The Education Gadfly Weekly

Matthew Chingos
Harvard University, Program on Education Policy and Governance
May 2010 

This working paper by Harvard research fellow Matthew Chingos is a crystal ball for Ohio. Chingos examines Florida’s 2002 class-size reduction (CSR) policy – a universal mandate similar to Ohio’s mid-2009 policy enacted via the governor’s evidence-based model of school funding—and finds that it had no impact on student achievement.

The report outlines previous class size research and puts the infamous Tennessee STAR experiment (frequently cited by CSR defenders) in context. Other research has since contradicted STAR’s findings, and minority children benefited more than non-minorities in that study – a finding that should caution against universal CSRs and encourage more targeted reductions, if anything.

Using student-level data from the Florida Department of Education, Chingos examines deviations from prior trends in student achievement at districts and schools required to reduce class size, as well as deviations among those districts and schools that weren’t require to do so. Simply put, the comparison group included districts and schools which had already reduced class size and so didn’t have to comply with the 2002 mandate. The treatment group focused on those that did have to reduce classes by at least two students per year. (Chingos explains clearly why looking at achievement before and after the policy isn’t appropriate since Florida also had all sorts of other policy changes at that time that may have contributed to rising achievement.)

To add insult to injury to CSR defenders, Chingo also found no positive outcomes on other indicators like absenteeism, suspension, or crime and violence. In exposing Florida’s class size disappointment, the report points out the poor assumptions underlying Ohio’s policy: it presupposes “that resources provided to reduce class size will have a larger impact on student outcomes than resources that districts can spend as they see fit.” Rather than requiring all Ohio schools to reduce class sizes dramatically, the state would be better off targeting CSRs to the students it would help the most – especially considering the exorbitant costs the mandate imposes on ailing school budgets.

Read the study here.

Editorial Projects in Education
June 2010

Did you know that just 2 percent of Americans earned high school diplomas in 1870? That’s just one of the tidbits you’ll find in this year’s Diplomas Count (find 2009, 2008, 2007, and 2006 here). In addition to the usual graduation statistics update, this edition attempts to chr onicle “data in action,” i.e., how the smart use of information can raise graduation numbers, mainly by identifying students at risk for dropping out. As usual the news isn’t good: The graduation rate hovers around 70 percent, having actually declined slightly from 2005 to 2007 (the most recent year of available data). But that’s somewhat misleading, because while the overall graduation rate has fallen, rates for every racial group have improved. This is Simpson’s Paradox in action, as the lower overall number can be attributed to the fact that the population of students most at risk of dropping out—minorities, especially Latinos—is composing an increasing percentage of the overall student body. Perhaps most interesting is that just twenty-five of the nations’ 11,000 school districts account for a whopping 20 percent of all dropouts, or 250,000 students. New York City and Los Angeles are the worst offenders, each failing to graduate more than 40,000 students every year. Admittedly, these districts are the nation’s largest and would statistically have more dropouts than smaller ones, but it’s also a lesson in how much difference data systems could make by being used well in just a handful of places. The accompanying journalistic pieces are heartwarming: School leaders responding to data with tutors, coaches, and mentors; teachers tracking down missing students and helping them enroll in summer school; and district and state leaders improving the use of data by tracking across borders, even state lines, to make sure no child falls through the cracks. The bottom line is this: Dropouts remain a huge problem, but robust and consistent use of data is a promising solution. Read it here and check out the accompanying web materials here.

Center for Reinventing Public Education
Robin Lake, Brianna Dusseault, Melissa Bowen, Allison Demeritt, Paul Hill
June 2010

This publication contains interim findings from a nationwide study of Charter Management Organization (CMO) effectiveness. Started in May 2008 and expected to conclude summer 2011,  the work was commissioned by New Schools Venture Fund and is being conducted jointly by Mathematica Policy Research, Inc., and The University of Washington’s Center on Reinventing Public Education. The point of the project is to examine the effectiveness of CMOs (non-profit organizations that manage “more than one school with a unified management team responsible for delivering the educational program and supervising school leaders”) by investigating their operations, internal structures, growth strategies, missions, and challenges.

