ODR Reductions and Ethnicity

Do elementary schools that document reductions in overall office discipline referrals document reductions across all student races and ethnicities?

Evaluation Brief

Claudia G. Vincent, Gwendolyn Cartledge, Seth May, and Tary J. Tobin1

It is a sobering and increasingly undeniable reality that, compared to their white peers, students from minority backgrounds, especially African-American students, are referred to the office more frequently (Cartledge & Johnson, 2004; Lo & Cartledge, 1996; Raffaele Mendez & Knoff, 2003; Skiba, Michael, Nardo, & Peterson, 2002), are referred for less severe offenses (Skiba et al., under review), are suspended or expelled more frequently (Skiba, Peterson, & Williams, 1997), and tend to be suspended for longer durations (Tobin & Vincent, under review; Vincent & Tobin, under review).

Efforts to make disciplinary practices equitable for students from all racial-ethnic backgrounds include the use of data collection systems that encourage disciplinary consistency across students and teachers and allow schools to review potential trends in their office discipline referral (ODR) data across student groups, locations, times, or behaviors. The School-wide Information System (SWIS; www.swis.org; May et al., 2003) is one ODR data collection system that relies on operational, mutually exclusive, and exhaustive definitions of behavioral violations to minimize potential bias in disciplinary practices due to varying interpretations of student behavior. In SWIS, two groups of behavioral violations exist: minor problem behaviors and major problem behaviors. Minor problem behaviors include low-intensity defiance, low-intensity disruption, inappropriate language, and inappropriate physical contact. Major problem behaviors include abusive language, insubordination, sustained disruption, and fighting.

To assess equity of their disciplinary practices across student ethnicities, SWIS users have the option to record ethnicity information for their overall school enrollment, as well as for each student who is entered into SWIS because he or she received an ODR. Prior to the 2009-2010 academic year, SWIS users could choose from the following racial-ethnic categories: (1) American Indian/Alaskan Native, (2) Asian, (3) African American, (4) Hispanic/Latino, (5) Pacific Islander/Native Hawaiian, (6) White, (7) Unknown, and (8) Not Listed. Because of the new federal race and ethnicity categories to be used by schools starting in 2010-2011, SWIS is currently revising its race and ethnicity categories to match federal guidelines. Beginning in 2009-2010, SWIS users will be able to choose from the following race and ethnicity categories: (1) American Indian/Alaskan Native, (2) Asian Race, (3) African American Race, (4) Hispanic Ethnicity, (5) Native Hawaiian or Other Pacific Islander Race, (6) White, and (7) Multiple Races (Dickey, 2009). Each ODR recorded in SWIS is linked to the race and ethnicity information recorded for the student who received the ODR. ODR data from schools that use SWIS and agree to share their data for research purposes become part of an extensive research database that lends itself to examining trends in disciplinary data across multiple years.

Given the overwhelming evidence of inequitable disciplinary practices and the availability of an extensive database of ODR, our goal was to examine if elementary schools that record a decrease of their overall ODR rate across a 3 year time span in SWIS show similar decreases in ODR rates for students from different racial and ethnic backgrounds.


Our sample included elementary schools that (a) were located in the United States, (b) were not labeled as “alternative” (c) continuously used SWIS during 3 consecutive academic years (2005-06 to 2007-08), (d) recorded race and ethnicity information for their overall school enrollment as well as for individual students, and (e) recorded an overall major ODR rate during 2007-08 that was at least 10% lower than their overall major ODR rate during 2005-2006. We chose to focus on major ODRs, because they tend to generate administrative decisions that can be of significant consequence to the student. Our focus on elementary schools allowed us to extract a maximally large sample, because the majority of SWIS users are elementary schools.

A total of 69 elementary schools met all inclusion criteria. These 69 schools were located in 16 states; the most frequently represented states were Illinois and Maryland with 14 (20.3%) schools each, followed by Missouri and North Carolina with 8 (11.6%) schools each. Table 1 provides an overview of the schools’ enrollment by race and ethnicity across the 3 years, as well as the total number of ODRs and mean ODR rates for each academic year.

