The Harlem Children’s Zone (HCZ) is a non-profit organization that funds and operates a neighborhood-based system of education and social services for the children of low-income families in a 100 block area in Harlem, New York.
The HCZ education components include early childhood programs with parenting classes; public charter schools; academic advisors and afterschool programs for students attending regular public schools; and a support system for former HCZ students who have enrolled in college. Health components include a fitness program; asthma management; and a nutrition program. Neighborhood services include organizing tenant associations, one-on-one counseling to families; foster care prevention programs; community centers; and an employment and technology center that teaches job-related skills to teens and adults.
Presidential candidate Barack Obama campaigned on replicating the HCZ as the first part of his plan to combat urban poverty:
The philosophy behind the project is simple — if poverty is a disease that infects an entire community in the form of unemployment and violence; failing schools and broken homes, then we can’t just treat those symptoms in isolation. We have to heal that entire community. And we have to focus on what actually works . . . . And it is working . . . . And if we know it works, there’s no reason this program should stop at the end of those blocks in Harlem.[iii]
True to his campaign promise, President Obama instituted a Promise Neighborhoods Initiative intended to replicate the HCZ in 20 cities across the country. The program received a $10 million appropriation from Congress in 2010, under which 339 communities applied to the U.S. Department of Education for planning grants to create Promise Neighborhoods. The administration has requested $210 million in new funding for the 2011 budget year to move from planning to implementation.[iv]
The influence of the HCZ is not limited to these shores. It is a regular stop for international visitors interested in education reform. The Hungarian government intends to replicate the program to address social and education problems with their largest ethnic minority, the Roma.[v]
What is unique and attention-getting about the HCZ is that it is designed on the assumption that it takes both effective, achievement-oriented schools and strong social and community services to support the educational achievement of children in poverty. The presumption is that effective schools alone are insufficient. In this the HCZ and Promise Neighborhoods are aligned with the Broader, Bolder Approach to Education, an advocacy position taken by an influential group of proponents of the view that public investment in the communities and society in which children are reared is a necessary condition for education reform.
The nation’s education policy has typically been crafted around the expectation that schools alone can offset the full impact of low socioeconomic status on learning . . . . [T]here is solid evidence that policies aimed directly at education-related social and economic disadvantages can improve school performance and student achievement. The persistent failure of policymakers to act on that evidence — in tandem with a schools-only approach — is a major reason why the association between disadvantage and low student achievement remains so strong.[vi]
Does the HCZ Work?
The entire rationale and appeal of the HCZ is its holistic, neighborhood-based approach to the educational achievement of low-income students. With the administration proposing hundreds of millions of dollars of new federal funding for Promise Neighborhoods, with the shape of the reauthorization of the Elementary and Secondary Education Act under influence from the Broader, Bolder philosophy, and with the academic future of a generation of poor children on the line, we should ask whether the HCZ works, and whether it works as advertised.
Whether the HCZ works and whether it works as advertised are different questions. Imagine that students who receive the full panoply of HCZ services have superior achievement to similar students who don’t receive those services. We would conclude that the HCZ works. But what if students who received the schools-only component of the HCZ did as well as students who received the full treatment? Then we would have to conclude that the HCZ works, but not as advertised. Under the latter scenario the HCZ would be an exemplar of the very schools-only approach that the Broader, Bolder proponents reject as ineffective.
HCZ works, at least to raise academic achievement among the population of students whose families try to enroll them in HCZ charter schools. Harvard researchers Dobbie and Fryer conducted a study of the HCZ that took advantage of a New York City regulation that requires public charter schools to select students by lottery when the demand for slots exceeds supply.[vii] By comparing academic outcomes for lottery winners vs. lottery losers, they were able to create the conditions of a randomized experiment, thus assuring that any differences among the two groups in academic outcomes were due solely to the opportunity for enrollment in the HCZ charter schools. The researchers found very large effects on academic achievement, particularly for math at the end of middle school. They conclude that, “the effects in middle school are enough to reverse the black-white achievement gap in mathematics.”
