race, we presume that disciplin-ary outcomes will be proportional across ra-cial/ethnic categories.

Methods

Data Source

The subjects for this investigation weredrawn from data generated by the School-wideInformation System (SWIS: May et al., 2006),which was being used in over 4000 schoolsacross the nation during the 2005–206 aca-

demic year (cf. http://www.swis.org, January2007).

The SWIS is a three-component deci-sion system for gathering and using schooldiscipline data for decision making. The com-ponents of SWIS are (a) a data collectionprotocol that schools adopt that uses opera-tionally defined categories for problem behav-ior, school-wide standards defining whichproblem behaviors are addressed in class-rooms versus sent to the office, and a structurefor team meetings in which data are used; (b)a Web-based computer application for enter-ing ODR data and retrieving summary reportsin graphic and tabular formats (May et al.,2006), and (c) a facilitator-based training sys-tem to help teams use data for active decisionmaking.

The entry of data into the SWIS com-puter application requires that students beidentified by name, district identification num-ber, grade, Individualized Education Program(IEP) status, and ethnicity. The content of anoffice discipline referral includes informationabout (a) the type of problem behavior leadingto the referral; (b) the time of day, location,referring adult, and others present during theevent; (c) the presumed maintaining conse-quence (e.g., access to attention, escape fromwork, response to taunting from peers); and(d) the primary administrative decision (e.g.,conversation, detention, loss of privilege, par-ent report, suspension) resulting from the re-ferral. This information is summarized in aseries of reports that allow an administrator,specialist, team, or faculty member to monitorthe rate of office discipline referrals; the typeof behaviors leading to referrals; the time ofday, location, and presumed maintaining con-sequence; and the administrative decisionpatterns.

The SWIS system also provides the op-tion for a school to compute a summary ofODRs by race/ethnicity. Race/ethnicity withinSWIS is determined by the family designationwhen a child is enrolled in school, but islimited in specificity to the six federal racecategories: African American, Asian, NativeAmerican, Pacific Islander, Hispanic/Latino,and Caucasian.


89

Selection of problem behavior for allschools using SWIS is based on a mutuallyexclusive and exhaustive list of 24 operation-ally defined “major problem behaviors” andthree operationally defined “minor problembehaviors.” School-based consequences forthese reported behaviors are coded into 14mutually exclusive and exhaustive categoriesof “administrative decisions.” Operationaldefinitions of both student behaviors and ad-ministrative decisions may be found on theSWIS Resources site at http://www.swis.org/index.php?page � resources;rid � 10121.

Participant Sample

As of January 2007, there were over4000 schools in the United States at varyingstages of SWIS adoption. During the fall of2007, we identified from this population ofschools a subset of 436 schools who (a) usedSWIS for the full 2005–2006 academic year,(b) reported ethnicity information, (c) hadgrade levels between kindergarten and sixthgrade (K–6) or sixth and ninth grade (6–9),and (d) agreed to share anonymous summariesof their data for evaluation purposes. Theseschools reported total enrollment of 120,148students in elementary grades (K–6)and 60,522 students in middle school grades(6–9). Spaulding et al. (2010), compared thedemographic features of the sample with73,525 comparable schools in the NationalCenter for Educational Statistics (NCES) sam-ple for 2005–2006 to assess bias in size, pro-portion of students with an IEP, SES, racial/ethnic distribution, and location (urban, sub-urban, or rural). They found the SWISdatabase to be within 5% of the NCES data inall categories except that the SWIS databasewas composed of fewer schools within thehigh SES category and had a larger number ofschools with high ethnic/racial diversity. Atthe K-6 and 6–9 levels, the sample did notdiffer from the NCES data in the percentage oflarge, midsized, or rural schools represented,but at the high school level there were moreschools with large enrollment.

It is important to note that schoolsadopting SWIS were self-nominated, and in

most cases were in varying stages of imple-menting school-wide positive behavior sup-port (Sugai et al., 2002; Lewis & Sugai, 1999).Adoption of SWIS does not require thatschools also adopt school-wide positive be-havior support, but often districts investing inadoptions of school-wide discipline systemsare more likely to also invest in adoption ofSWIS. Again, the purpose of this investigationwas not to explore any attributes, effects, oreffectiveness of the use of SWPBS in theseschools. No information was available withinthe SWIS database to indicate the length orfidelity of SWPBS adoption.

For purposes of analysis, the 436schools were organized into an elementarylevel (K–6) and a middle school level (6–9).Schools that overlapped to some extent (e.g.,Grades 5–9) were placed in the group with thelargest degree of overlap. Schools with a de-gree of overlap that did not permit sorting withconfidence (e.g., schools serving Grades K–8)were dropped from the sample. The final sam-ple included 272 K–6 level schools and 926–9 level schools.

Research Questions and Variables

The primary research questions focusedfirst on the pattern of ODRs by race and thenon the pattern of administrative decisions byrace. The following questions guided thestudy.

ODRs. Two questions were addressedthrough descriptive data and logistic regres-sion analyses.

1. To what extent does racial/ethnic statusmake a contribution to rates of ODR inelementary or middle schools?

2. In which categories of ODRs are racialor ethnic disparities evident?

The original SWIS data included 27 cat-egories of disciplinary infraction that could beentered as the reason for an ODR. For con-ceptual and analytic clarity, we categorizedthose infractions into the categories minormisbehavior, disruption, noncompliance,moderate infractions, major violations, use/possession, tardy/truancy, and other/unknown


90

(Note: specific infractions included in eachcategory may be found in Table 3).

Administrative decisions. Three ques-tions were addressed through descriptive data,simply logistic regression, and multinomiallogit models.

3. To what extent does racial/ethnic statusmake a contribution to administrativedecisions concerning disciplinary con-sequences in elementary or middleschools?

4. In which categories of disciplinary con-sequence are racial or ethnic disparitiesevident?

5. What are the racial disparities in theinteraction of infraction types and ad-ministrative decisions regarding conse-quence? In which infraction/conse-quence pairs do such disparities occur?

The original SWIS data included 14 cat-egories of administrative decision regardingprimary disciplinary consequence. For con-ceptual and analytic clarity, we categorizedthose decisions into the categories of minorconsequences, detention, moderate conse-quences, in-school suspension, out-of-schoolsuspension and expulsion, and other/unknown.(Note: Specific administrative decisions com-prising each category may be found inTable 4.)

Data analyses. Descriptive analysesand a series of logistic and multinomial logitregression analyses (Greene, 2008) were usedto describe the extent of disproportionality instudent infractions, administrative decisions,and their interaction. The first logistic equa-tion addressing Research Question 2 (Table 2)was designed to test the extent to which raceproved a factor in the probability of an ODR.

The second analysis addressing Re-search Question 2 (Table 3) was a multinomiallogit model designed to explore the contribu-tion of race to referrals for specific types ofinfraction. The third regression, addressingQuestions 3 and 4, was a multinomial logitregression testing the influence of type of in-fraction and race on the probability of receiv-

ing a given consequence (Table 4). The finalmultinomial logit regression, addressing Re-search Question 5, was designed to test thespecific influence of race on the probability ofall possible infraction/consequence interac-tions (Table 5).

Early analyses showing differences inpatterns of results by school level led us toconduct separate analyses for K–6 and 6–9schools. The measure used to index dispropor-tionality throughout the analyses is the oddsratio (OR) drawn from the logistic and multi-nomial logit regression equations, with valuesgreater than 1.0 indicating overrepresentationand values less than 1.0 indicating underrep-resentation. In contrast to risk ratios, the ORmay offer a more stable and accurate estimateof disproportionality, because it accounts forboth occurrence and nonoccurrence of theevent being measured (Finn, 1982; Oswald,Coutinho, Best, & Singh, 1999). For all anal-yses involving racial/ethnic categories, the in-dex category was “White,” while the indexcategory for infractions or administrative de-cisions varied and is noted in the footnote ofthe table for that analysis. Finally, note thatdescription of the results refers to “overrepre-sentation” or “underrepresentation” of a givenracial/ethnic group with respect to a giveninfraction or disciplinary consequence. In de-scriptive results, presented as composition in-dices, proportionality is compared to thatgroup’s representation in the population.Across all logistic and multinomial logit anal-yses, disproportionality is framed in terms ofover- or underrepresentation in comparison tothe index group. This is not intended to con-vey any judgment concerning how frequentlya given consequence ought to be applied (e.g.,Latino students being underrepresented in de-tention does not in any way suggest theyshould be placed in detention more fre-quently), but rather is simply a numerical rep-resentation of the probability of occurrencerelative to the index group (White students).The Box Tidwell Transformation Test (Cohen,Cohen, West, & Aiken, 2003) was performedto test for the assumption that there is a linearrelationship between the independent vari-ables and the log odds (logit) of the dependent


91

measure. Tests across all models failed toyield any significant results, indicating thatthere is no nonlinearity issue for this sample.

