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, June 27, 2014

Charter School Cream Skimming Doesn't Always Pay Off

There's a bit of a "he said/she said" thing going on in Jersey City between the school district and a charter school:
After telling Jersey City parents two weeks ago that the school district was investigating a possible data breach of personal student information, and then saying two days later that there was no data breach, school officials now allege that a local charter school did indeed improperly access the personal information. 
In a letter sent to parents on Friday, School Superintendent Marcia V. Lyles said a district investigation has concluded that M.E.T.S. Charter School officials have admitted they "accessed" the district's computer system to obtain personal information on district students. 
The district subsequently restricted the charter school's access to the system and is meeting with its software vendor to determine the extent of the data breach and how to protect student information, according to the letter. 
"The district will continue to work to ensure that our student information is secure," Lyles said in the letter, adding that the district has alerted the state Department of Education about the matter. 
Board of Education member Ellen Simon, speaking on behalf of the nine-member board, said the data breach was done "without district knowledge or consent." 
"The district acted diligently and in good faith," Simon said in an email. "I'm not sure the same can be said about M.E.T.S., at this point." [emphasis mine]
M.E.T.S., for their part, said they were allowed to get the student information:
A Jersey City charter school accused of improperly accessing the personal information of district students is denying the claim in a statement sent to The Jersey Journal tonight, with charter school officials saying the district allowed the school to access the data. 
"The M.E.T.S. Charter school was an authorized user of the district's computer system," reads the statement, the school's first regarding the alleged data breach. "There was no breach." 
On Friday, the Jersey City school district sent a letter to parents saying that M.E.T.S. Charter School, located on Sherman Avenue, improperly accessed personal student data. The district began investigating the matter after parents said M.E.T.S. mailed home registration information in the students' names. 
"The address information used was marketing information that is widely available to the general public through paid and free service companies," the M.E.T.S. statement reads. 
Parents told The Jersey Journal that at a June 11 meeting, Schools Superintendent Marcia V. Lyles said M.E.T.S. obtained information that included students' names, addresses, phone numbers and possibly social security numbers. 
In today's statement, Richard Borkowski, a member of M.E.T.S. Charter School's board, said the school had access only to students' names, addresses and telephone numbers, not social security numbers. [emphasis mine]
"Marketing information," huh? Well I guess we know what's being marketed; according to the Jersey Journal: "The district began investigating the matter after parents said M.E.T.S. mailed home registration information in the students' names." Whether or not the charter obtained the information with consent, it's clear they wanted the data so they could sell themselves to the families of the district.

Perhaps the largest controversy surrounding the charter school sector is whether or not charters "skim the cream": whether they deliberately screen out students who suffer from economic disadvantage (as measured by free-lunch (FL) eligibility), or Limited English proficient (LEP), or have special education needs (SpecEd).

The usual response is: "No, we don't, because we have lotteries when there are more applicants than slots open." To me, that's way too facile, for two reasons:

- First, it's easy to ward off families with special needs students: just don't provide the services their children need. Don't offer an extensive bilingual program, and you won't get many LEP applicants. Don't offer an ABA program with opportunities for mainstreaming, and you won't get many autistic applicants. "Suggest" that parents should be prepared to donate a specific amount of time to assist in school, and you won't get many FL applicants. It's really that simple.

- Second, the "relinquishers" often underplay the amount of social capital necessary to navigate a "choice" system of education. Immigrant families and the working poor are far less likely to be as sophisticated in making "choices" than those who are in a higher socio-economic status group.

Whether you buy into this or not, it's clear that something is causing the charters in places like Jersey City to wind up with substantially different student populations than the local public schools.

I'm going to piggy-back off of Bruce Baker's* work on Jersey City charters from the past fall, and some of his earlier work on charters throughout the state. All of the data following comes from the NJDOE. Let's start with some descriptive stuff:

M.E.T.S. enrolls grades 6 to 12, so I'm limiting this sample to schools that enroll 8th Graders; I have some graphs that show high school comparisons below. Four of the five "middle" schools in Jersey City that serve the lowest proportions of free lunch-eligible students are charter schools. Can this possibly be an accident? Especially when considering the enrollment of LEP students?

The Jersey City charters that serve grade 6 and higher have only two LEP students on their rolls (BelovED Community CS has 52 LEP students, but its oldest students are in Grade 3). Here's the SpecEd data:

100% of Regional Day's students are SpecEd so it's obviously an anomaly. M.E.T.S. Charter School has a special education rate lower than even The Academy I, Jersey City's designated "gifted and talented" school.