Does the HCZ work as advertised? Dobbie and Fryer approached this question with research strategies that are far less definitive than the lottery-based randomized trial they used to assess the effect of charter school attendance. Nevertheless their results are informative and not encouraging for the Broader, Bolder philosophy. In the most compelling analysis, they compare outcomes for students attending the HCZ charter schools living inside vs. outside the geographical boundaries of the HCZ. Those inside the boundaries are eligible for the complete HCZ package of social and community supports whereas those outside the boundaries receive only the charter school component. They find that students outside the Zone garner the same benefit from the HCZ charter schools as the students inside the Zone. In other words, proximity to the community programs had no effect. In light of this finding and other results that show no relationship between community services and academic outcomes, they provide a conclusion that appears overly generous to the possible role of the community components of the HCZ: “We conclude . . . that high-quality schools or high-quality schools coupled with community investments generate the achievement gains. Community investments alone cannot explain the results.” The more parsimonious conclusion might have been that the results can be explained without recourse to community investments.
A New Analysis
Dobbie and Fryer’s analysis is restricted to students who participated in a lottery to attend an HCZ charter school. They did not examine the effectiveness of HCZ charter schools relative to other charter schools in NYC. However the effectiveness of the HCZ charters relative to other charter schools is directly relevant to the Broader, Bolder hypothesis and the Obama administration’s Promise Neighborhoods initiative. Reforming neighborhoods and making schools the center of social service networks is challenging and expensive. The HCZ has benefitted from over $100 million in philanthropy to support its holistic approach. If a schools-only approach works as well or better than a schools plus community approach, this has huge consequences for education policy. It goes to the heart of how public funds should be allocated to enhance educational achievement and reduce socioeconomic disparities.
This question, whether charter schools with a school-centric approach do as well or better than the HCZ schools with their associated neighborhood and social services, can be answered in a satisfactory manner given certain patterns of data even though specific causal estimates are not possible.
Causal estimates of the effect of HCZ charter schools relative to other charter schools would require the random assignment of students to charter schools, or a proxy for randomization. No such randomization has occurred or is likely to occur in New York City. Under the district’s enrollment policies parents select the charter schools they wish their children to attend and the schools actively recruit. Thus the enrollment in any particular school is a joint function of parental interest in that school and the school’s recruiting efforts. This means that the performance of the students in a particular school is determined both by the educational experiences provided by the school and the characteristics of the students enrolled in that school. Characteristics of students can be assumed to vary across schools.
These two sources of influence on student outcomes, schools and student backgrounds, cannot be separated definitively absent randomization. Thus if students in HCZ charter schools performed at higher levels than students in other charter schools one could not know for sure whether that was because the students and their families attending the HCZ charters were different from those attending other charter schools, or because the HCZ charter school plus community package was superior to the schools-centric approach of other charters, or both. However, this pattern of data, superior student outcomes in HCZ charters compared to other charters, would be consistent with the Broader, Bolder philosophy of investment in community services.
Another pattern of data that would be consistent with the HCZ operating as advertised would be better than expected performance of HCZ schools given the background characteristics of the students served. In this scenario, students attending HCZ charters would not have to perform better than students attending other charter schools; they would only have to perform better than students of similar racial and socioeconomic backgrounds attending other charter schools. Again, this would not prove that it takes a neighborhood to make an effective school, but it would be consistent with that point of view.
Other patterns of data would be inconsistent with the theory of action underlying the HCZ, Promise Neighborhoods, and the Broader, Bolder philosophy. Presidential candidate Obama asserted that we can’t treat failing schools in isolation — we have to heal the entire community – and he used the success of the HCZ to support his view. That position would not square with evidence that the HCZ schools, which are unique in their schools plus community services model, are unexceptional in terms of performance.
Our question is simply this: Does the HCZ produce exceptional academic achievement? If it does, that is promising for Promise Neighborhoods. If it doesn’t, it isn’t.
We utilize the most recent administrative data from the New York State Department of Education, which includes state test results for all schools in New York.[viii] The State database contains the English language arts and math test score data for the 2007 through 2009 school years as well as a variety of school demographic information. The State database does not contain a designation of whether a school is a charter school or traditional public school. We used the Common Core of Data from the National Center for Education Statistics to identify which schools in the State database were charter schools.