Results

Disproportionate representation inschool discipline can occur at either the pointof referral or administrative decision. To trackdisproportionality through these two points inthe disciplinary process, results are organizedinto two sections: ODRs and administrativedecisions.

ODRs

Table 1 presents the total enrollment,number of students referred to the office, andnumber of ODRs disaggregated by race forK–6 and 6–9 level schools. The percentagesin Columns 2, 4, and 6 are column percent-ages, reflecting the percent of enrollment, stu-dents referred, or total number of ODRs re-spectively for each racial/ethnic group. Eachpercentage in Column 4 or 6 therefore repre-

sents a composition index (Donovan & Cross,2002) that can be interpreted by comparing itto the percent overall enrollment in Column 2.Thus, at the K–6 level, African Americanstudents appear to be overrepresented, relativeto their proportion in the population, amongthose referred to the office, represent-ing 25.8% of total enrollment (Column 2),but 35.3% of those referred to the office (Col-umn 4). White and Hispanic/Latino studentsare underrepresented relative to their enroll-ment among those referred to the office at theK–6 level. At the 6–9 level, African Ameri-can students appear to be overrepresented, andWhite students appear to be underrepresentedin their rate of ODRs as compared to theirpercentage in the population. Hispanic/Latinostudents appear to be roughly proportionatelyrepresented in middle school ODRs. The levelof overreferral of African American studentsbecomes even more apparent if one examinesthe absolute number of referrals to the office(Column 6). The discrepancy between individ-

Table 1Enrollment, Number of Students Referred, and Number of Referrals

Disaggregated by Racial Ethnic Group

Group

Enrollment Students Referred Referrals

N % N % N %

K–6 Schools (N � 272)Hispanic/Latino 25,051 20.9 4,311 13.1 12,863 9.6African American 30,961 25.8 11,577 35.3 57,601 43.0White 54,690 45.5 11,703 35.7 45,900 34.3Unknown/all othersa 9,446 7.9 5,212 15.9 17,518 13.1

Total N of students 120,148 100.0 32,803 100.0 133,882 100.0

6–9 Schools (N � 92)Hispanic/Latino 10,332 17.1 4,245 16.9 18,419 14.5African American 13,228 21.9 8,024 32.0 52,894 41.7White 32,975 54.5 9,542 38.1 42,605 33.6Unknown/all others 3,987 6.6 3,260 13.0 12,842 10.1

Total N of students 60,522 100.0 25,071 100.0 126,760 100.0

Note. K–6 � kindergarten through Grade 6; 6–9 � Grade 6 through Grade 9.aUnknown/all others: American Indian/Alaskan Native, Asian, Pacific Islander/Native Hawaiian, Not Listed, andUnknown.


92

ual referrals and total referrals suggests ahigher rate of multiple referrals to the officefor African American students as opposed toWhite or Hispanic/Latino students at both theelementary and middle school levels.

Differences in ODRs across racial/eth-nic categories were more directly compared inthe ORs drawn from a logistic regression anal-ysis predicting the probability of at least oneODR (Table 2). The index group for all anal-yses is White students. All ORs in Table 2 aresignificant at the 0.01 level. At the K–6 level,African American students’ odds of being re-ferred to the office are 2.19 times that of Whitestudents; the overrepresentation of AfricanAmericans in ODRs relative to White studentsappears to increase (OR � 3.79) at the 6–9level.

A somewhat different pattern of dispro-portionality is in evidence for Hispanic/Latinostudents. At the K–6 level, Hispanic/Latino

students are underrepresented in their rate ofreferral to the office relative to White students(OR � 0.76). At the 6–9 level, however, His-panic/Latino students in this sample are over-represented (OR � 1.71) in their rate ofODRs. These results at the 6–9 level appear atfirst glance to be somewhat at odds with com-position indices (Table 1) that appear to showrates of ODRs for Hispanic/Latino studentsthat are roughly proportionate at the 6–9 level.Thus, significant Latino overrepresentationrelative to White students at the middle schoollevel appears to be from, not the absoluteoverreferral of Latino students, but rather tothe substantial underreferral to the office ofWhite students as compared to their represen-tation in the population.

The ORs associated with ODRs brokendown by infraction comparing African Amer-ican and Latino students with White studentsare presented in Table 3. With the exception ofTardy/Truancy, Major Violations, and Use/Possession for Hispanic/Latino students at theK–6 level, all ORs are significant at the p �.01 level. At both the K–6 and 6–9 levels,African American students are significantlyoverrepresented in ODRs across all infractiontypes, with the highest ORs compared toWhite students for the infraction types of Tar-dy/Truancy, Disruption, and Noncompliance.At the K–6 level, Hispanic/Latino students areunderrepresented as compared with White stu-dents in ODRs for Minor Misbehaviors, Dis-ruption, Noncompliance, and Moderate Infrac-tions. At the 6–9 level, in contrast, Hispanic/Latino students appear to be overrepresentedrelative to White students for all ODRcategories.

Administrative Decisions

Table 4 presents the results of a multi-nomial logistic regression predicting the like-lihood of a particular administrative decisionusing two models. In Model 1, the likelihoodof an administrative consequence is predictedsolely from type of infraction. In Model 2,race/ethnicity is added to type of infraction topredict the likelihood of an administrativeconsequence. For both Model 1 and Model 2,

Table 2Logistic Regression of the Influence

of Race/Ethnicity on Referrala

GroupReferred:

Odds Ratio

K–6 SchoolsHispanic/Latino 0.76*African American 2.19*Unknown/all others NANumber of cases 120,148Model �2 7,152Nagelkerke Pseudo R2 0.084% Correctly predicted 73.5

6–9 SchoolsHispanic/Latino 1.71*African American 3.79*Unknown/all others NANumber of cases 60,522Model �2 6,925Nagelkerke Pseudo R2 0.146% Correctly predicted 67.4

Note. K–6 � kindergarten through Grade 6; 6–9 �Grade 6 through Grade 9; NA � not available.aReference category is Not Referred for outcome and isWhite for Race/Ethnicity.*p � .05.


93

Tab

le3

Mu

ltin

omia

lL

ogit

Reg

ress

ion

ofth

eIn

flu

ence

ofR

ace/

Eth

nic

ity

onR

efer

rals

byIn

frac

tion

s:O

dds

Rat

iosa

Gro

upM

inor

Mis

beha

vior

sD

isru

ptio

nN

onco

mpl

ianc

eM

oder

ate

Infr

actio

nsM

ajor

Vio

latio

nsU

se/P

osse

ssio

nT

ardy

/T

ruan

cyO

ther

/U

nkno

wn

K–6

Scho

ols

His

pani

c/L

atin

o0.

66**

0.67

**0.

60**

0.77

**0.

911.

220.

840.

80**

Afr

ican

Am

eric

an1.

77**

3.96

**3.

32**

2.62

**2.

87**

2.96

**6.

58**

2.66

**U

nkno

wn/

all

othe

rsN

AN

AN

AN

AN

AN

AN

AN

An

142,

451

Mod

el�

212

,848

Nag

elke

rke

Pseu

doR

20.

093

%C

orre

ctly

pred

icte

d61

.3

6–9

Scho

ols

His

pani

c/L

atin

o1.

50**

1.31

**1.

87**

1.76

**1.

80**

1.93

**2.

44**

1.30

**A

fric

anA

mer

ican

3.72

**5.

60**

5.43

**4.

76**

3.91

**2.

02**

4.40

**4.

55**

Unk

now

n/al

lot

hers

NA

NA

NA

NA

NA

NA

NA

NA

n89

,279

Mod

el�

212

,001

Nag

elke

rke

Pseu

doR

20.