Table 1: Student enrollment by race and ethnicity, total number of major ODR per year, and mean ODR rates per year.










Nat American





















Afr American










































Number of major ODR




Mean rate/100 students/day (SD)

.38 (.30)

.31 (.27)

.23 (.22)

To assess if overall reductions in major ODR meant reductions in major ODR for students from varying ethnicities, we calculated the following two metrics for each ethnicity: Number of major ODR per 100 students enrolled [(number of major ODR/student enrolled)*100] and number of students with major ODR per 100 students enrolled [(number of students with major ODR/students enrolled)*100]. Number of major ODR indicates the frequency with which a school gives referrals to students from a specific race or ethnicity; individual students might receive multiple ODRs. Number of students with major ODR indicates the number of unique students from a specific race or ethnicity that received at least one major ODR. To assess the extent to which ODR were proportionately distributed across student races and ethnicities in our sample, we calculated the percent of students enrolled and the percent of students with major ODR by ethnicity.

As follow-up analyses, we also examined if there were differences in ODR patterns due to gender within racial-ethnic categories. Because SWIS does not record overall school enrollment by gender, we had to merge SWIS ODR data with enrollment data from the National Center for Education Statistics (NCES) which reflect enrollment by gender within each race and ethnicity. No NCES data were available for 3 schools; thus the sample size for the follow-up analyses was n = 66. Because NCES data are lagged by 2 years, enrollment by gender data for only the 2005-2006 and 2006-2007 academic years were available.

Because African-American students tend to be disciplined more than any other student group and Latino students are the fastest growing student group (Gandara & Contreras, 2009), our follow-up analyses focused on comparing these student groups to White students.


Overall, results indicated that elementary schools that documented a reduction in their overall major ODR rates also showed reductions in numbers of major ODR for each racial-ethnic category of interest as well as reductions in number of students receiving major ODR. Figure 1 presents changes in major ODR/100 students by race and ethnicity as well as by gender for African-American, Latino, and White students. The pattern of reductions from year to year appears to hold across most races and ethnicities. African-American students clearly received the highest number of ODR within each year.

Figure 1: Major ODR per 100 students enrolled by race and ethnicity and by gender within race and ethnicity.

Disaggregation of the data by gender within ethnicity showed that the vast majority of major ODRs as well as the majority of reductions involved male students. Male students received approximately 4-5 times as many ODR as did female students; this ratio did not appear to change across the two years for which data were available. Of all male students, African-American boys clearly received the greatest number of ODR in both years. Of all female students, African-American girls clearly received the greatest number of ODR in both years.

Figure 2 presents changes in number of students with major ODR per 100 students enrolled. The pattern of reductions from year to year appears to hold for all races and ethnicities; however, among all students with ODR, African-American students were clearly the most frequently represented. In year 1, approximately 32 of 100 African-American students received a major ODR, while approximately 15 of 100 White students did. In year 3, approximately 24 of 100 African-American students received an ODR, while approximately 12 of 100 White students did.

Disaggregation of the data by gender within race or ethnicity showed that among all male students with major ODR, African-American boys were the most frequently represented, and among all female students with major ODR, African-American girls were the most frequently represented.

Figure 2: Students with major ODR per 100 students enrolled by race and ethnicity and by gender within race and ethnicity.

Figure 3 provides an estimate about the proportionality of students with major ODR relative to their overall enrollment for African-American, Latino, and White students. African-American students were clearly over-represented among students with major ODR in all 3 years, while Latino and White students were underrepresented. Disaggregation of the data by gender within race and ethnicity showed that African-American boys were clearly over-represented among students with major ODR in the 2 years for which data were available, while Latino and White boys were approximately proportionately represented. African-American girls were approximately proportionately represented in the two years for which data were available, while Latino and White girls were under-represented.

