These are remarks as prepared for my testimony today on One Newark. More to come. Cross-posted with Jersey Jazzman.
Testimony before the Joint Committee on the Public Schools
New Jersey Legislature
Tuesday, March 10, 2015
Good morning. My name is Mark Weber; I am a New Jersey public school teacher, a public school parent, a member of the New Jersey Education Association, and a doctoral student in Education Theory, Organization, and Policy at Rutgers University’s Graduate School of Education.
Last year, I was honored to testify before this committee regarding research I and others had conducted on One Newark, the school reorganization plan for the Newark Public Schools. Dr. Bruce Baker, my advisor at Rutgers and one of the nation’s foremost experts on school finance and policy, joined me in writing three briefs in 2014 questioning the premises of One Newark. Dr. Joseph Oluwole, a professor of education law at Montclair State University, provided a legal analysis of the plan in our second brief.
I would like to state for the record that neither myself, Dr. Baker, nor Dr. Oluwole received any compensation for our efforts, and our conclusions are solely our own and do not reflect the views of our employers or any other organization.
Our research a year ago led us to conclude that there was little reason to believe One Newark would lead to better educational outcomes for students. There was little empirical evidence to support the contention that closing or reconstituting schools under One Newark’s “Renew School” plan would improve student performance. There was little reason to believe converting district schools into charter schools would help students enrolled in the Newark Public Schools (NPS). And we were concerned that the plan would have a racially disparate impact on both staff and students.
In the year since my testimony, we have seen a great public outcry against One Newark. We’ve also heard repeated claims made by State Superintendent Cami Anderson and her staff that Newark’s schools have improved under her leadership, and that One Newark will improve that city’s system of schools.
To be clear: it is far too early to make any claims, pro or con, about the effect of One Newark on academic outcomes; the plan was only implemented this past fall. Nevertheless, after an additional year of research and analysis, it remains my conclusion that there is no evidence One Newark will improve student outcomes.
Further, after having studied the effects of “renewal” on the eight schools selected by State Superintendent Anderson for interventions in 2012, it is my conclusion that the evidence suggests the reforms she and her staff have implemented have not only failed to improve student achievement in Newark; they have had a racially disparate impact on the NPS certificated teaching and support staff.
Before I begin, I’d like to make a point that will be reiterated throughout my testimony: my analysis and the analyses of others actually raise more questions than they answer. But it shouldn’t fall to independent researchers such as me or the scholars I work with to provide this committee or other stakeholders with actionable information about Newark’s schools.
Certainly, we as scholars stand ready to provide assistance and technical advice; but the organization that should be testing the claims of NPS and State Superintendent Anderson is the New Jersey Department Of Education. The students and families of Newark deserve nothing less than a robust set of checks and balances to ensure that their schools are being properly managed.
One Newark can be thought of as containing four components: the expansion of charter schools; a “renewal” program for schools deemed to be underperforming; a system of consumer “choice,” where families select schools from a menu of public and charter options; and continuing state control of the district.
This last component is clearly a necessary precondition for the first three. Given the community outcry against State Superintendent Anderson and One Newark, it’s safe to say that none of the other three components would have been implemented were it not for continuing state control.
The critical questions I ask about these components are simple: do they work, are there unintended consequences from their implementation, and is One Newark being properly monitored and evaluated? Let me start by addressing the expansion of charter schools in Newark.
This past fall, I authored a report on New Jersey charter school demographics with Dr. Julia Sass Rubin of the Bloustein School of Planning and Public Policy at Rutgers University. This report was commissioned by the Daniel Tanner Foundation. Using publicly available data, we found that Newark’s charter schools, like charter schools throughout the state, serve a different population of students on average than their host districts.
This slide, from my first report with Dr. Rubin, shows that Newark’s charter sector serves fewer students eligible for free lunch, a proxy measure for economic disadvantage. Charters serve very few Limited English Proficient (LEP) students, fewer boys, and a substantially different racial profile of students than NPS’s schools.
This disparity in student populations has been acknowledged by State Superintendent Anderson herself, who said last fall: “I’m not saying they [the charter schools] are out there intentionally skimming, but all of these things are leading to a higher concentration of the neediest kids in fewer [district] schools.” The data does, indeed, back up the State Superintendent’s claim.
Another important difference between charter and district schools is the proportion of special education students they serve. Overall, no charter school serves as large a proportion of special needs students as NPS. Yes, there is variation between the district schools, but this is to be expected: some NPS schools specialize in serving students with particular learning disabilities.