129

%C

orre

ctly

pred

icte

d40

.5

Not

e.M

inor

Mis

beha

vior

s�

min

orin

appr

opri

ate

verb

alla

ngua

ge,

min

orph

ysic

alco

ntac

t,m

inor

defia

nce/

disr

espe

ct/n

onco

mpl

ianc

e,m

inor

disr

uptio

n,m

inor

prop

erty

mis

use,

othe

rm

inor

mis

beha

vior

s;D

isru

ptio

n�

disr

uptio

n;N

onco

mpl

ianc

e�

defia

nce/

disr

espe

ct/in

subo

rdin

atio

n/no

ncom

plia

nce;

Mod

erat

eIn

frac

tions

�ab

usiv

ela

ngua

ge/in

appr

opri

ate

lang

uage

,fig

htin

g/ph

ysic

alag

gres

sion

,ly

ing/

chea

ting,

hara

ssm

ent/b

ully

ing;

Maj

orV

iola

tions

�pr

oper

tyda

mag

e,fo

rger

y/th

eft,

vand

alis

m,

bom

bth

reat

/fal

seal

arm

,ar

son;

Use

/Pos

sess

ion

�us

e/po

sses

sion

ofto

bacc

o,al

coho

l,co

mbu

stib

leite

ms,

wea

pons

,dr

ugs;

Tar

dy/T

ruan

cy�

tard

y,sk

ipcl

ass/

trua

ncy,

dres

sco

devi

olat

ion;

Oth

er/U

nkno

wn

�ot

her

beha

vior

,un

know

nbe

havi

or;

K–

6�

kind

erga

rten

thro

ugh

Gra

de6;

6–9

�G

rade

6th

roug

hG

rade

9;N

A�

not

avai

labl

e.a E

ntri

esin

each

cell

repr

esen

tth

eod

dsra

tiodr

awn

from

the

mul

tinom

ial

logi

t,ho

ldin

gco

nsta

ntth

eco

ntri

butio

nof

all

othe

rva

riab

les.

Ref

eren

ceca

tego

ryis

Not

Ref

erre

dfo

rou

tcom

ean

dis

Whi

tefo

rR

ace/

Eth

nici

ty.

**p

�.0

1.


94

the odds of receiving a suspension/expulsionfor committing a minor infraction are very lowat both the K–6 and 6–9 levels, and appear toincrease proportionally as the seriousness ofinfraction increases. Thus, the odds of receiv-ing a suspension or expulsion for Use/Posses-sion are very high at both the K–6(OR � 16.60) and 6–9 (OR � 53.01) level.

Model 2 in Table 4 adds race/ethnicityto the model, and results in little change to theORs or significance for type of infraction ascompared to Model 1. Race/ethnicity entersthe equation significantly for most administra-tive decisions. Both African American andLatino students are overrepresented in suspen-sion/expulsion relative to White students atboth K–6 and 6–9 levels. African Americanstudents are underrepresented in the use ofdetention at the K–6 level, and underrepre-sented in all administrative consequences ex-cept suspension/expulsion at the 6–9 level. Incontrast, Hispanic/Latino students are under-represented relative to White students in theuse of moderate consequences, but overrepre-sented in detention at both the K–6 and 6–9levels. The continuing significance of race/ethnicity in Model 2 even after controlling fortype of behavior indicates that race/ethnicitymakes a contribution to administrative deci-sions regarding discipline independent of typeof infraction, above and beyond any prior dis-parity in classroom referral.

Table 5 presents the ORs for receivingvarious administrative consequences brokendown by race and type of infraction drawnfrom a series of multinomial logit regressionanalyses at the K–6 and 6–9 levels, testing forthe presence of differential administrativetreatment for the same offense. The resultsdescribe a complex pattern of variation acrosstype of infraction, race/ethnicity, and schoollevel. At the elementary level, African Amer-ican students were more likely than Whitestudents to receive out-of-school suspension/expulsion for all types of infractions tested(note that tardiness/truancy and use/possessioncould not be estimated in the model becausezero cells). In particular, results indicated thatAfrican American elementary school studentswere more likely than White students to be

suspended out-of-school for minor misbehav-ior (OR � 3.75, p � .01). They were also lesslikely to receive in-school suspension for dis-ruption or noncompliance, less likely to re-ceive moderate consequences for noncompli-ance, and less likely to receive detention forminor misbehavior or moderate infractions.Latino students at the elementary level weremore likely to be suspended/expelled thanWhite students across all infractions exceptdisruption, and also more likely to receivedetention than White students for minor mis-behavior, noncompliance, and moderate in-fractions. Finally, Latino students were morelikely than White students to receive in-schoolsuspension for minor misbehavior, and lesslikely to receive in-school suspension fornoncompliance.

A slightly different pattern of infraction/consequences appears at the middle schoollevel. The overrepresentation of AfricanAmerican students in suspension/expulsion forspecific offenses is less consistent at the 6–9level, with ORs significantly greater than 1.00for only disruption, moderate infractions, andtardy/truancy. The pattern of African Ameri-can underrepresentation in less serious conse-quences was more pronounced at the 6–9level, however, with ORs significantly lessthan 1.00 for almost all less serious conse-quences across most types of infractions. His-panic/Latino disproportionality in suspension/expulsion at the 6–9 level appeared to be moreconsistent with elementary level findings, withORs significantly greater than one across alltypes of infractions except use/possession.Significant underrepresentation of Hispanic/Latino students was found in the use of mod-erate consequences for disruption and moder-ate infractions, and in in-school suspension forminor misbehavior and tardy/truancy.

Discussion

We conducted a disaggregated analysisof a detailed, nationally representative data setin order to provide a more comprehensivepicture of disproportionality in disciplineacross racial/ethnic categories and school lev-els. The results indicate that, across an exten-


95

Tab

le4

Mu

ltin

omia

lL

ogit

Reg

ress

ion

ofth

eIn

flu

ence

ofB

ehav

ior

and

Rac

eon

Adm

inis

trat

ive

Dec

isio

ns:

Odd

sR

atio

sa

Var

iabl

e

Mod

el1

Mod

el2

Det

entio

nM

oder

ate

Con

sequ

ence

In-S

choo

lSu

spen

sion

OSS

and

Exp

ulsi

onO

ther

/U

nkno

wn

Det

entio

nM

oder

ate

Con

sequ

ence

In-S

choo

lSu

spen

sion

OSS

and

Exp

ulsi

onO

ther

/U

nkno

wn

K–6

Scho

ols

Infr

actio

nsb

Min

orM

isbe

havi

ors

0.96

0.64

0.17

**0.

02**

0.03

**0.

880.

660.

17**

0.03

**0.

03**

Dis

rupt

ion

0.76

*1.

400.

790.

59**

0.01

**0.

73*

1.41

0.78

0.63

**0.

01**

Non

com

plia

nce

0.88

1.51

0.88

0.78

0.01

**0.

841.

530.

880.

850.

01**

Mod

erat

eIn

frac

tions

1.04

1.42

1.28

1.55

**0.

01**

0.96

1.44

1.28

1.73

**0.

01*

Maj

orV

iola

tions

1.19

2.53

**1.

44*

1.14

0.01

**1.

072.

60**

1.44

*1.

270.

01**

Use

/Pos

sess

ion

0.94

1.84

4.34

**16

.60*

*0.

02**

0.82

1.91

4.35

**18

.82*

*0.

02**

Oth

er/U

nkno

wn

1.33

*2.

25*

1.19

1.32

0.02

**1.

202.

30**

1.19

1.48

**0.

02**

Rac

eb

His

pani

c/L

atin

o1.

24**

0.71

**1.

011.

52**

0.64

**A

fric

anA

mer

ican

0.78

**1.

030.

991.

64**

1.03

Oth

er/U

nkno

wn

NA

NA

NA

NA

NA

Nof

case

s13

3,88

213

3,88

2M

odel

�2

33,8

9935

,681

Nag

elke

rke

Pseu

doR

20.