Again: I'm not saying M.E.T.S is engaging in cream-skimming on purpose; I'm saying that a "choice" system is inevitably going to create segregated schools like M.E.T.S., deliberately or not.

Now, all this begs a question: what advantage does M.E.T.S. gain from their segregated population? Are they posting up better results because they serve fewer FL, LEP, and SpecEd students? Let's look at their grade 8 scores in English Language Arts (ELA):

I am always loathe to point out the relative deficiencies on test scores for any school, charter or public: there's a lot more to a "good" school than test scores. But if we're going to use data to tar-and-feather public schools, let's use it fairly on charters. The truth is M.E.T.S., despite its low FL, LEP, and SpecEd percentages, isn't all that impressive.

"But wait!" say the charteristas. "Look on the right side of the graph! Those charters are spectacular! All we have to do is make more of them!" Well, sure, those average scale scores are great... but we're not accounting for variations in student populations, are we? I mean, it's much easier to get good scores when you have fewer kids, proportionately, who are FL, LEP, and SpecEd, right?

So let's account for that. We'll run a basic regression model that accounts for these three factors -- free lunch eligibility percentage, Limited English Proficiency percentage, and special education percentage -- and come up with a prediction for each school based on that model. We'll then see whether each school comes in above or below prediction, and by how much (geeks, the regression output is below).

The first thing to notice is that, as Bruce Baker pointed out in his post, there are two different types of charters in JC. Look along the x-axis: it's the same data as the bar graph above. Lena Edwards and Jersey City Community charters are much more similar to the JCPS schools than the rest of the charters in the city in the number of kids they enroll who qualify for FL .

But, again, what advantage did that give those segregated charters? Well, M.E.T.S. is way below where we'd expect them to be, given their student population; in this model (which is very limited and only gives us a clue about any school's "efficiency"), M.E.T.S. is one of the lowest-performing schools in the city.

But it's not like the rest of Jersey City's charter sector is doing much better. There are plenty of JCPS schools that do just as well, if not better, within this model -- and they are serving far more children in economic disadvantage.

Let's look at a few more tests. Here's Grade 8 math:

If the Jersey City charters were all that, we'd be seeing more of them above the prediction line. As it is, they perform from about where we'd expect to substantially below.

Grade 7 math and ELA:

Grade 6:

Now, an argument you'll hear is that what really counts is "growth": how much a school's students gained or lost relative to their peers on test scores. The state has one growth measure for schools: median Student Growth Percentiles (mSGP). Contrary to the state's position, we know for a fact that SGPs correlate to measures of student poverty and special education rates. So let's try to control for that and see how Jersey City's charters do:

I'll give props to Jersey City Golden Door on this last graph: they do substantially beat prediction on ELA SGPs. Taken together, however, the Jersey City charter sector is quite underwhelming.

Let's review:

  • While some of Jersey City's charters come close to matching the student populations of JCPS schools, the sector, as a whole, is not educating its fair share of children in economic disadvantage, who have limited proficiency in English, or who have special education needs.
  • When accounting for their differing student populations, Jersey City's charters are not getting particularly impressive test-based results, either in absolute scores or in measures of growth.
Yet here is M.E.T.S., advertising itself to the families of Jersey City in mailers, attempting to lure them away from the public schools which, as a whole, take more children who have special needs and who, in many cases, do a better job than the charters.

But if M.E.T.S. can't get better results even after skimming the cream (unintentionally or otherwise), what's the point? Why allow charters to recruit when they aren't doing any better than the local public schools?

And the point is...?

ADDING: Here are some demographics for the high school population of Jersey City. If a school enrolls Grade 12 students, it's on these charts.

Give University Academy Charter their props: their student populations look much more like the rest of the city than M.E.T.S.'s does.

Here's the regression output from Stata for that Grade 8 ELA model:

IndepVar  |  Coef.  |  Std. Err.  |  P > |t| (*p<.05, **p<.01)

PctFL | -23.83651 |  9.913562  |  *
PctLEP |  -6.111324  | 19.34249  |  
SpecEdPct2013 |  -99.77056 |  32.04121  |  **
Constant  |   235.3792 |  6.159327

R-sq = 0.5615
N = 24

 * Bruce is my advisor at Rutgers GSE in the PhD program.

1 comment:

Green Gables said...

Thank you a lot for sharing this with all of us you actually realize what you’re talking approximately! Bookmarked. Please also discuss with my site
Green Gables is one of the Secondary Schools in Hyderabad and Schools in Hitech City