In 2011, the NJDOE commissioned a report that outlined the costs of serving students with a variety of learning disabilities. The costs of Specific Learning Disabilities (SLDs) and Speech/Language Impairments (SPL) were found to be low compared to other impairments. As this graph shows, Newark’s charter schools serve proportionately more students with low-cost disabilities compared to NPS.
This is both a cost and logistical burden on NPS that the charter schools do not share. As we shall see, this difference likely has a profound effect on school finances in Newark.
The following graphs come from analyses that will be presented in an upcoming report on New Jersey charter school finances, authored by myself and Dr. Sass Rubin, to be released later this year. According to NJDOE data, Newark charter schools do, on average, spend less per pupil than NPS schools. However, charters spend less on student support services, and far more on administration. We must ask, at a time when New Jersey is under great budgetary stress, whether it is prudent to replicate independently managed schools within the same city, particularly when their administrative costs are so high.
Data from NJDOE’s “Taxpayers Guide to Education Spending” (TGES) shows that the budgetary costs per pupil – the Department’s preferred metric for comparing education spending across districts – are greater at NPS than the Newark charter sector as a whole. This trend is seen in cities across the state.
However, we must once again remember that district schools serve a different population of students than charter schools. These special education figures comes from the TGES; again, in Newark there is a substantial gap between the percentage of special needs students in the charters and in NPS schools. Undoubtedly, this affects per pupil spending costs.
This slide shows the differences in spending on student support services between charter schools and district schools. Support services include attendance, social work, health, guidance, educational media/school library, child study team, and so on. These are precisely the sorts of services we would expect to be provided more extensively by schools that serve larger populations of at-risk, special needs, and LEP students.
As in every other city, NPS far outspends its city’s charter sector on these services. NPS spends $3,963 more per pupil on support services than the charter sector as a whole. Clearly, the responsibility NPS has to educate more students with more costly disabilities relative to the charter schools is affecting school finances in Newark.
This begs a question: where else do Newark’s charter schools spend their money?
This slide gives us a clue. Newark’s charter sector spends, on average, $1,795 more per pupil on administrative costs. Again, it is possible that Newark’s charters simply can’t leverage the economies of scale NPS schools can. But there is another possible explanation:
Newark’s charters spend, on average, $1,098 more per pupil on administrative salaries than NPS. Again, this is typical of the trend across the state.
This raises the question of efficiency: are charters actually more efficient than district schools? In other words: given differences in student characteristics and available resources, which schools achieve the best test-based outcomes? Which schools really “do more with less”?
Dr. Bruce Baker has created a model, using a standard statistical technique called a linear regression, that allows for the comparison of efficiencies between Newark charter schools and NPS schools. Dr. Baker explains this model in a series of briefs; you can find links in my written testimony.
Basically, this method of comparison uses several inputs – special education percentages, free-lunch eligibility, staffing costs per pupil, and school size – to “hold all things constant.” In other words, Dr. Baker’s model attempts balance the scales for schools that serve more special need students, or more at-risk students, or spend less on staff, so that these schools aren’t disadvantaged in a comparison of test-based outputs. This, then, is a statistical model that makes comparisons fair.
With Dr. Baker’s permission, I have annotated his work here. The red bars represent Newark charter schools; the blue bars are NPS schools. I’ve further modified the graph so the “Renew Schools,” which I will discuss shortly, are in light blue.
The schools with bars that point upward are schools that are “more efficient”: given their student populations, their spending on staff, and their size, they produce better growth on student test scores than we would predict.
The schools with bars that point downward are schools that are “less efficient”: given their student populations, their spending on staff, and their size, they produce lower growth on student test scores than we would predict.
How does the charter sector fare overall? Certainly, some charters do well. But Robert Treat Academy, often touted in news reports as one of the highest performing charter schools in the city, is a relatively poor performer in this efficiency model. TEAM Academy, affiliated with the national charter management organization KIPP and often cited as another high-performing school, is quite average in this comparison.
Let me be clear: neither Dr. Baker nor I would ever claim that this analysis should be used as the final word on which schools perform well and which do not. As we shall see next, there are many other factors, not included in this model, which can affect test score growth.
What is evident here, however, is that the simplistic claim that charter schools “do more with less” is a gross mischaracterization of a highly complex interaction between student characteristics, resources, and test-based results. Simple claims that allowing charter schools to expand will lead to more great schools in Newark are just not warranted.
