On understanding inconsistent disciplinary behaviour in schools Authors: Anton Bekkerman & Gregory Gilpin This is an Accepted Manuscript of an article published in Applied Economics Leters in 2015, available online: htp:/www.tandfonline.com/10.1080/13504851.2014.978065. Bekkerman, Anton, and Gregory Gilpin. "On understanding inconsistent disciplinary behaviour in schools." Applied Economics Leters 22, no. 10 (2015): 772-776. DOI: htps:/dx.doi.org/10.1080/13504851.2014.978065. Made available through Montana State University’s ScholarWorks scholarworks.montana.edu On understanding inconsistent disciplinary behaviour in schools Anton Bekkerman and Gregory Gilpin Department of Agricultural Economics and Economics, Montana State University, Bozeman, MT 59715–2920, USA Inconsistent disciplinary administration across schools can inequitably impact students’ education access opportunities by separating certain students from familiar learning environments, especialy in misconduct cases that result in longer-term removal. We empiricaly estimate whether such inconsistencies are atributable to heterogeneity in student body demographic characteristics. The results indicate that a greater number of disciplines that remove students from school for an extended period of time are observed in schools with a higher proportion of black students, but no significant diferential punishment efects are observed in schools with a higher Hispanic student population. Furthermore, results of decom- posing the marginal efects into conditional and unconditional elasticities indicate that it is not the case that schools with predominantly white student bodies have the least severe punishments and schools with more minority students have the most severe punishments. Rather, schools with inconsistent disciplinary behaviour have a proportion of the inconsistency atributable to the race of the student body. I. Introduction Students unable to access their familiar education setings are more likely to experience diminished learning outcomes and are less likely to remain in school (Finn,1989; Fitzpatricket al.,2011). Bernburg and Krohn (2003) show that these efects are also evident when students are removed from schools due to misconduct discipline decisions, implying that schools’punishment practices can play a role in students’educational opportunities. Schools often have some discretion in their disciplin- ary decision-making process, which could lead to inequitable punishment outcomes if this process is influenced by factors that are not directly related to the evaluation of misconducts, such as student body demographic and socio-economic characteristics. Previous studies provide some evidence that school administrators’disciplinary decisions are afected by factors external to a misconduct case, including stu- dents’racial characteristics (e.g. see Welch and Payne,2010; Kinsler,2013). However, these studies evaluate inconsistencies in short-term punishment decisions, which have less impact on students’abil- ity to atend school. Only limited research exists on potential inconsistencies in punishments that remove students for extended time periods, which can have significantly greater consequences on access to education. This study investigates whether disciplinary inconsistencies across U.S. public high schools can be atributed to diferences in the student body char- acteristics for misconducts that lead to long-term or permanent removal from the school. Wefind evi- dence that inconsistencies exist and that they, in part, can be atributed to the racial composition of the student body. Specificaly, in schools with a higher proportion of black students, more miscon- ducts are punished by long-term or permanent removal and more severe discipline is imposed in lieu of lesser punishments. Additionaly, we decom- pose thisfinding to show that disciplinary variation atributable to student body demographics is only identified in schools where punishment inconsisten- cies exist, rather than in schools that simply have a higher proportion of minority students. I. Data and Empirical Specification Table 1presents the variables used in this study, which represent combined data from the restricted-access versions of the 2003–2004, 2005–2006, 2007–2008 and 2009–2010 School Survey on Crime and Safety (SSOCS) for 3200 U.S. public high schools, the U.S. Census Bureau, the U.S. Bureau of Labor Statistics and the Federal Bureau of Investigation’sUniformCrime Reports. The SSOCS provides information about misconducts reported to school administrators, which were commited on school grounds, buses, or at school-sponsored events. These misconducts are categorized as one of four ofenses: the pos- session or use of a nonfirearm weapon, ilegal drugs, alcohol or being involved in a physical altercation with an intent to harm others.1Due to the seriousness of these misconducts, punish- ments were either an expulsion or transfer of a student to an alternative school (permanent removal), a suspension exceedingfive days (pro- longed removal) or a short-term discipline (e.g. detention, suspension of fewer than 5 days, no punishment). Table 2provides preliminary insights about the heterogeneity in disciplinary decisions across schools with diferent student demographics. As expected, more serious misconducts are more severely punished, but across al ofense categories, schools with significant black or Hispanic student bodies (i.e. schools where black or Hispanic student bodies are at least 25%) expel or transfer students approximately 3–15 