240

0.25

1%

Cor

rect

lypr

edic

ted

57.5

57.5


96

Tab

le4

Con

tin

ued

Var

iabl

e

Mod

el1

Mod

el2

Det

entio

nM

oder

ate

Con

sequ

ence

In-S

choo

lSu

spen

sion

OSS

and

Exp

ulsi

onO

ther

/U

nkno

wn

Det

entio

nM

oder

ate

Con

sequ

ence

In-S

choo

lSu

spen

sion

OSS

and

Exp

ulsi

onO

ther

/U

nkno

wn

6–9

Scho

ols

Infr

actio

nsb

Min

orM

isbe

havi

ors

0.48

**0.

27**

0.27

**0.

27*

1.06

0.49

**0.

27**

0.27

**0.

28**

1.07

Dis

rupt

ion

0.30

**0.

38**

0.58

**1.

030.

20**

0.33

**0.

40**

0.59

**1.

090.

20**

Non

com

plia

nce

0.29

**0.

41**

0.74

**1.

40**

0.24

**0.

31**

0.43

**0.

75**

1.45

**0.

24**

Mod

erat

eIn

frac

tions

0.30

**0.

67**

1.20

**6.

40**

0.33

**0.

31**

0.66

**1.

22**

6.70

**0.

34**

Maj

orV

iola

tions

0.38

**2.

04**

1.04

**6.

59**

0.50

**0.

38**

1.93

**1.

44**

6.89

**0.

50**

Use

/Pos

sess

ion

0.27

**1.

97*

2.05

**53

.01*

*0.

61*

0.25

**1.

74*

1.98

**55

.65*

*0.

58*

Oth

er/U

nkno

wn

0.49

**1.

20*

1.04

2.95

**0.

49**

0.51

**1.

19*

1.05

3.13

**0.

50**

Rac

eb

His

pani

c/L

atin

o1.

06*

0.77

**0.

991.

58**

0.90

**A

fric

anA

mer

ican

0.58

**0.

53**

0.82

**1.

12**

0.73

**O

ther

/Unk

now

nN

AN

AN

AN

AN

AN

ofca

ses

126,

760

126,

760

Mod

el�

228

,905

31,1

23N

agel

kerk

ePs

eudo

R2

0.21

10.

226

%C

orre

ctly

pred

icte

d32

.533

.3

Not

e.D

eten

tion

�de

tent

ion;

Min

orC

onse

quen

ces

�tim

ein

offic

e,lo

ssof

priv

ilege

s,co

nfer

ence

with

stud

ent,

pare

ntco

ntac

t,in

divi

dual

ized

inst

ruct

ion;

Mod

erat

eC

onse

quen

ces

�Sa

turd

aysc

hool

,bu

ssu

spen

sion

,re

stitu

tion;

In-S

choo

lSu

spen

sion

�in

-sch

ool

susp

ensi

on;

OSS

and

Exp

ulsi

on�

out-

of-s

choo

lsu

spen

sion

,ex

puls

ion;

Oth

er/U

nkno

wn

�ot

her

adm

inis

trat

ive

deci

sion

,un

know

nad

min

istr

ativ

ede

cisi

on;

K–

6�

kind

erga

rten

thro

ugh

Gra

de6;

6–9

�G

rade

6th

roug

hG

rade

9;N

A�

not

avai

labl

e.a R

efer

ence

cate

gory

isM

inor

Con

sequ

ence

sfo

rou

tcom

e.bR

efer

ence

cate

gory

isT

ardy

/Tru

ancy

for

Infr

actio

ns,

Whi

tefo

rR

ace/

Eth

nici

ty.

*p�

.05.

**p

�.0

1.


97

Tab

le5

Mu

ltin

omia

lL

ogit

Reg

ress

ion

ofth

eIn

flu

ence

ofR

ace

onA

dmin

istr

ativ

eD

ecis

ion

sin

Indi

vidu

alIn

frac

tion

:O

dds

Rat

iosa

Rac

eb

K–6

Scho

ols

6–9

Scho

ols

Det

entio

nM

oder

ate

Con

sequ

ence

sIn

-Sch

ool

Susp

ensi

onO

SSan

dE

xpul

sion

Oth

er/

Unk

now

nD

eten

tion

Mod

erat

eC

onse

quen

ces

In-S

choo

lSu

spen

sion

OSS

and

Exp

ulsi

onO

ther

/U

nkno

wn

Min

orM

isbe

havi

ors

His

pani

c/L

atin

o1.

22**

0.85

3.01

**2.

06*

0.51

**1.

050.

800.

36**

1.83

**0.

27*

Afr

ican

Am

eric

an0.

58**

1.24

*1.

76**

3.75

**1.

030.

70**

0.46

**0.

52**

0.95

0.62

*U

nkno

wn/

All

Oth

ers

NA

NA

NA

NA

NA

NA

NA

NA

NA

NA

N56

,884

23,1

81M

odel

�2

1,67

31,

209

Nag

elke

rke

Pseu

doR

20.

033

0.05

4

Dis

rupt

ion

His

pani

c/L

atin

o1.

271.

030.

781.

281.

361.

080.

63*

0.99

1.59

**1.

12A

fric

anA

mer

ican

0.94

0.94

0.83

*1.

54**

0.84

0.54

**0.

53**

0.93

1.35

**1.

19**

Unk

now

n/A

llO

ther

sN

AN

AN

AN

AN

AN

AN

AN

AN

AN

AN

8,20

318

,570

Mod

el�

252

498

Nag

elke

rke

Pseu

doR

20.

007

0.02

8

Non

com

plia

nce

His

pani

c/L

atin

o1.

30**

0.72

0.72

**1.

24*

1.06

1.09

0.82

1.05

1.25

**0.

94A

fric

anA

mer

ican

1.04

0.70

**0.

81**

1.22

**0.

920.

47**

0.44

**0.

88**

0.99

0.90

*U

nkno

wn/

All

Oth

ers

NA

NA

NA

NA

NA

NA

NA

NA

NA

NA

N21

,374

34,6

42M

odel

�2

141

887

Nag

elke

rke

Pseu

doR

20.

007

0.02

6


98

Tab

le5

Con

tin

ued

Rac

eb

K–6

Scho

ols

6–9

Scho

ols

Det

entio

nM

oder

ate

Con

sequ

ence

sIn

-Sch

ool

Susp

ensi

onO

SSan

dE

xpul

sion

Oth

er/

Unk

now

nD

eten

tion

Mod

erat

eC

onse

quen

ces

In-S

choo

lSu

spen

sion

OSS

and

Exp

ulsi

onO

ther

/U

nkno

wn

Mod

erat

eIn

frac

tions

His

pani

c/L

atin

o1.

27**

0.48

**0.

891.

59**

1.08

1.01

0.72

*1.

031.

74**

1.34

**A

fric

anA

mer

ican

0.91

*1.

32**

0.98

1.84

**1.

14*

0.52

**0.

59**

0.88

**1.

13**

1.28

**U

nkno

wn/

All

Oth

ers

NA

NA

NA

NA

NA

NA

NA

NA

NA

NA

N34

,792

24,5

20M

odel

�2

504

481

Nag

elke

rke

Pseu

doR

20.

015

0.02

0

Maj

orV

iola

tions

His

pani

c/L

atin

o1.

121.

010.

691.

87**

1.03

1.47

1.11

0.94

1.93

**0.

96A

fric

anA

mer

ican

0.85

0.73

0.90

2.02

**0.

870.

52**

0.45

**0.

801.

301.

14U

nkno

wn/

All

Oth

ers

NA

NA

NA

NA

NA

NA

NA

NA

NA

NA

N2,

978

2,12

0M

odel

�2

4784

Nag

elke

rke

Pseu

doR

20.

017

0.04

0

Use

/Pos

sess

ionc

His

pani

c/L

atin

o1.

420.

391.

002.

320.

67A

fric

anA

mer

ican

0.34

0.33

0.29

**0.

550.

38U

nkno

wn/

All

Oth

ers

NA

NA

NA

NA

NA

N1,

015

Mod

el�

240

Nag

elke

rke

Pseu

doR

20.