Figure 3: Percent of students enrolled and percent of students with major ODR by race and ethnicity and by gender within race and ethnicity.


It is encouraging to see that patterns of reductions in overall ODR rates apply to students from varying racial-ethnic backgrounds. It is less encouraging to see that the disproportionately high number of African-American students with ODR persisted despite overall reductions in ODR. Our analysis by gender within ethnicity indicated that these disproportionately high numbers of students with ODR affected primarily African-American boys. Our dataset did not include information about schools’ disciplinary practices beyond usage of SWIS. However, our outcomes underscore the necessity for research into culturally sensitive disciplinary practices. A number of questions present themselves. For example: Why might operationally defined behavioral violations as they are used in SWIS lead to differential outcomes for African-American students? Do behavioral definitions on which ODR are based take the racial-ethnic backgrounds of a school’s population into consideration? Do students and teachers interpret inappropriate behaviors differently? What might contribute to potential differences in interpretation? How might those potential differences be reconciled? Are the schools’ disciplinary practices adjusted for the cultural differences of their school population? How could such an adjustment occur conceptually and practically? Much work remains to be done to create schools where all students regardless of their race or ethnicity and gender can succeed behaviorally. Future research might focus on examining data from schools and districts with consistently low rates for all student groups.

It is important to note that our dataset had a number of limitations that compromise interpretability of outcomes. Between 5% and 10% of the ODR in our dataset were linked to students whose ethnicity was recorded as “not listed or “unknown,” and approximately 2% of schools’ overall enrollment was recorded as “not listed” or “unknown” in SWIS. Because these students could belong to any ethnic category, the calculated ratios (number of ODR/students enrolled, students with ODR/students enrolled) likely contain error. In addition, discrepancies between NCES and SWIS enrollment data commonly exist. While these discrepancies are usually small, the merging of the two data sources also likely introduced error into our analysis.


Cartledge, G. & Johnson, C.T. (2004). School violence and cultural sensitivity. In J. C. Conoley & A. P. Goldstein (Eds.) School violence intervention: A practical handbook. 2nd ed. (pp. 441-482). New York: Guilford Press.

Dickey, C.R. (2009). SWIS ethnicity categories 2009-2010. Retrieved on October 14, 2009 from http://www.swis.org/index.php?page=resources;rid=10139

Gandara, P. & Contreras, F. (2009). The Latino education crisis. Cambridge: Harvard University Press.

Lo, Y. & Cartledge, G. (2006). FBA and BIP: Increasing the behavior adjustment of African American boys in schools. Behavioral Disorders 31(1), 147-161.

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

Raffaele Mendez, L. M. & Knoff, H. M. (2003). Who gets suspended from school and why: A demographic analysis of schools and disciplinary infractions in a large school district. Education and Treatment of Children, 26, 30-51.

Skiba, R.J., Horner, R.H., Chung, C., Rausch, M.K., Seth, M., & Tobin, T. (2008). Race is not neutral: A national investigation of African American and Latino disproportionality in school discipline. Manuscript submitted for publication.

Skiba, R.J., Michael, R.S., Nardo, A.C., & Peterson, R. (2002). The color of discipline: Sources of racial and gender disproportionality in school punishment. Urban Review, 34, 317-342.

Tobin, T.J. & Vincent, C.G. (under review). Strategies for preventing disproportionate exclusion of students from minority backgrounds.

Vincent, C. G. (2008). Do schools using SWIS take advantage of the “school ethnicity report”? Evaluation Brief. Retrieved on May 14, 2009 fromhttp://pbis.org/evaluation/evaluation_briefs/default.aspx

Vincent, C.G. & Tobin, T.J. (under review). The relationship between implementation of school-wide positive behavior support and disciplinary exclusion of students from various ethnic backgrounds with and without disabilities.

1We gratefully acknowledge the contributions of Dr. Jennifer L. Frank to the development of this brief.