Tuesday, October 4, 2011

Racial Disparities in Special Education

If you are a parent in the Ann Arbor Public Schools, last week you got a letter that was nearly unintelligible.

You might have wondered what that meant--I know I did, and I know that others asked me what it meant.

Let's start with the basics. School districts are supposed to track and explain whether, or if, some racial/ethnic groups are disproportionately enrolled in special education program. There's a really nice explanation of how this gets calculated in this document, but if you don't want to read the whole thing, here is a thumbnail sketch. (If you understand risk ratios, skip the part that is in red.)

The whole point of using these statistics is to describe and understand sub-groups in the school environment. Don't be scared that math is involved.

Composition: Let's say that 10% of the district population is Hispanic/Latino, but 20% of the students who are identified as having a disability are Hispanic/Latino. Then we know that there is a disparity, because twice as many Hispanic/Latino students are enrolled in special education compared to the Hispanic/Latino population. (This is a good basic number, but it may be less accurate when we are talking about groups that make up a small percentage of a district's population.)

Risk Ratio: A Risk Ratio compares one population (say, African American students diagnosed with autism) to a comparison group. According to what I read, generally the risk comparison that is preferred is to compare one population group to the risk for all other students combined, because that way you can use a consistent comparison group for all calculations.
So. . . the equation for the risk ratio would be:

Risk ratio = Risk for racial/ethnic group for disability category / Risk for comparison group for disability category

Let's say that the autism risk for Black students is 4%, but for all other students is 2%. Then the Risk Ratio is 2.0. In other words, in this example, Black students are two times more likely than all other students to receive services for autism. A risk ratio of 1.0 means there is no difference between the two groups. (Risk ratio >1 equals increased risk for the identified sub-group; risk ratio <1 equals decreased risk for the identified sub-group.)

Weighted Risk Ratios allow you to compare data across school districts, where (as we know) the racial/ethnic composition varies greatly. Read about the math of the weighted risk ratios in the link I gave you earlier--but they are similar to Risk Ratios in the way that they work: weighted risk ratio >1 equals increased risk for the identified sub-group; weighted risk ratio <1 equals decreased risk for the identified sub-group.

So--now that we've taken a statistical detour, what exactly did that letter that we got mean?

1. We know that it was written because it was state-mandated.
1. We know that it was based on data from 2009-2010.
2. The issues raised center on suspensions, expulsions, and the category of students labeled as "cognitively-impaired."
3. The first few times I read the letter, I assumed that the problem with suspensions and expulsions was specific to African American students, but let's note that the letter doesn't say that. I therefore believe that this is a problem for more than one group. The letter also doesn't state the weighted risk ratio for suspensions and expulsions.
4. In the cognitive impairment group, African Americans are more than 3 times more likely than the statewide comparison group (of all other students, I believe--although it is not stated in the letter!) to be identified as cognitively impaired.

To put it baldly, the letter is terribly-written, and it was clearly written just to meet the letter of the law and not the spirit.  I'd say that the district missed an important chance to really explain (not just in vague platitudes) what it is doing to improve things.

The district writes:
"Ann Arbor Public Schools was also determined to have significant disproportionality because..."

The English teacher in me notes the passive tense here--who exactly is responsible for the problem?

The letter doesn't explain this at all, but I dug up an old Ann Arbor News article which says that in 2007-2008, 47% of the students who were identified as cognitively impaired were African American.  The article (David Jesse, 3/12/2009) says, "the district was not complying with federal and state law in the way it determined which students were cognitively impaired." So was misdiagnosis involved? The letter doesn't say. If kids were misdiagnosed, were they moved out of the "cognitive impairment" group and is that responsible for the slight improvement in numbers?

Based on the ratios disclosed in the letter, I believe that while things may have improved somewhat, I am not sure they have improved at all significantly. And in using the word "significantly," here, I'm talking about statistical significance. Is it just a fluke the numbers have dropped slightly or are the results related to something specific?

Further, the letter only addresses the areas of disparity which reached a state-mandated benchmark. There are other areas of disparity--many other areas of disparity--in special education, that fall short of the benchmark.

But any time there is a big disparity, we need to ask why the disparity exists:

1. Is it because different races/ethnicities are differentially diagnosed? In other words, is the bar for "cognitive impairment" set equally for everbody?

Translation: This could be due to obvious or subtle racism. Historically, kids who had trouble switching from Black English to Standard English were sometimes labeled cognitively impaired or placed in other special education categories.

2. Does the disparity represent a real difference? In other words, using the same exact standards, are more African Americans in the Ann Arbor schools cognitively impaired than other kids in the Ann Arbor schools?

Translation: The numbers could be describing real differences. For instance--more African American babies are born prematurely than White or Hispanic/Latino babies, and premature babies are more likely to grow up to have cognitive impairments.

3. Does the racial disparity mask a different disparity?

Translation: Is this problem really about poverty, and because proportionally more African American kids live in poverty, a disparity based on class shows up as a disparity based on race? For instance--do poor housing conditions lead to increased exposure to lead paint and thereby to cognitive impairments?

You can read an earlier post I wrote on disparities here, but to cut to the chase: racial disparities have been with the district for over 30 years. There are no easy solutions, or we would have solved the problem.

On the other hand, I think the conversations about how to solve the problems need to take place with more parental and taxpayer involvement than has been the case previously. This letter is a weak start.


  1. Thank you for all this digging and clarifying!

    In order to have more parental and taxpayer involement, the district has to (or we have to) find a way to make this a parental and taxpayer concern. This seems to be the wall that seems so impossible to climb - if your child does well, this is not your concern, if you don't have a child in the system, this is not your concern, etc... I feel the district had an opportunity here (and elsewhere) to educate about why this is eveyone's concern and botched it by (in this case) sending out a letter of gobilty gook.

    So I truly appreciate your taking the time to explain what this letter means - perhaps AAPS should copy and send out what you have written to the community.

  2. Don't shoot the messenger, scary enough I think I understood what she was saying, and I especially like the emphasis on PD, positive behavior interventions, and continue analysis of implementation...disporportionality is a real problem at this district. Either a kid is in or out in this district.

  3. Regarding point 3, there is a lot of evidence that racial achievement gaps exist independent of other demographic factors. The McKinsey Achievement Gap report, for example, points out that test scores for non-poor blacks are lower than scores for poor whites. There may be other disparities at play, but race does seem to be a factor in why our schools don't educate students of color as well as they do white students.