047


99

Tab

le5

Con

tin

ued

Rac

eb

K–6

Scho

ols

6–9

Scho

ols

Det

entio

nM

oder

ate

Con

sequ

ence

sIn

-Sch

ool

Susp

ensi

onO

SSan

dE

xpul

sion

Oth

er/

Unk

now

nD

eten

tion

Mod

erat

eC

onse

quen

ces

In-S

choo

lSu

spen

sion

OSS

and

Exp

ulsi

onO

ther

/U

nkno

wn

Oth

er/U

nkno

wn

His

pani

c/L

atin

o1.

320.

550.

62*

2.12

**0.

990.

940.

710.

933.

99**

1.70

**A

fric

anA

mer

ican

1.10

0.83

1.38

**1.

69**

0.61

**0.

57**

1.19

1.00

1.67

**0.

81*

Unk

now

n/A

llO

ther

sN

AN

AN

AN

AN

AN

AN

AN

AN

AN

AN

4,50

66,

328

Mod

el�

218

640

3N

agel

kerk

ePs

eudo

R2

0.04

20.

064

Tar

dy/T

ruan

cyc

His

pani

c/L

atin

o1.

34**

1.06

1.86

**1.

66**

2.28

**A

fric

anA

mer

ican

0.63

**0.

48**

0.62

**1.

37**

0.38

**U

nkno

wn/

All

Oth

ers

NA

NA

NA

NA

NA

N16

,384

Mod

el�

282

3N

agel

kerk

ePs

eudo

R2

0.05

1

Not

e.D

eten

tion

�de

tent

ion;

Mod

erat

eC

onse

quen

ces

�Sa

turd

aysc

hool

,bus

susp

ensi

on,r

estit

utio

n;In

-Sch

oolS

uspe

nsio

n�

in-s

choo

lsus

pens

ion;

OSS

and

Exp

ulsi

on�

out-

of-s

choo

lsu

spen

sion

,exp

ulsi

on;O

ther

/Unk

now

n�

othe

rad

min

istr

ativ

ede

cisi

on,u

nkno

wn

adm

inis

trat

ive

deci

sion

;K–

6�

kind

erga

rten

thro

ugh

Gra

de6;

6–9

�G

rade

6th

roug

hG

rade

9;N

A�

not

avai

labl

e.E

ach

OD

R/s

choo

lle

vel

com

bina

tion

inth

isan

alys

isre

pres

ents

ase

para

tem

odel

.Thu

s,w

ithse

para

teN

valu

esfo

rea

chm

odel

,it

isim

port

ant

tono

teth

atth

eod

dsra

tios

acro

ssm

odel

sar

eno

tdi

rect

lyco

mpa

rabl

e.a O

utco

me

refe

renc

eca

tego

ryis

Min

orC

onse

quen

ces.

bR

ace/

Eth

nici

tyre

fere

nce

cate

gory

isW

hite

.c M

odel

coul

dno

tbe

estim

ated

beca

use

ofze

roco

unts

inon

eor

mor

ece

lls.

*p�

.05.

**p

�.0

1.


100

sive national sample, significant disparities ex-ist for African American and Latino studentsin school discipline. Patterns are complex andmoderated by type of offense, race/ethnicity,and school level. Nevertheless, the overall pat-tern of results indicates that both initial refer-ral to the office and administrative decisionsmade as a result of that referral significantlycontribute to racial and ethnic disparities inschool discipline.

Across a national sample, AfricanAmerican students have twice the odds com-pared to White students of receiving ODRs atthe elementary level, and almost four times theodds of being referred to the office at themiddle school level. A different pattern ofdisproportionality emerges for Hispanic stu-dents, with significant overrepresentation(OR � 1.71) at the middle school level, butsignificant underrepresentation (OR � 0.76) atthe elementary school level. These results areconsistent with previous research indicatingubiquitous overrepresentation in school disci-pline for African American students, but in-consistent evidence of disparities for Latinostudents (Gordon, Della Piana, & Keleher,2000; Skiba & Rausch, 2006). It is possiblethat the striking shift found in the currentstudy from Hispanic under- to overrepresenta-tion in ODRs as one moves to the middleschool level may help explain some of theinconsistency in previous findings.

For African American students at theelementary school level, and for both AfricanAmerican and Latino students at the middleschool level, disparities in rates of referralwere widespread across referral types. Oneexplanation for racial and ethnic dispropor-tionality in school discipline is that such dis-parities are primarily a result of socioeco-nomic disadvantage (National Association ofSecondary School Principals, 2002); that is,African American students, overexposed tothe stressors of poverty, are more likely to beundersocialized with respect to school normsand rules. Yet previous research has found noevidence that disciplinary disproportionalitycan be explained to any significant degree bypoverty (Wallace et al., 2008; Wu et al.,1982). More important, there appears to be

little support for a hypothesis that AfricanAmerican students act out more in similarschool or district situations (McFadden et al.,1992; McCarthy & Hoge, 1987; Wu et al.,1982). The current results at the middle schoollevel, that the most likely types of ODR lead-ing to disparate African American disciplineare disruption and noncompliance, are consis-tent with a growing body of previous researchin suggesting that the types of referrals inwhich disproportionality is evident are mostlikely to be in categories that are more inter-active and subjectively interpreted, such asdefiance (Gregory & Weinstein, 2008) anddisrespect (Skiba et al., 2002).

One important premise of the presentresearch is that effective disciplinary systemsare more likely to have at their core a gradu-ated model, in which more serious conse-quences are reserved for more serious infrac-tions. Although zero tolerance disciplinaryphilosophy and practice has focused on “send-ing a message” to potentially disruptive stu-dents by applying relatively harsh punish-ments for both minor and more serious infrac-tions (Skiba & Rausch, 2006), reviews of theevidence regarding school discipline havefound little evidence supporting the effective-ness of such an approach (American Psycho-logical Association, 2008). Rather, a gradu-ated discipline model whereby the severity ofconsequences are scaled in proportion to theseriousness of the infraction, often in conjunc-tion with a tiered model of discipline (Sugai,2007), appears to hold far more promise as aneffective and efficient method for organizingschool disciplinary policy and practice.

Prior to disaggregation of the data(Model 1 in Table 4), the current data seem toshow some evidence of such a pattern of grad-uated discipline. That is, the odds for all stu-dents in this sample of receiving a suspensionor expulsion for minor misbehavior are ex-tremely low at both the elementary and middleschool levels, and gradually increase such thatit becomes highly likely that use and posses-sion of weapons or drugs will result in a sus-pension or expulsion. Although given the ab-sence of information on the extent of imple-mentation of SWPBS for this sample it is


101

impossible to know whether SWPBS imple-mentation was related to that finding, thesedata do suggest that many of the school disci-plinary systems in the current sample are ingeneral organized in a way that could be ex-pected to be efficient and effective.

In the logistic regressions regarding stu-dent infraction, consequence, and race/ethnic-ity (Table 4), both the various types of infrac-tions and the racial/ethnic categories enter theequation significantly at both the elementaryand middle school level. The failure of race tosignificantly affect the pattern of ORs for in-fractions means that the nature of the recordedinfraction is an important contributor to theseverity of consequences received, regardlessof student race or ethnicity. At the same time,the significance of race in predicting suspen-sion and expulsion even after the inclusion ofinfraction means that, regardless of the type ofinfraction, race/ethnicity makes a significantcontribution to the type of consequence cho-sen for a given infraction.

The pattern of differential treatment iseven more clearly articulated by the pattern ofadministrative decisions for various infrac-tions (Table 5). At the elementary schoollevel, African American students were morelikely than White students to be suspended orexpelled for any offense, and Latino studentsmore likely to be suspended for all offensesexcept disruption. In particular, African Amer-ican students have almost four times the odds,and Hispanic students twice the odds, of beingsuspended or expelled for a minor infraction atthe elementary school level. Although the pat-tern of overrepresentation in out-of-schoolsuspension/expulsion is somewhat less pro-nounced at the middle school level, there isstill substantial evidence of differential treat-ment. African American students were signif-icantly more likely than White students to besuspended or expelled for disruption, moder-ate infractions, and tardy/truancy, while La-tino students were more likely to be suspendedor expelled in Grade 6–9 schools for all in-fractions except use/possession. In addition,underrepresentation in the use of minor ormoderate consequences appears to be morepronounced at the middle school level, espe-

cially for African American students. Al-though specific patterns differ by race/ethnic-ity and school level, these findings are againconsistent with previous investigations (e.g.,McFadden et al., 1992; Shaw & Braden, 1990)that have found evidence of differential pro-cessing (Gregory et al., 2010) at the adminis-trative level.

In summary, these results suggest thatboth differential selection at the classroomlevel and differential processing at the admin-istrative level make significant contributionsto the disproportionate representation of Afri-can American and Latino students in schooldiscipline. For African American students,disproportionality at both the elementary andmiddle school levels begins at referral, mostparticularly in the areas of tardiness/truancy,noncompliance, and general disruption; forLatino students, disparities in initial ODRsemerge at the middle school level. Yet regard-less of previous disproportionality at referral,the type of infraction, or the school level, thefindings from this study indicate that studentsof different races and ethnicities are treateddifferently at the administrative level, withstudents of color being more likely to receivemore serious consequences for the same in-fraction. An investigation of extant data, thisstudy was able to identify the existence ofdisproportionate outcomes at the classroomand administrative level, but without local ob-servation, was unable to specify the classroomor school variables that create such imbal-ances. Further research, in particular ethno-graphic or observational studies that can iso-late specific teacher–student or administrator–student interactions, are essential forincreasing understanding of the variables con-tributing to racial and ethnic disparities inschool discipline.

The focus of this article was not onintervention per se, but these results may holdimportant implications for monitoring the ef-fects of interventions intended to address dis-ciplinary disproportionality. Although the ef-ficacy of SWPBS in reducing rates of ODRshas been well demonstrated (Barrett, Brad-shaw, & Lewis-Palmer, 2008; Bradshaw,Koth, Thornton, & Leaf, 2009; Bradshaw,


102

Mitchel, & Leaf, 2009 Horner et al., 2009;Nelson, Martella, & Marchand-Martella,2002; Safran & Oswald, 2003 Taylor-Green &Kartub, 2000), few investigations (Jones et al.,2006) have explored the issue of PBS andcultural variation, or sought to explore howthe application of such school-wide systemswill affect rates of disciplinary disproportion-ality. Kauffman, Conroy, Gardner, and Os-wald (2008) have argued that there is no evi-dence that behavioral interventions operatedifferently based on ethnicity, gender, or reli-gion, but also noted that differential effectsbased on race and ethnicity have been under-studied in the behavioral literature, and that“many studies in leading behavioral jour-nals. . . do not report sufficient detail about thecultural identities of participants” (p. 255).Until such time as a sufficient database hasbeen accumulated on interventions for reduc-ing disproportionate representation in schooldiscipline outcomes, it seems logical that im-plementations of interventions designed to af-fect student behavior in school should disag-gregate their results, in order to empiricallyexplore the extent to which those interventionswork equally well for all groups.

The current data demonstrate a markeddiscrepancy between the aggregated data, inwhich the severity of infraction and conse-quence are relatively well matched, and thedisaggregated data, showing that AfricanAmerican and Latino students receive moresevere punishment for the category “minormisbehavior.” Such a pattern of results isconsistent with emerging research on cultur-ally responsive pedagogy and classroommanagement (e.g., Harris-Murri, King, &Rostenberg, 2006; Serpell, Hayling, Steven-son, & Kern, 2009; Utley, Kozleski, Smith,& Draper, 2002) in suggesting that it cannotbe assumed that interventions intended toimprove behavior will be effective to thesame degree for all groups. Existing racialand ethnic differences in the use of currentdisciplinary interventions strongly indicatethat, for any intervention strategy aimed atreducing such disparities, disciplinary out-come data should be disaggregated, in orderto explicitly evaluate whether SWPBS, or

indeed any general intervention, is equallyeffective for all racial/ethnic groups.

Limitations

The present investigation was not ableto explicitly test the influence of SES on thetested relationships. Despite widespread be-liefs to the contrary, there is no previous evi-dence that the overrepresentation of AfricanAmerican or Latino students in school disci-plinary outcomes can be fully explained byindividual or community economic disadvan-tage (Skiba et al., 2002; Wallace et al., 2008;Wu et al., 1982). Further investigation isneeded, however, to parse the relative contri-bution of individual, classroom, and schoolcharacteristics to disciplinary disproportional-ity, including both SES and the complex ef-fects of gender (see e.g., Wallace et al., 2008).In addition, although we were able to explorevariations for different educational levels orracial/ethnic categories to arrive at a morecomplex rendering of disciplinary dispropor-tionality, we were not able to analyze the databy geographic location or school locale. Itseems highly likely that the variables contrib-uting to racial and ethnic disparities will varyconsiderably by location and locale, especiallyfor groups such as Hispanic students that haveshown inconsistency in previous research. Itmay well be that the specific causes of racialdisparities are regionally unique, requiring lo-cal analysis of causes and conditions (Skiba etal., 2008), in much the same way that func-tional behavior analysis is used at the individ-ual level in order to develop individualizedbehavior plans tailored to the needs of eachchild in each situation.

Finally, using school as the unit of anal-ysis restricted our ability to investigate thecontribution of prior infractions, a variablethat might well be expected to have a signifi-cant effect on administrative decisions regard-ing disciplinary consequences. It is importantto note, however, that there is no previousresearch we are aware of that explores theassociation of students’ prior record of schoolinfraction with racial and ethnic disproportion-ality in school discipline. There is an excep-


103

tionally long history in this nation of acceptinga stereotype of African Americans, especiallyAfrican American males, as being more proneto disordered behavior or criminality (see e.g.,Muhammed, 2010), often with little or no sup-porting evidence. With no evidence that sup-ports the notion that there are concurrentlyhigher levels of disruption among AfricanAmerican students, we see no reason to pre-sume that disparate rates of discipline betweenracial and ethnic groups can be explained bydifferential behavioral histories.

Summary and Recommendations

The fact of racial/ethnic disproportion-ality in school discipline has been widely and,we would argue, conclusively demonstrated.Across urban and suburban schools, quantita-tive and qualitative studies, national and localdata, African American and to some extentLatino students have been found to be subjectto a higher rate of disciplinary removal fromschool. These differences do not appear to beexplainable solely by the economic status ofthose students, nor through a higher rate ofdisruption for students of color.

Opportunity to remain engaged in aca-demic instruction is arguably the single mostimportant predictor of academic success. Inthe absence of an evidence-based rationalethat could explain widespread disparities indisciplinary treatment, it must be concludedthat the ubiquitous differential removal fromthe opportunity to learn for African Americanand Latino students represents a violation ofthe civil rights protections that have developedin this country since Brown v. Board of Edu-cation. We propose here that the existing em-pirical evidence for disproportional school dis-cipline by race, and the severe effect of exclu-sionary discipline on educational success,make disproportional application of exclusion-ary discipline an issue in need of immediateand substantive response. At the school level,(a) data on discipline by race should be re-ported regularly (monthly) to faculty, (b) pol-icies focused on prevention and culturally re-sponsive practice should be encouraged, and(c) investment in developing appropriate so-

cial behaviors should be made before resortingto exclusionary consequences. At the districtand state level, (a) disaggregated data on dis-cipline patterns should be available and dis-seminated, (b) policies addressing disciplinaryinequity and promoting equity should be es-tablished, and (c) personnel development op-tions should be made available to minimizethe disproportionate application of discipline.At the federal level, (a) research funding isneeded to move beyond mere description ofdisproportionality to clear documentation ofcausal mechanisms and functional interven-tions for reducing disparate outcomes; (b) re-sources are needed to document the technicalassistance and implementation strategies thatwill allow state- and district-wide responses todisproportionate use of discipline; and (c) as iscurrently the case for disproportionality inspecial education, federal monitoring practicesshould regularly require disaggregated report-ing of discipline patterns, and mandate thedevelopment and implementation of correctiveaction plans where disparities are found.