percentage points more fre- quently than schools with a higher proportion of white students. Moreover, schools with a higher pro- portion of black and Hispanic students are more likely to use tougher punishments than schools with predominantly white student bodies, where suspen- sions are more prevalent than expulsions or transfers. We separately model punishments across each of the four misconduct categories as the proportion of misconductjin schooliduring school yeartthat was assigned a discipline typek,dkijt¼ # Disciplines kijt # Misconductsijt . This measure is modeled as a function of student body characteristics (SBit), school-level and Table 1. Variables in the empirical analysis Dependent variablea Misconduct management/ prevention programs Expulsion/transfer Parental input program Suspension, >5 days Student mentoring Short-term discipline Assistance for parents Student body characteristicsb Perceived crime prevention limitations Black Hispanic Eligible for free lunch Below 15th percentile of standardized tests Inadequate funding to implement programs Fear of state reprisal Limited access to alternative placement Fear of student retaliation Gang activity Community characteristics Median income ($1000 s of 2010 dolars) Metropolitan location indicator Adult crime rate (per 1000 adults) Rural location indicator Unemployment rate Notes:aDiscipline per misconduct. bIn per cent. Supplemental data provides detailed definitions for these variables. 1The SSOCS also includes information aboutfirearm ofenses. It is excluded from our analysis because the 1994 Gun-Free Schools Act mandates federal standardized punishments for these ofenses, significantly reducing schools’discretion in disciplining these misconducts. These results are available upon request. where the unobserved latent discipline outcome is dkijt. A Tobit model with sampling weights provided by the SSOCS yields unbiased and consistent para- meters estimates. The Tobit model conveniently provides an oppor- tunity to decompose marginal efects into condi- tional and unconditional elasticities between discipline outcomes and the variables of interest (McDonald and Moffit,1980), that is, εkjðCÞ¼ @E½05 days Black −1.241*** −0.396* 0.906* 0.437*** Hispanic−0.612 −0.362 −0.221 −0.086 Short-term discipline Black 0.218 0.170 −0.369*−0.360*** Hispanic 0.152 0.407**−0.081 0.046 Note: *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels. 2We do not present elasticity estimates with respect to the Hispanic student body because the Tobit models’marginal efect estimates indicate a weak relationship between disciplinary inconsistencies and changes in the Hispanic student population. Table 3. Tobit discipline model estimates of latent marginal efects IV. Conclusions While cals for greater discretionary power for evalu- ating student misconduct have increased, inconsistent disciplinary behaviours across schools can inequitably impact students’access to education opportunities. We show that a portion of observed disciplinary inconsistencies can be atributed to diferences in the student body’s racial composition and that these efects are concentrated in schools with larger black student bodies. Moreover, wefind that diferential disciplinary behaviour is not unconditionaly related to student demographics; rather, student body racial compositions are more likely to influence disciplinary decisions in schools that already have variability in making those decisions. Consequently, public policy that seeks to reduce educational inequities must be crafted with particular atention to the complexities underlying schools’existing predispositions to disci- plinary inconsistencies. Supplemental data Supplemental data for this article can be accessed at htp:/dx.doi.org/10.1080/13504851.2014.978065 References Bernburg, J. and Krohn, M. (2003) Labeling, life chances, and adult crime: the direct and indirect efects of official intervention in adolescence on crime in early adulthood, Criminology,41, 1287–318. doi:10.1111/j.1745-9125.2003.tb01020.x Finn, J. (1989) Withdrawing from school,Review of Educational Research,59,117–42. doi:10.3102/ 00346543059002117 Fitzpatrick, M. D., Grissmer, D. and Hastedt, S. (2011) What a diference a day makes: estimating daily learning gains during kindergarten andfirst grade using a natural experiment, Economics of Education Review,30, 269–79. doi:10.1016/j. econedurev.2010.09.004 Kinsler, J. (2011) Understanding the black-white school discipline gap,Economics of Education Review,30, 1370–83. Kinsler, J. (2013) School discipline: a source or salve for the racial achievement gap?,International Economic Review,54, 355–83. doi:10.1111/ j.1468-2354.2012.00736.x McDonald,J.and Moffit, R. (1980) The uses of tobit analysis, The Review of Economics and Statistics,62, 318–21. doi:10.2307/192 4766 Welch, K. and Payne, A. (2010) Racial threat and punitive school discipline,Social Problems,57,25–48. doi:10.1525/sp.2010.57.1.25 Table 4. Discipline elasticity estimates with respect to the per cent of the black student body Nonfirearm weapon Ilegal drugs Alcohol Physical altercations Expulsion/transfer Unconditional elasticity 0.021 0.008 0.013 0.028 Conditional elasticity 0.001** 0.001** 0.001* 0.003*** Suspension, >5 days Unconditional elasticity −0.038 −0.016 0.013 0.022 Conditional elasticity −0.001*** −0.001*** 0.001* 0.002*** Short-term discipline Unconditional elasticity 0.011 0.011 −0.019 −0.023 Conditional elasticity 0.000 0.001 −0.001* −0.002*** Note: *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels.