I will protect your pensions. Nothing about your pension is going to change when I am governor. - Chris Christie, "An Open Letter to the Teachers of NJ" October, 2009

Friday, October 31, 2014

UPDATED: TEAM/KIPP Charter Academy's Special Education Data: The Facts

UPDATED: TEAM responds, as do I. See below.

There's been a big to-do over social media about my latest report on New Jersey charter schools, co-authored with Julia Sass Rubin. I have come to loathe debate by Twitter, so let me spell out my thoughts here:

I'm not one for playing games with data. I'm not going to tell you I have some special, secret stash of numbers and that you have to trust me because I have the real true facts, and anyone who says otherwise must live in the suburbs and hate city children...

What I'm about to do is something anyone can do quite easily with a copy of Excel and a few spreadsheets you can download here. Everything in this analysis is completely replicable. If I got something wrong, tell me and I'll fix it. I am an open book.


First, download the 2013 "District Classification Rates, Ages 3-21" file from the New Jersey Department of Education's Special Education Data webpage. Pick out two relevant data points: the district classification rate for the Newark Public Schools, and TEAM Academy Charter School, the Newark branch of the national charter management organization, KIPP.

Then make your graph:

You can go on all day about how there are some NPS schools that have a special education rate lower than 17.8%, but that's completely irrelevant. The district, as a whole, must educate more students who are classified with a special education need than TEAM. The money has to come out of the district's budget, the qualified teachers and other staff have to be assigned out of the district's workforce, and the facilities and equipment have to be purchased by the district. In addition, the test scores for the district as a whole are very likely impacted by this higher concentration of students with special needs.


Next, go get the 2013 "Placement Data By Eligibility Category" file for Ages 6 - 21. Open it up and look at the reporting for TEAM Academy:

The "Eligibility" column has an abbreviation for each of 12 categories of special education disability:
  • AUT: Autism 
  • DB: Deaf Blindness 
  • EMN: Emotional Disturbance 
  • HI: Hearing Impairment 
  • MD: Multiple Disabilities 
  • ID: Intellectual Disability 
  • OHI: Other Health Impairment 
  • OI: Orthopedic Impairments 
  • SLD: Specific Learning Disability*
  • SPL: Speech or Language Impairment*
  • TBI: Traumatic Brain Injury 
  • VI: Visual Impairment
The row tells us the placement of the students for each eligibility: 12 x 7 = 84 cells. I've marked SPL and SLD to show these are lower-cost eligibilities, as confirmed by a report commissioned by the NJDOE itself. 

You'll notice 13 dashes in this dataset. This represents "suppressed" data, ostensibly hidden to protect the privacy of students. As the file itself explains:
Note:  Cells sizes containing counts of students 5 or less have been suppressed and are marked with "-" symbol.  
But cells with "0" have not been suppressed. So we know that every suppressed cell has at least a "1," and potentially a "5." Let's start filling in the blanks by putting a "1" into every suppressed cell.

You'll notice I moved the rows around a little, but the numbers are still all the same. I wanted SPL and SLD -- remember, those are the lower-cost disabilities -- to be together. You'll see why in a minute.

Now, we also know that the total count here for ages 6-21 can't be greater than the count for ages 3-21 in the previous file. That file said TEAM had a total of 275 classified students for 2013, which means if we total up every cell in this dataset, it can't be more than 275. That means we are left with 28 students whose special education eligibility is unknown.

From this, it's easy to make a chart showing how TEAM's special education students break out:

Remember: those "unknown" special education students can only go into cells that are suppressed. Which means TEAM has no hearing impaired or visually impaired students, and no emotionally disturbed students on campus (I am very curious as to why TEAM is placing at least one child into a private school; is this a data error?). There are open cells for SLD and SPL, but let's make a gigantic assumption and say that all of those 10 percent of "unknown" students are in higher-cost eligibilities.

That means TEAM's lower-cost eligibility rate is 66 percent. 

Let's now do the same thing for NPS:

To be as generous as possible to TEAM in this analysis, I simply put a "1" in every suppressed cell in a higher-cost eligibility and left it. Then I put a "5" in every lower-cost eligibility cell and left that (those cells, incidentally, are for out-of-district placements, which can be expensive -- but we won't quibble). I am, therefore, quite likely overstating the lower-cost percentage, and understating the higher-cost percentage. Even still...

NPS's lower-cost eligibility rate is 53% -- that's 13 percentage points lower than TEAM. 

So what does this publicly available, entirely replicable data tell us?

1) TEAM Academy does not serve as many children proportionately with a special education need as NPS.


2) The special education students TEAM does serve are more likely to have lower-cost disabilities than NPS's special education population.

This is not an "attack" -- it is a look at the facts, and it has profound consequences. If, for whatever reason, TEAM is not serving the same proportion of classified students and the same proportion of higher-cost classified students, that affects both the budget and the test-based outcomes of NPS.

I don't know why anyone would be surprised by any of this. It makes no sense to think TEAM would take on children who were far along on the autism spectrum, or who had severe emotional disturbances, or who were profoundly hearing impaired. And it makes no sense to think the district wouldn't, to at least some degree, try to concentrate these children within certain buildings, leaving others with below-average higher-cost special education rates. 

I am a huge proponent of special education inclusion, but I know these children often need personalized instruction in addition to time mainstreamed with general education students. Putting a program for children with a particular learning disability into one building with general education students creates the best of both worlds: personalized instruction and a chance to assimilate.

But you can't expect a charter to do this. You shouldn't expect a charter to do this: they just don't have the scale to be able to make it work. It's not their fault, and it doesn't mean they don't care about special education students; it just means they're not set up for the job.

Is this really so controversial? Why are we so surprised that in a "choice" model, the families of students who are similar to each other "choose" to go to the same schools?

To my fellow educators over at TEAM: no one -- well, me at least -- is accusing you of putting your thumb on the scale. True, there have been some really outrageous examples of charters engaging in practices that influence their student enrollments. But I've never heard any of them attributed to you, and I'd never accuse you of engaging in them without hard evidence.

But the facts are the facts. Yes, this data may be noisy and dirty, but it's all I've got, and it tells the same story time and again: you're not educating the same types of students as NPS.


ADDING: From the response below by a TEAM representative:
Your comparison here is off-base because you include "Public Separate & Private Day" counts. That includes NPS special ed schools like JFK, Bruce, out-of-district placement, etc. 
No. No, no, no.

You must include those placements, because the district pays the costs. NPS could hire more staff and open another building or two and educate those students in-house, but they've chosen to instead contract out their education. This is money that comes out of the general fund, just like payments to charters.

TEAM is trying to have it both ways: they want to be independent like a school district, getting their payment from NPS and spending it however they wish. But then they want to be considered like just another neighborhood school when comparing special education rates. Sorry, you don't get to choose what you are as it suits your purposes.

NPS is a system of schools that work together but are subject to a central authority when resources are allocated. TEAM is not part of that system. NPS can't say to TEAM: "Oh, sorry, need to cut your budget back because we need more money for out-of-district placements." NPS also can't say: "Oh, sorry, we're making you take all the autistic students in the larger area for an inclusion program."

TEAM runs its lottery, takes its kids -- who may stay or may leave -- picks up its check from NPS, and goes its merry way. No NPS school is run in this matter. Any analysis must acknowledge this difference.

You can read the entire comment from TEAM, and my response, in the comments below.

ADDING MORE: Also, read this from Bruce Baker. This notion of charters having it both ways fiscally has got to stop.


Andrew said...

Hi, Andrew from TEAM here.

Your comparison here is off-base because you include "Public Separate & Private Day" counts. That includes NPS special ed schools like JFK, Bruce, out-of-district placement, etc.

I don't think anyone is shocked to hear that TEAM isn't running a special ed campus yet - NPS is 10x bigger, and they only have a handful. As Ryan said on twitter, we're currently considering running a special campus for kids with extraordinary needs, like JFK

For low-incidence low-incidence, high-needs conditions, everyone needs to share in the costs. NJ does this via the extraordinary aid system, and there a bunch of funding streams available to NPS that charters don't see -- more discussion at http://www.uaedreform.org/wp-content/uploads/2014/charter-funding-inequity-expands-nj.pdf

TEAM kids are Newark kids. Our population looks a lot like the median district school. A more reasonable comparison would be to use the first three columns of those reports -- they cover all the traditional NPS schools. I did the calcs for 2013 using this method, and I get 58% low cost for NPS, 66% for TEAM. 8% difference.

I shared our current numbers (2014 submission) already -- we are at 57% low cost this year.

I don't appreciate your tone re: 'playing games with data' and 'special, secret stash'. That's not what's happening here. I'm quoting you figures from our student enrollment system - the legal system of record for our kids. Those numbers are to-the-minute (state publications are often a year or more behind), and you don't have to make any assumptions about cells with '--'. Those are our kids.

You conclude that "special education students TEAM does serve are more likely to have lower-cost disabilities than NPS's special education population." I don't think that's what the data shows - in 2013, we're talking about an 8% difference (or 22 kids out of an enrollment of 2200+), and in 2014, our high cost SPED share is precisely even with NPS's number from last year.

Classification is a proxy for the special ed needs of your students, but it's not 1:1 --
NPS themselves say they over-classify students as special ed. http://www.nps.k12.nj.us/wp-content/uploads/2014/08/SABBudgetFinal1314ReadOnly.pdf
We work extremely hard to staff our schools to serve special education kids, and meet the needs of all kids.

Giuseppe said...

The latest talking point for the charter school propagandists seems to be:"...but, boo hoo, we don't get the same funding as the traditional public schools, sob, sob.." Well, if you don't like the funding then don't open a charter school or go begging to the many billionaires who are for school privatization. Some of those supposed "miracle" charter schools in NYC get significant funding from the hedge fund billionaires who hate the real public schools.

Giuseppe said...

Charter school propagandists assert that they are public schools because they get public money except for the times when they claim to be private entities. From edweek.org: "So let's end this hoax. Charter schools are happy to accept public dollars, but reject the oversight and accountability that comes with operation as a public school. As the California Charter School Association insists, they are private entities. As Diane Ravitch suggests, if they are going to claim that in court, then that is good enough for me. They are private entities. Not public schools."

Giuseppe said...

It's not just NJ. From the miamiherald.com: From South Dade to the northern reaches of Broward County, only a handful of students with profound disabilities make it into charter schools, according to a Miami Herald/StateImpact Florida analysis of student enrollment data. The trend holds true across the state, where 87 percent of charter schools don’t serve any students with the most intense support needs. [snip] The trend has also been observed in New York, Los Angeles and New Orleans, where the few children with intense support needs who are in charter schools are clustered in schools that specialize in their disabilities.

“If we had a similar pattern of exclusion of kids by gender or race, there would be much more outrage than there is,” said Harvard University Professor Thomas Hehir, who headed the federal Department of Education’s office of special education under President Bill Clinton. [snip] Hehir and others say it comes down to money.

“There is a disincentive [for charter schools] to enroll these kids because they cost more money to educate,” Hehir said.

Read more here: http://www.miamiherald.com/news/special-reports/cashing-in-on-kids/article1939221.html#storylink=cpy

Read more here: http://www.miamiherald.com/news/special-reports/cashing-in-on-kids/article1939221.html#storylink=cpy

Read more here: http://www.miamiherald.com/news/special-reports/cashing-in-on-kids/article1939221.html#storylink=cpy

Duke said...


1) Your current data is useless for a comparison to NPS without the same data reported in the same way from NPS. So let's not even go there unless and until we can do that. Sorry you didn't like the snark, but this is an obvious point, and it's frustrating to have to make it.

2) Will the state require suppression of data under the same rules as last year? If so, publishing this data with cells of counts under 5 is, frankly, unethical.

3) Using "Public Separate & Private Day" is not only relevant, it is necessary. NPS bears the cost of these placements, as they bear the costs of the charters. The relevant question is whether charters should be getting the amount they get when they are not responsible for educating the most expensive children, which the district must pay to place in special schools.

In the district presentation you reference, you'll notice NPS spent over $50 million on out-of-district tuition, all coming from the general fund. That's over a third of the cost of charter payments. This matters.

4) Regarding the UArk report:


Bruce spells out why you need to include the students who are being placed out of district in the comparison. The funding streams are not the issue. Look at Figure 1.

5) The eligibilities aside, you still leave out the second part of the difference between TEAM and NPS - the district's higher overall classification rate. So even if you were serving the same proportion of high-cost students (you aren't), you'd still be serving proportionately fewer of these students.

6) Talking about TEAM as a school, and not a district, masks the issue I explain in the post. It's perfectly reasonable to assume NPS will cluster special education students in particular schools for reasons of efficiency. We must, therefore, look at the district special education percentages in any comparison that is used to make inferences about effects on budgets.

7) NPS may casually say it over-classifies its children with no evidence whatsoever presented. You'll forgive me when, given the district's track record:


I don't simply take that statement as fact without supporting proof. There is, frankly, no incentive for the district to do so, as SFRA allocates special education funds based on a census approach:


I've heard this argument quite a bit from the "choice" set, and it's well past time to start backing it up with some concrete evidence.

