Department of Education Equity Report
(June 27, 2013; rev. July 13, 2013)
This is a subpage to the Discipline Disparities page of jpscanlan.com. That page and all its subpages are related to subject this page. Also relevant to the subject of this page is the Educational Disparities. For the lack of understanding of statistics that is the specific subject of this page is also pertinent to the agency’s efforts to monitor disparities related to academic performance.
***
The Department of Education’s actions with respect to discipline disparities make clear that it does not understand how to analyze data on demographic differences in outcome rates. Specifically, it encourages public schools to relax discipline standards in order to reduce racial differences in discipline rates. Yet, just as lowering a test cutoff (thereby lowering overall failure rates) will tend to increase relative differences in failure rates while reducing relative differences in pass rates, relaxing discipline standards (thereby lowering overall discipline rates) will tend to increase relative differences in discipline rates. Unaware that reducing the frequency of suspensions and expulsions will tend to increase relative differences in discipline rates, the Department continues to monitor the fairness of discipline policies on the basis of relative differences in discipline rates.
Recent articles explaining this matter include “Misunderstanding of Statistics Leads to Misguided Law Enforcement Policies” (Amstat News, Dec. 2012); “Racial Differences in School Discipline Rates” (The Recorder, June 22, 2012). The Los Angeles SWPBS subpage of the Discipline Page discusses the way a program aimed at reducing discipline rates in Los Angeles led to larger racial differences in discipline rates. The Suburban Disparities subpage discusses the pattern whereby relative differences in suspension rates tend to be larger in suburban schools (where suspension are less common) than in city schools. The Oakland Agreement subpage discusses the Department of Education agreement with the Oakland Unified School District that calls for the general reduction in suspensions and expulsion while providing for that monitoring of disparities that will certainly be done in terms of relative differences in rates of suspension and expulsion.
The Department’s lack of understanding in this area is no more serious than that found in the health and healthcare disparities research of Harvard Medical School and the Harvard School of Public Health (see Harvard University Measurement Letter (Oct. 9, 2012)), the enforcement of various civil rights laws including those related to school discipline by the Department of Justice (see United States Department of Justice Measurement Letter (Apr. 23, 2012)), and the enforcement of Fair Lending Laws by the Federal Reserve Board (Federal Reserve Board Measurement Letter (March 4, 2013). But the DOE’s lack of understanding may have broader effects.
The Office of Civil Rights November 2012 document titled “Helping to Ensure Equal Access to Education: Report to the President and Secretary” is a typical reflection of the agency’s failure to understand the statistical issues. Notably, it site five states (Colorado, Texas, Virginia, North Carolina, and Delaware) as having recently taken up reforms of zero tolerance policies, noting that many cited data from the Civil Rights Data Collection project in support of their initiatives.
But the report does provide some useful information for appraising the Department of Education’s belief that zero tolerance policies lead to large racial differences in discipline rates. At page 28, it notes both that in the CRDC sample blacks comprised 18 percent of students and 39 percent of student expelled and that in districts with at least one expulsion under zero tolerance blacks comprised 19 percent of students and 33 percent of students expelled.
One would prefer to have the actual expulsion rates and also to have data broken down completely by race in order that one could compare the black and white rates in the two settings. But, as discussed in the Harvard University Measurement Letter (at 27), one can derive from these figures the ratio of the black expulsion rate to the expulsion rate of other students. Table 1 shows those ratios. It shows that, as one with a sound understanding of statistics would expect, the ratio is larger overall than it is for zero tolerance settings. But that is the opposite of what the Department of Education has been telling the public is the case and it is the opposite of the perceptions of the jurisdictions that are relaxing school discipline policies because of concerns about large relative differences in discipline rates.
Table 1: Proportions Blacks Students Comprise of Students and Expelled Students Overall and in Districts with Zero Tolerance Expulsions with Ratio of Black Expulsion Rate to White Expulsion Rate
|
|
Situation
|
Black Percent of Students
|
Black Percent of Expulsions
|
Black/Non-Black Expulsion Ratio
|
|
Overall
|
18%
|
39%
|
2.91
|
|
Zero Tolerance Schools
|
19%
|
33%
|
2.10
|
As discussed in the Harvard Letter, data solely on representation among the potentially affected population and the affected population do not allow one to calculate relative differences for failing to be expelled. But one can assume that the relative difference in failing to be expelled would be smaller overall than at zero tolerance schools.
In other circumstances, there could be reason for concern that fact that the overall ratio is greater overall than at zero tolerance schools is simply a function of the disproportionate concentration of black students in zero tolerance schools, where expulsion rates of black and whites are generally higher than at other schools. Table 2 illustrates a hypothetical situation where notwithstanding that the black/white expulsion ratio is smaller at non-zero tolerance schools than at zero tolerance schools, because of the greater representation of blacks at zero tolerance schools, the overall black/white expulsion ratio is smaller than at zero tolerance schools.
Table 2: Hypothetical Illustration of Situation Where Relative Differences in Overall Expulsion Rates Could be Larger than Relative Differences in Expulsion Rates at Zero Tolerance Schools Even Though Relative Differences in Expulsion Rates are Greater at Zero Tolerance Schools than at Other Schools.
|
|
Type
|
Num B Stud
|
Num W Stud
|
B Perc of Stud
|
Num B Expulsions
|
Num W Expulions
|
B Perc of Expulsions
|
Perc of B Expelled
|
Perc of W Expelled
|
B/W Exp Rate Ratio
|
|
Zero
|
3000
|
10000
|
23.08%
|
600
|
1000
|
37.50%
|
20.00%
|
10.00%
|
2.00
|
|
Non-Zero
|
1000
|
10000
|
9.09%
|
100
|
700
|
14.29%
|
10.00%
|
7.00%
|
1.43
|
|
Total
|
4000
|
20000
|
16.67%
|
700
|
1400
|
33.33%
|
17.50%
|
7.00%
|
2.50
|
But while the pattern shown in Table 2 is possible in circumstances where blacks comprise a much higher proportion of students in zero tolerance schools than other schools, the fact that, as shown in Table 1, blacks comprise approximately the same proportion of students at zero tolerance schools that they comprise overall suggests that that larger rate ratio overall than at zero tolerance schools is not a function of the disproportionate concentration of black students at zero tolerance schools.