Racial and ethnic disparities that leavestudents of color behind remain ubiquitous inAmerican education. The national reportBreaking Barriers (Caldwell, Sewell, Parks, &Toldson, 2009; Toldson, 2008) found thatwhile personal, family, and community factorsall make a contribution to such disparities, sodo school and teacher characteristics, such asstudent perceptions of being respected andsupported by teachers, and perceptions ofschool safety. To the extent that the policiesand practices of schools maintain or widenracial gaps, it is imperative that policy makersand educators search for school-based solu-tions that can contribute to reducing racial andethnic disparities in important educationaloutcomes.

All children deserve access to effectiveeducational settings that are predictable, pos-itive, consistent, safe, and equitable. Access toeducational achievement requires the supportneeded to be socially successful in school.This typically involves not simply ensuringthat problem behavior is addressed equitably,but investing in building school cultures whereappropriate behavior is clearly defined, ac-


104

tively taught, and consistently acknowledged.For race to become a socially neutral factor ineducation, all levels of our educational systemmust be willing to make a significant invest-ment devoted explicitly to altering currentlyinequitable discipline patterns, to ensure thatour instructional and disciplinary systems af-ford all children an equal opportunity forschool learning.

References

American Psychological Association Zero Tolerance TaskForce. (2008). Are zero tolerance policies effective inthe schools? An evidentiary review and recommenda-tions. American Psychologist, 63, 852–862.

Barrett, S., Bradshaw, C., & Lewis-Palmer, T. (2008).Maryland state-wide PBIS initiative. Journal of Posi-tive Behavior Interventions, 10, 105–114.

Blanchett, W. J., Mumford, V., & Beachum, F. (2005).Urban school failure and disproportionality in a post-Brown era: Benign neglect of the constitutional rightsof students of color. Remedial and Special Education,26, 70–81.

Bradshaw, C., Koth, C., Thornton, L., & Leaf, P. (2009).Altering school climate through School-wide PositiveBehavioral Interventions and Supports: Findings froma group randomized effectiveness trial. Prevention Sci-ence, 10, 100–115.

Bradshaw, C., Mitchell, M., & Leaf, P. (2009). Examiningthe effects of school-wide positive behavioral interven-tions and supports on student outcomes: Results from arandomized controlled effectiveness trial in elementaryschools. Journal of Positive Behavior Interventions,12, 133–148.

Brantlinger, E. (1991). Social class distinctions in adoles-cents’ reports of problems and punishment in school.Behavioral Disorders, 17, 36–46.

Brophy, J. E. (1988). Research linking teacher behavior tostudent achievement: Potential implications for in-struction of Chapter 1 students. Educational Psychol-ogist, 23, 235–286.

Browne, J. A., Losen, D. J., & Wald, J. (2002). Zerotolerance: Unfair, with little recourse. In R. J. Skiba &G. G. Noam (Eds.), New directions for youth develop-ment (No. 92, Zero tolerance: Can suspension andexpulsion keep schools safe?) (pp. 73–99). San Fran-cisco: Jossey-Bass.

Caldwell, L. D., Sewell, A. A., Parks, N., & Toldson, I. A.(2009). Before the bell rings: Implementing coordi-nated school health models to influence the academicachievement of African American males. Journal ofNegro Education, 78, 204–215.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. (2003).Applied multiple regression: Correlation analysis forthe behavioral sciences (3rd ed.). Mahwah, NJ: Law-rence Erlbaum.

Donovan, M. S., & Cross, C. T. (Eds.). (2002). Minoritystudents in special and gifted education. Washington,DC: National Academies Press.

Elementary and Secondary Education Act. Pub. L. 89–10, 79 Stat. 77, 20 U.S.C. ch.70 (0000).

Ekstrom, R. B., Goertz, M. E., Pollack, J. M., & Rock,D. A. (1986). Who drops out of high school and why?Findings from a national study. Teachers College Re-cord, 87, 357–373.

Ferguson, A. A. (2001). Bad boys: Public schools and themaking of Black masculinity. Ann Arbor: University ofMichigan Press.

Finn, J. D. (1982). Patterns in special education placementas revealed by the OCR survey. In K. A. Heller, W. H.Holtzman, & S. Messick (Eds.), Placing children inspecial education: A strategy for equity (pp. 322–381).Washington, DC: National Academy of Sciences Na-tional Academy Press.

Gordon, R., Della Piana, L., & Keleher, T. (2000). Facingthe consequences: An examination of racial discrimi-nation in U.S. public schools. Oakland, CA: AppliedResearch Center.

Gottfredson, D. C., Gottfredson, G. D., & Hybl, L. G.(1993). Managing adolescent behavior: A multiyear,multischool study. American Educational ResearchJournal, 30, 179–215.

Greene, W. H. (2008). Econometric analysis (6th ed.).Upper Saddle River, NJ: Pearson/Prentice Hall.

Greenwood, C. R., Horton, B. T., & Utley, C. A. (2002).Academic engagement: Current perspectives on re-search and practice. School Psychology Review, 31,328–349.

Gregory, A., Skiba, R., & Noguera, P. (2010). Theachievement gap and the discipline gap: Two sides ofthe same coin? Educational Researcher, 39, 59–68.

Gregory, J. F. (1997). Three strikes and they’re out: Af-rican-American boys and American schools’ responsesto misbehavior. International Journal of Adolescenceand Youth, 7(1), 25–34.

Gregory, A., & Weinstein, S. R. (2008). The disciplinegap and African Americans: Defiance or cooperation inthe high school classroom. Journal of School Psychol-ogy, 46, 455–475.

Harris-Murri, N., King, K., & Rostenberg, D. (2006).Reducing disproportionate minority representation inspecial education programs for students with emotionaldisturbances: Toward a culturally responsive responseto intervention model. Education and Treatment ofChildren, 29, 779–799.

Hawkins, J. D., Doueck, H. J., & Lishner, D. M. (1988).Changing teaching practices in mainstream classroomsto improve bonding and behavior of low achievers.American Educational Research Journal, 25, 31–50.

Horner, R. H., Sugai, G., Smolkowski, K., Eber, L., Na-kasato, J., Todd, A., & Esperanza, J. (2009). A ran-domized, waitlist-controlled effectiveness trial assess-ing school-wide positive behavior support in elemen-tary schools. Journal of Positive Behavior Interventions,11(3), 133–144.

Individuals with Disabilities Education Improvement Actof 2004. Pub. L. No. 108–446, 20 U.S.C. 1400 et seq.(2004).

Jones, C., Caravaca, L., Cizek, S., Horner, R., & Vincent,C. G. (2006). Culturally responsive schoolwide posi-tive behavior support: A case study in one school witha high proportion of Native American students. Multi-ple Voices for Ethnically Diverse Exceptional Learn-ers, 9, 108–119.

Kauffman, J. M., Conroy, M., Gardner, R., & Oswald, D.(2008). Cultural sensitivity in the application of behav-


105

ior principles to education. Education and Treatmentof Children, 31, 239–262.

Ladson-Billings, G. (2006). From the achievement gap tothe education debt: Understanding achievement in U.S.schools. Educational Researcher, 35, 3–12.

May, S., Ard, W., Todd, A. W., Horner, R. H., Glasgow,A., Sugai, G., et al. (2006). School-wide informationsystem. Eugene: Educational and Community Sup-ports, University of Oregon.

McCarthy, J. D., & Hoge, D. R. (1987). The social con-struction of school punishment: Racial disadvantageout of universalistic process. Social Forces, 65, 1101–1120.

McFadden, A. C., Marsh, G. E., Prince, B. J., & Hwang,Y. (1992). A study of race and gender bias in thepunishment of handicapped school children. UrbanReview, 24, 239–251.

McLoyd, V. C. (1998). Socioeconomic disadvantage andchild development. American Psychologist, 53, 185–204.

Morrison, G. M., Anthony, S., Storino, M., Cheng, J.,Furlong, M. F., & Morrison, R. L. (2001). Schoolexpulsion as a process and an event: Before and aftereffects on children at-risk for school discipline. NewDirections for Youth Development: Theory, Practice,Research, 92, 45–72.

Muhammed, K. G. (2010). The condemnation of black-ness: Race, crime, and the making of modern urbanAmerica. Cambridge, MA: Harvard University Press.

National Association of Secondary School Principals.(2000, February). Statement on civil rights implica-tions of zero tolerance programs. Testimony presentedto the United States Commission on Civil Rights,Washington, DC.