The State database contains different test score information for schools across grades and years. Because our interest was in comparing the effectiveness of the HCZ schools with other charter schools, we utilized the data for the grades and years that included test scores for the HCZ schools. These were:
2007: Grades 6, 7, and 8
2008: Grades 3, 7, and 8
2009: Grades 3, 4, 5, and 8
Two HCZ schools are included in the State database. The Harlem Children’s Zone Promise Academy is the longest established HCZ public charter school and serves students from elementary school through high school. It has data in the State database for each of the years and grades listed above. The Harlem Children’s Zone Academy II is an elementary school that has data in the State database only for grade 3 in 2008 and grades 3 and 4 in 2009. To avoid confusing the results from the newer Academy II with those from the Promise Academy and to circumvent the HCZ from competing against itself in the school rankings that are the heart of our method, we included only data from the HCZ Promise Academy in our analysis.
Our initial analysis compared the average test scores for the HCZ Promise Academy with the average test scores of all other charter schools in Manhattan and the Bronx for the grades and subjects in the State database. Because different schools are in the State database for each year and subject tested our comparisons were performed separately for each grade, year, and subject (i.e., a calculation using mathematics in Grade 6 in 2007; a separate calculation using English language arts in Grade 6 in 2007, etc). This resulted in 20 separate comparisons: the 10 grades and years listed above for each of the 2 subjects. For each comparison we converted the mean test score for each charter school into a percentile rank. For example, if a particular charter school had a mean score on the mathematics assessment at grade 8 in 2007 in the exact middle of the distribution of scores such that half the charter schools had higher scores and half the charter schools had lower scores, it would have received a percentile score of 50. We then averaged the HCZ Promise Academy’s percentile ranks across the 10 separate grade and year distributions for mathematics. We did the same for English language arts. The number of charter schools with which the HCZ Promise Academy could be compared varied from year to year and grade to grade, but averaged 14.
We conducted a parallel set of analyses with statistical adjustments for the demographic background of students in the schools. Using data from all regular public and charter schools in New York City, a prediction equation for school test scores was generated for each of the 20 grade, year, and subject possibilities. Percent free lunch, percent reduced lunch, percent limited English proficient, percent African American, and percent Hispanic were the predictors. Predicted scores for each of the Manhattan and Bronx charter schools in our HCZ comparison sample were derived from these equations using each school’s demographic information. Each school’s predicted score was subtracted from its actual test score for a particular subject, year, and grade to create a difference score. A positive difference score meant that the school’s actual performance on the test was higher than the score predicted from the demographics of the school’s student population for that subject, year, and grade. Conversely, a negative difference score indicated that the school’s students did worse than predicted from their demographics. Finally, these difference scores were transformed into percentile ranks and averaged using the same procedure that was applied to the actual scores as described previously.
The following table summarizes the results. The column labeled “Actual” presents the HCZ Promise Academy’s average percentile rank on mathematics and English language arts relative to charter schools in Manhattan and the Bronx using actual scores. The column labeled “Adj.” presents corresponding data for the difference scores, i.e., performance adjusted for the demographic characteristics of the students in the schools.
Percentile Scores of the HCZ Promise Academy Charter School
Mathematics, relative to charter schools in Manhattan and the Bronx
English language arts, relative to charter schools in Manhattan and the Bronx
Considering mathematics and English language arts jointly (the grand mean in the table) half or more of the public charter schools in Manhattan and the Bronx produce test scores on state assessments that are superior to those produced by the HCZ Promise Academy. This is true both for actual scores as well as scores adjusted for student demographics. The same general pattern holds for math and English language arts considered separately, but it appears that mathematics is HCZ’s stronger suit. The inescapable conclusion is that the HCZ Promise Academy is a middling New York City charter school.
There are two credible studies demonstrating that charter schools in New York City are strong performers as a group, producing superior gains for students compared to traditional schools in that city.[ix] Thus the HCZ Promise Academy is up against strong competition. That it is in the middle of the pack is not an indictment of its effectiveness by any means. This is illustrated in the following figure, which plots the performance of charter schools in our sample on mathematics at grade 8 in 2009 as the difference between their actual scores and the scores predicted from their student demographics. The majority of the charter schools have positive scores, which is to say that they do better than the average across the city for a school with their demographic profile. The HCZ Promise Academy, represented by the white bar in the chart, scores 10 points above its predicted score on the state assessment, which is about .6 of the school-level standard deviation for all NYC schools for grade 8 math in 2009. Thus students attending the HCZ Promise Academy are doing impressively better than students of their backgrounds attending a typical public school in NYC. However, the charter school at the top of the list, which happens to be a KIPP school, scores 30 points above its predicted score. There are 3 KIPP schools represented in the graph. All score higher than the HCZ Promise Academy. None provide or depend on community and social services to achieve their academic mission.
These findings create a large question mark for the theory of action of the HCZ. If other charter schools generate outcomes that are superior to those of the HCZ and those charter schools are not embedded in broad neighborhood improvement programs, why should we think that a neighborhood approach is superior to a schools-only approach?
There is no compelling evidence that investments in parenting classes, health services, nutritional programs, and community improvement in general have appreciable effects on student achievement in schools in the U.S. Indeed there is considerable evidence in addition to the results from the present study that questions the return on such investments for academic achievement. For example, the Moving to Opportunity study, a large scale randomized trial that compared the school outcomes of students from poor families who did or did not receive a voucher to move to a better neighborhood, found no impact of better neighborhoods on student academic achievement.[x] The Nurse-Family Partnership, a highly regarded program in which experienced nurses visit low-income expectant mothers during their first pregnancy and the first two years of their children’s lives to teach parenting and life skills, does not have an impact on children’s reading and mathematics test scores.[xi] Head Start, the federal early childhood program, differs from other preschool programs in its inclusion of health, nutrition, and family supports. Children from families enrolled in Head Start do no better academically in early elementary school than similar children whose parents enroll them in preschool programs that do not include these broader services.[xii] Even Start, a federal program that combines early childhood education with educational services for parents on the theory that better educated parents produce better educated kids, generates no measureable impact on the academic achievement of children.[xiii]
This is not to suggest that factors such as parental education and income, family structure, parental employment, exposure to crime, and child health are not related to student achievement. Such statistical associations are at the empirical heart of the Broader, Bolder claims. However, evidence, for example, that single parenthood is negatively associated with children’s academic achievement is no evidence at all that investment in a community service that intends to keep parents together will succeed in doing so, much less have a cascading positive impact on the academic achievement of children in families that are served by the marital counseling program. Per our recitation of findings from studies of Moving to Opportunity, Head Start, et al., efforts to affect achievement in school through broad interventions outside of school have little evidence of success.
In contrast to disappointing results for Broader, Bolder initiatives, there is a large and growing body of evidence that schools themselves can have significant impacts on student achievement. The most powerful educational effects over which we have any societal control occur within the walls of schools. They are the effects produced by good teachers,[xiv] effective curriculum,[xv] and the changes in leadership, management, culture, and time to learn that are incorporated into schools that beat the odds, including successful charter schools.[xvi]
President Obama was a community organizer before he was a politician, so it is natural that his instincts are to invest in community programs. But President Obama has repeatedly called for doing what works. Doing what works depends on evidence not instincts. There is no evidence that the HCZ influences student achievement through neighborhood investments. There is considerable evidence that schools can have dramatic effects on the academic skills of disadvantaged children without their providing broader social services. Improving neighborhoods and communities is a desirable goal in its own right, but let’s not confuse it with education reform.
[iv] It appears likely based on initial action in the House that substantially less than the requested amount will be forthcoming. Given the expansive and expensive model of services embodied in the HCZ it isn’t clear that the $210 million the administration wanted would have been sufficient (each grantee was to receive a 5-year grant – $2 million per year on average for 20 grantees). Funding the program for far less will undermine any chance of replicating the HCZ unless the number of funded sites is scaled back in proportion.
[ix] http://credo.stanford.edu/reports/NYC%202009%20_CREDO.pdf ; http://www.nber.org/~schools/charterschoolseval/how_NYC_charter_schools_affect_achievement_sept2009.pdf