Unknown said...

What's amazing about Andrew's analysis is he takes the most postive spin you could possibly take from the numbers by selectively deciding what numbers get included, and STILL his charter doesn't match up to the district. So even if we have to accept your reasoning for choosing the numbers you did, although JJ has done a stellar job explaining why you can't just choose the numbers you did, in the most positive light you could frame the data, charters STILL not serving the same population.

John Thompson said...

I note that TEAM has 40% fewer students with emotional disturbances. This, more than the costs, is most important for teachers with large percentages of IEP and other high-challenge students, who have more children with conduct disorders.

Andrew replies that his school is close to the median in Newark. The real issue is how would they do with equally large concentrations of kids from extreme poverty who have endured extreme trauma.

Andrew said...

Hi Mark,

So there are two questions that we're talking about here - representative student body, and cost. Let's break them out individually.

Representative student body
TEAM schools are free, open enrollment schools. Our enrollment coordinator works really hard to make sure that we get the word out about our schools (flyers, door tags, etc) in the needier neighborhoods in Newark. We visit Title 1 pre-school programs to make sure that low-income families know about our schools. Why do we do this? Because the people in our schools believe pretty profoundly in expanding educational opportunity, and want to reach families that need us most. This is a sentiment that I've seen you characterize as insincere or somehow not genuine, but it describes our people to a T. You can read an interview with one of our school leaders here. This is why we were so supportive of the Universal enrollment program - it puts all schools on the same footing, and actually gives low-income kids a boost when matching kids and schools.

Our kids are Newark kids. Our student body looks a lot like a regular neighborhood school. We think that's the right target to shoot for - the median Newark school. When asked about demographics, that's been a pretty consistent response. We run two middle schools. We're less than a tenth as big as NPS. If I'm not mistaken, NPS only runs one or two of those campuses. If I'm not mistaken, you won't find JFK in the NJASK files, because all or nearly all of their students qualify for alternative assessments with their IEPs. It's just not the right comparison from a policy perspective. There's a reason why the state breaks out out-of-district sped numbers into a separate column -- it's an apples and oranges thing.

We are working really hard to do the right thing, which is why we defend our record when people raise the question. TEAM is committed to meeting the needs of special education students - and we staff our schools accordingly. Our mix of students with special needs this year has the same proportion of high/low cost kids as the NPS average for all the neighborhood schools.

You raise a separate question about cost allocation. With respect, I think that you should revise your language above, because it's not a very good description of how special education funding works in New Jersey.

It's true that "the district pays the cost", in the sense that the district does cut a check for kids in special placements.

But there's a special revenue stream precisely created to offset those expenditures - the extraordinary aid system. I mentioned this in my comment above. I am not 100% familiar with the current regs, but I think that the formula is that the state reimburses 90% of the cost for extraordinary special ed expenditures above and beyond 40K in expenditures - sort of like an insurance program. This is a totally sensible policy solution for high-cost, low-incidence special ed classifications. It distributes costs across districts, so that everyone shares the cost burden.

A final thought on tone: could you please take it down a notch? Throwing out accusations like 'unethical' is kind of a conversation stopper. We ran those counts by our lawyer and he said that they were OK to publish. The state has the N < 10 policy in place because when releasing large, multi-dimensional (by race, by income, etc etc) tables counts can sometimes become personally identifiable. That's not the case here - there's nothing personally identifiable in what I shared.

Duke said...

Andrew, let's make this short:

- I don't believe I have ever questioned the commitment of TEAM's staff to their students. If I have, show me where I did, and I will apologize and correct the assertion.

- As I said above: NPS is a system of schools. TEAM and the other charters are not in that system. It is, therefore, wholly appropriate to look at TEAM and other charters as their own systems when determining the impact of disparate student populations.

- As your own source makes clear, extraordinary special education aid does not cover anywhere near the full cost of a special education placement out of district (or in district). There most certainly is a cost to districts for students with costly special needs, and there is a very good argument to be made that the cost burden increases if these students are concentrated in district schools because they are not being served by the charter sector.

- You will note that above I asked whether the state will continue to suppress data under the same rules as before; maybe the rules have changed, I don't know.

But if they have, you are not adhering to the same protocols. Your lawyer may think that's fine; I find it unethical. You shouldn't get to decide for yourselves which rules you think you should and shouldn't follow; district schools don't run that way (or at least, they are not supposed to run that way).