I mentioned that Dr. Baker’s model does not account for many factors that may explain the relative successes of certain Newark charter schools. One of those factors is attrition. There has been quite a bit written about this issue, so I’d like to make sure we get our terms straight before we look at this factor.
“Cohort attrition” is the year-over-year loss of students at a school within the same grade level. If, for example, a school enrolls 100 fifth grade students in 2014, and then 90 sixth grade students in 2015, that is a cohort attrition rate of 10 percent.
The role of student attrition in explaining charter school outcomes has been a source of national debate. Admittedly, it is impossible to get precise estimations of the effects of cohort attrition without student level data.
Even though we have incomplete data, however, we can look at it to discern whether there is enough evidence to warrant an investigation. Here is the cohort attrition for the Class of 2014 at NPS schools, TEAM Academy Charter School, and North Star Academy Charter School. The graph shows how the number of students declines each year for the grade level that was on track to graduate in the spring of 2014, starting in Grade 5. Each year shows the number of enrolled students in that class as a percentage of students enrolled in Grade 5 back in 2006-07.
NPS’s Class of 2014 was 76 percent of the size it was back when it was enrolled in Grade 5 in 2006-07. In contrast, TEAM’s Class of 2014 was only 59 percent of its size by it senior year; North Star’s was only 56 percent of its Grade 5 size.
To be fair, this class was somewhat unusual for TEAM. While its cohort attrition for the Class of 2013 was still greater than NPS’s, the gap between the two systems was not as great as with the Class of 2014. North Star, however, only retained 43 percent of its original class size.
The question we should ask is whether this attrition affects test score outcomes: are low performers leaving charter schools, helping to boost their average test scores? Again, we need student-level data to answer this question; however, we can look at publicly available data for some interesting clues.
This graph superimposes cohort attrition for the Class of 2018 – the last class for which we have NJASK Grade 8 data – with average scale scores for each year’s NJASK English Language Arts (ELA) test. Note that North Star starts with higher scale scores for this class in Grade 5 than NPS; however, the gap increases as the size of North Star’s cohort shrinks. The NPS Class of 2018 cohort, in contrast, barely changes during this time.
The critical question then is this: were students who left North Star a drag on the school’s average test scores? Does the school retain high performers while lower performers leave? Again, there is no way to know without individual student data; however, there is, in my opinion, more than enough evidence for the NJDOE to begin a serious investigation into the role of cohort attrition on test score outcomes.
I’d like to turn my attention now to the next component of the One Newark plan: “Renew” schools. In March of 2012, State Superintendent Anderson announced a plan that would, among other reforms, require all staff members at eight NPS schools to reapply for their jobs.
The wholesale turnover of a teaching staff is known as “reconstitution.” While not all teachers were necessarily replaced in the plan, our analysis indicates there was a significant change in the staff of the eight Renew schools.
My review of the research shows that there is no evidence that reconstitution is a consistently successful strategy for improving schools. In fact, reconstitution can often be risky, leading to students enrolling in schools that underperform compared to where they were previously enrolled.
In December of 2014, the Alliance for Newark Public Schools published a report by Dr. Leonard Pugliese, regional vice-president of the American Federation of School Administrators and a faculty member at Montclair State University. This table is from that report.
Dr. Pugliese found that, in most cases, the passing rates on NJASK tests for the Renew schools actually decreased over the two years of their renewal.
Extending this work, I examined the Renew schools through several other lenses. Revisiting Dr. Baker’s efficiency model, I changed the parameters to only look at school outputs over the last two school years. Obviously, it is too early to make definitive conclusions about the efficacy of the Renew strategy; however, this early look shows that there is no consistent pattern of Renew schools demonstrating any more efficiency, as a group, than the rest of Newark’s schools.
Further: in their first year of “renewal.” the eight schools showed, on average, a sharp drop in their median Student Growth Percentile (mSGP) scores compared to the rest of NPS’s schools. This slide shows the drop in ELA mSGP scores for the Renew schools. It is worth noting that SGPs compare students – and, consequently, schools – to other students with similar test score histories. In other words, the bounce back up in mSGP in year two of renewal does not likely indicate a return to where the schools were before renewal, because the schools are now being compared to lower-performing schools. It is, likely, easier for the Renew schools to show growth, because their growth was low in year one of renewal.
Here we see the same pattern for Math mSGP scores. Again, the bounce in year 2 likely indicates that it is easier for Renew schools to now show growth as their new comparison schools are lower performing.
Test-based outcomes were not the only changes at the Renew schools. My analysis of NJDOE staffing data shows several remarkable trends after renewal. The average experience of the staff at these schools, following reconstitution, declined significantly, and the percentage of staff with less than three years of experience increased.
