Study Purpose. In response to school violence, American schools have adopted strict discipline policies and often implemented school suspension. At the same time, there have been growing concerns over frequent use of school suspension. Of particular concern are persistent racial-ethnic disparities. Some scholars interpret such disparities as evidence of differential treatment against minority students. Others explain them with compounding effects of a 3rd factor. They posit that minority students are vulnerable to school suspension because they are more likely, than white students, to display problem behaviors, have family disadvantages, or attend unsafe and disorderly schools. The study purpose is to twofold: one is to determine the effects of student race/ethnicity on school suspension; the other is to test the 3rd factor hypothesis. If the association between student race/ethnicity and school suspension is fully explained by the 3rd factor(s), logical grounds of the differential treatment hypothesis will be challenged. If not, the differential treatment hypothesis can be considered plausible.

Method. The data for this study were drawn from Education Longitudinal Study of 2002. The study sample is composed of 8,424 white, 1,942 African-American, and 2,149 Hispanic high school sophomores. The primary analysis method is logistic regression where the outcome variable is whether a student has been suspended during the past semester.

Results. Student race/ethnicity has significant effects on school suspension. Compared to white students, African-American students were approximately three times more likely (odd ratio = 2.97, p < .001), and Hispanic students were two times more likely (odd ratio = 1.96, p < .001) to be suspended. When student problem behavior was added to the baseline logistic regression model, the odds for the African-American and Hispanic students decreased to 2.52 and 1.35, respectively. Yet, the effects remained significant (p < .001). Similarly, including family SES in the baseline model attenuated the likelihood of school suspension among racial-ethnic minority students (2.35 for African-Americans & 1.40 for Hispanics), but the effects were still significant (p < .001). Adding school factors to baseline model also reduced the odds of racial-ethnic minority students (2.52 for African-Americans & 1.35 for Hispanics), but did not remove their significances (p < .001). When student problem behavior and family SES were added to the baseline model simultaneously, Hispanic students' higher likelihood of school suspension, compared to white students, became insignificant, but there was no change in African-American students' significance. Even when student problem behavior, family SES, and school characteristics were all taken into account, African-American students' likelihood of school suspension is two times higher than that of white students at the alpha level of .001.

Implications. The study results show that 3rd factors explained Hispanic students' vulnerability to school suspension fully but African-American students' only partially. With regard to African-American students, therefore, the differential treatment hypothesis is plausible. To remedy racial-ethnic disparities in school suspension, a combination of interventions is needed: micro-level to help at-risk students adapt to the situation, macro-level to reduce racial-ethnic gaps in SES, and meso-level to increase cultural competency among school teachers and administrators.