Neal, L. I., McCray, A. D., Webb-Johnson, G., &Bridgest, S. T. (2003). The effects of African Ameri-can movement styles on teachers’ perceptions and re-actions. Journal of Special Education, 37, 49–57.

Nelson, J. R., Martella, R. M., & Marchand-Martella, N.(2002). Maximizing student learning: The effects of acomprehensive school-based program for preventingproblem behaviors. Journal of Emotional and Behav-ioral Disorders, 10, 136–148.

No Child Left Behind Act of 2001, 20 U.S.C. § 6319(2008).

Oswald, D. P., Coutinho, M. J., Best, A. M., & Singh,N. N. (1999). Ethnic representation in special educa-tion: The influence of school-related economic anddemographic variables. The Journal of Special Educa-tion, 32, 194–206.

Pigott, R. L., & Cowen, E. L. (2000). Teacher race, childrace, racial congruence, and teacher ratings of chil-dren’s school adjustment. Journal of School Psychol-ogy, 38, 177–196.

Raffaele Mendez, L. M., & Knoff, H. M. (2003). Whogets suspended from school and why: A demographicanalysis of schools and disciplinary infractions in alarge school district. Education and Treatment of Chil-dren, 26, 30–51.

Safran, S. P., & Oswald, K. (2003). Positive behaviorsupports: Can schools reshape disciplinary practices?Exceptional Children, 69, 361–373.

Serpell, Z., Hayling, C. C., Stevenson, H., & Kern, L.(2009). Cultural considerations in the development ofschool-based interventions for African American ado-

lescent boys with emotional and behavioral disorders.Journal of Negro Education, 78, 321–332.

Shaw, S. R., & Braden, J. P. (1990). Race and gender biasin the administration of corporal punishment. SchoolPsychology Review, 19, 378–383.

Skiba, R. J., & Rausch, M. K. (2006). Zero tolerance,suspension, and expulsion: Questions of equity andeffectiveness. In C. M. Evertson & C. S. Weinstein(Eds.), Handbook of classroom management: Re-search, practice, and contemporary issues (pp. 1063–1089). Lawrence Erlbaum Associates.

Skiba, R. J., Michael, R. S., Nardo, A. C., & Peterson, R.(2002). The color of discipline: Sources of racial andgender disproportionality in school punishment. UrbanReview, 34, 317–342.

Skiba, R. J., Simmons, A. B., Ritter, S., Gibbs, A. C.,Karega Rausch, M., Cuadrado, J., et al. (2008).Achieving equality in special education: History, sta-tus, and current challenges. Exceptional Children, 74,264–288.

Spaulding, S., Irvin, L., Horner, R., May, S., Emeldi, M.,Tobin, T., & Sugai, G. (2010). School-wide socialbehavioral climate, student problem behavior, and ad-ministrative decisions: Empirical patterns from 1510schools nationwide. Journal of Positive BehavioralInterventions, 12, 69–85.

Sugai, G. (2007, August). Response to intervention ap-proach to school-wide discipline. Paper presented atthe OSEP Response to Intervention Summit, Washing-ton, DC.

Sugai, G., & Horner, R. H. (2006). A promising approachfor expanding and sustaining school wide positive be-havior support. School Psychology Review, 35, 246–259.

Sugai, G., Horner, R. H., & Gresham, F. (2002). Behav-iorally effective school environments. In M. R. Shinn,G. Stoner, & H. M. Walker (Eds.), Interventions foracademic and behavior problems: Preventive and re-medial approaches (pp. 315–350). Silver Spring, MD:National Association for School Psychologists.

Sugai, G., & Lewis, T. J. (1999). Developing PositiveBehavioral Support Systems. In G. Sugai & T. J. Lewis(Eds.), Developing positive behavioral support for stu-dents with challenging behaviors. Council for Childrenwith Behavioral Disorders, 15–23.

Taylor-Greene, S. J., & Kartub, D. T. (2000). Durableimplementation of school-wide behavior support: Thehigh five program. Journal of Positive Behavior Inter-ventions, 2, 233–245.

Todd, A. W., Haugen, L., Anderson, K., & Spriggs, M.(2002). Teaching recess: Low-cost efforts producingeffective results. Journal of Positive Behavioral Inter-ventions, 4, 46–52.

Toldson, I. A. (2008). Breaking barriers: Plotting thepath to academic success for school-age African-American males. Washington, DC: CongressionalBlack Caucus Foundation.

Townsend, B. L. (2000). The disproportionate disciplineof African-American learners: Reducing school sus-pensions and expulsions. Exceptional Children, 66,381–391.

Utley, C. A., Kozleski, E., Smith, A., & Draper, I. L.(2002). Positive behavior support: A proactive strategyfor minimizing behavior problems in multiculturalyouth. Journal of Positive Behavior Interventions, 4,196–207.


106

Vavrus, F., & Cole, K. (2002). “I didn’t do nothin’”: Thediscursive construction of school suspension. The Ur-ban Review, 34, 87–111.

Wald, J., & Losen, D. J. (2003). Defining and redirectinga school-to-prison pipeline. In J. Wald & D. J. Losen(Eds.), New directions for youth development (No. 99;Deconstructing the school-to-prison pipeline) (pp.9–15). San Francisco: Jossey-Bass.

Wald, J., & Losen, D. J. (2007). Out of sight: The journeythrough the school-to-prison pipeline. In S. Books(Ed.) Invisible children in the society and its schools(3rd ed.) (pp. 23–27). Mahwah, NJ: Lawrence ErlbaumAssociates.

Wallace, J. M., Jr., Goodkind, S. G., Wallace, C. M., &Bachman, J. (2008). Racial/ethnic and gender differ-ences in school discipline among American highschool students: 1991–2005. Negro Educational Re-view, 59, 47–62.

Wu, S. C., Pink, W. T., Crain, R. L., & Moles, O. (1982).Student suspension: A critical reappraisal. The UrbanReview, 14, 245–303.

Zimmerman, R. S., Khoury, E. L., Vega, W. A., Gil,A. G., & Warheit, G. J. (2006). Teacher and parentperceptions of behavior problems among a sample ofAfrican American, Hispanic, and Non-Hispanic Whitestudents. American Journal of Community Psychology,23, 181–197.

Zumwalt, K., & Craig, E. (2005). Teachers’ characteris-tics: Research on the indicators of quality. In M. Co-chran-Smith & K. M. Zeichner (Eds.), Studyingteacher education: The report of the AERA Panel onResearch and Teacher Education (pp. 111–156). Mah-wah, NJ: Lawrence Erlbaum.

Date Received: September 23, 2009Date Accepted: January 6, 2011

Action Editor: George Bear �

Russell J. Skiba is Professor in the School Psychology Program and Director of the EquityProject at Indiana University. His current research focus includes racial and ethnicdisproportionality in school discipline and special education, and developing effectiveschool-based systems for school discipline and school violence.

Robert H. Horner is Almuni-Knight Endowed professor of Special Education at theUniversity of Oregon. His research focuses on severe disabilities, instructional design,positive behavior support, and implementation science.

Choong-Geun Chung is Statistician at the Center for Evaluation and Education Policy atIndiana University in Bloomington. His research interests are analytical issues in racialand ethnic disproportionality in special education and in school discipline, and statisticalmodels for access and persistence in higher education.

M. Karega Rausch currently serves as the Director of the Office of Education Innovationfor Indianapolis Mayor Greg Ballard. He is currently completing his doctorate in schoolpsychology at Indiana University-Bloomington. Rausch’s research interests include eq-uity in school discipline and special education.

Seth L. May is a research assistant and software developer at Educational and CommunitySupports in the University of Oregon’s College of Education. His efforts currently focuson developing research databases and large software systems for managing, collecting,and reporting on student behavior.

Tary J. Tobin is a Research Associate and an Adjunct Assistant Professor of SpecialEducation at the University of Oregon. Her current research interests include staffdevelopment, cultural and linguistic diversity, and use of prevention science and tech-nology in school and community interventions.


107

Copyright of School Psychology Review is the property of National Association of School Psychologists and its

content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for individual use.

race is not neutral: a national investigation of african ...equity/docs/skiba et al race is not...

Documents