The average of teacher experience before renewal was 14.8 years; after renewal, average experience dropped to 11.1 years, and stayed roughly the same in the next year.
Before renewal, 11% of teachers had less than three years of experience; after renewal, 26% of teachers were similarly inexperienced. The research consensus is clear: teachers gain most in effectiveness during their first few years of teaching. The large increase in novice teachers likely made the overall teaching corps for the Renew schools less effective.
Experience, however, was not the only change in staff characteristics following renewal. In the program’s first year, the percentage of black teachers at these schools dropped substantially. Keep in mind that most of the Renew schools serve a majority black student population.
A recent article in the peer-reviewed journal Urban Education reviews the literature and concludes that black students benefit from having teachers of their own race. While there is no indication that the change in the racial composition of the Renew schools’ staffs was deliberate, there may still have been unintended consequences.
This graph shows the student populations for the eight Renew schools; all but one have majority black student populations. In those seven, at least 75% of the student body is black.
And yet, in the first year of renewal, the proportion of black teachers declined by seven percentage points.
What did “renewal” ultimately mean for these schools? Intentionally or not, it meant fewer black teachers with experience – this in schools with large proportions of black students. It also meant a significant drop in growth scores, and a decline in proficiency on state tests.
Again: it is too early to come to a definitive conclusion about the efficacy of the Renew schools program. All early indications, however, are not promising.
ONE NEWARK and “CHOICE”
I turn now to the third component of One Newark: school “choice.” The One Newark plan called for students and families to choose their schools from a menu of charter and district schools, using a single application. I won’t recount the many problems with this application system – nor the subsequent staffing, transportation, and logistical problems – as those have been well reported in the press.
I will, however, refer to a classic economics paper from George Akerlof titled “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.” Akerlof used the used car market to explain that a consumer model only works when there is adequate and impartial information available to consumers. Without this information, consumers are not only likely to fall victim to unscrupulous providers; providers of quality goods are less likely to enter the market.
A market system of choice for schools, then, requires that families have high-quality information about the schools they are choosing. NPS attempted to provide that information on the One Newark application, labeling schools at three different tiers. “Falling Behind” schools are those that allegedly lag in student outcomes. “On the Move” schools are supposedly improving in their performance; “Great” schools supposedly serve their student well.
It is reasonable to think that Newark’s families leaned heavily on this application when making their school choices. But what was NPS actually measuring? Were they taking into account the differences in student populations when judging test score outcomes?
To judge this, I conducted an analysis using a linear regression model, and published the results in a brief this past spring. By using a statistics tool to “hold all things equal,” I’m able to show which schools performed above or below where we would predict them to be, given their student populations.
Logically, we would expect the “Falling Behind” schools to perform below prediction, and the “Great” schools to perform above. In fact, however, the ratings are all over the map: there are “great” schools that under-perform, and “Falling Behind” schools that over-perform.
This question, then, is how NPS was judging whether schools were “Great,” “On the Move,” or “Falling Behind”?
This graph gives us a clue. “Great” schools have fewer free lunch eligible students, fewer boys, fewer black students, and fewer students with special needs. The One Newark application, arguably, wasn’t evaluating the effectiveness of a school; it was, instead, judging the characteristics of its student population.
I would argue this is not the sort of information that a family needs when making a school choice. No school should be penalized simply because it serves a different student population.
Today, I will admit that I have raised more questions than I could answer. The truth is that the body that should be bringing you the information you need is the New Jersey Department of Education. The appropriate role of the Department is to provide the data and analysis that you, the policy makers of this state, need to inform your decisions.
To that end, the Department must be an impartial overseer of Newark’s, and every district’s, schools. There needs to be a system of checks and balances put in place to ensure that NPS and State Superintendent Anderson are pursuing programs that have a good chance of succeeding.
I know I speak for many education researchers in New Jersey in stating that we are ready and willing to assist all of the policy makers responsible for Newark’s schools in formulating programs that can be successful.
One resource for you, your staffs, the NJDOE, and NPS to use is the New Jersey Education Policy Forum, a collaborative effort of education policy scholars throughout the state. Dr. Baker established this resource because he believes, as do I, that scholars and researchers should make our work available to policy makers and other stakeholders.
We will continue to monitor Newark’s progress as best we can, and we stand ready to assist you and all other policy makers in serving the children of Newark.
Thank you for your time.
Please refer to Jersey Jazzman for all footnote links.