The findings come from 43 of the 82 known CMOs (as of 2007) and highlight educational programs, school cultures, teacher accountability systems, and the economics of starting and sustaining CMO operations. Most interesting among the takeaways is that while many of the CMOs firmly believe that their particular educational approach is critical to the success of their programs, their approaches differ widely. This confirms what, anecdotally, we’ve seen from an authorizer perspective: the school leadership and people at the front of the classroom matter most. It also gives credence to another anecdotal observation: all individual schools have different needs – even schools under the umbrella of the same management organization - and individual schools must have the ability to adjust quickly to meet internal (e.g., human resource) needs and the needs of students. Although this study focuses on charter schools, the latter observation likely applies to non-charter public schools as well.

The report also notes the challenges CMOs face with growth, particularly the fine line between developing internal procedures to efficiently manage operations and layering on rules and procedures, thereby becoming the very bureaucracy many of these CMOs sought to avoid. Interestingly, the larger the CMO, the more control the central office is likely to exert over the individual schools and their operations. 

The authors also set forth next steps in their research, which include four issues that definitely make this work worth revisiting in 2011 in the final report: 1) how differences between CMOs and districts relate to outcomes, 2) the costs and benefits of increased instructional time, 3) whether there is an optimal CMO size, and 4) an analysis of CMO cost structures and sustainability. Read it here.

Mathematica Policy Research, Inc.
Christina Clark Tuttle, Bing-ru Teh, Ira Nichols-Barrer, Brian P. Gill, & Philip Gleason
June 2010 

This is the interim report of an ongoing (until 2014) longitudinal study of achievement in a quarter of KIPP's eighty-two schools. Though KIPP schools have been the focus of previous research (see here and here, for example), this is by far the largest and most rigorous study to date. And the results are encouraging. Using matched student achievement data from twenty-two middle schools that had been open since at least 2005-06, Mathematica analysts found statistically significant impacts on reading in fifteen of the twenty-two, and on math in eighteen. Conversely, just two schools had a significantly negative impact on reading, while one school had a significantly negative impact on math (in year 1), which actually reversed into a positive impact by year three. These positive effects are sizable, especially in math. After three years in a KIPP school, a student will have made on average 4.2 years of growth in math and 3.9 years of growth in reading. This was true even though KIPP included in its treatment group all students who were ever enrolled in a KIPP school during the study, including those who spent just one year at KIPP and subsequently left, as well as the results for two schools that lost their KIPP affiliation during the study and subsequently closed. That means these results are probably conservative in terms of students who remained enrolled at KIPP all four years of middle school because they hold KIPP accountable for students who actually were not at KIPP for the majority of their middle school years. Though KIPP surely deserves praise for these results, it should also be applauded for subjecting itself to such a rigorous assessment. Read it here.

The truth is finally setting in about the health of Ohio’s public pension systems, but one shouldn’t be too optimistic that major reforms are on the horizon.

In 2007 the Fordham Institute commissioned the well-respected economists Robert Costrell and Michael Podgursky to analyze Ohio’s State Teachers Retirement System. Key findings from their report, Golden Peaks and Perilous Cliffs: Rethinking Ohio’s Teacher Pension System, included that:

• The system is too pricey to sustain in its current manifestation;
• The system encourages early retirement and “double-dipping”;
• The system is out of step with the state’s current teacher needs, labor markets, and career patterns; and
• The system lacks transparency.

Representatives of both the STRS and the teachers unions publicly questioned not only the veracity of the report’s findings but also the integrity of the authors themselves. Much of the media bought STRS’s line, including the Dayton Daily News, which said in an editorial about our report:

The Fordham Institute is wrong to imply that the retirement system is in crisis, or that Ohio taxpayers suddenly could be saddled with huge liabilities. Such alarmist talk is factually inaccurate and does not fairly characterize what the researchers found.

The retirement system has ample assets to pay retirement benefits. There is no risk that Ohio taxpayers could be stuck with huge, unforeseen pension liabilities.

Fast forward a few years.

By law, state pensions must be able to cover their liabilities within a 30 year period. But in its 2009 annual report, STRS reported that its unfunded liabilities were at “infinity,” meaning they can never be paid off unless changes are made to the system.

Last week, a joint analysis and examination of Ohio’s five public pension systems by the state’s major newspapers agreed. Ohio Senate President Bill Harris weighed in, saying that Ohio’s current public-employee pension systems are “not sustainable” and will require legislative action.

In 2007, Costrell and Podgursky used the power of economics and rational analysis to point out the serious problems confronting STRS. They have subsequently done deeper analysis in other states and have been at the forefront of badly needed pension reform efforts nationally.

Ohio’s political debate and its newspapers have caught up to the pessimistic scientists. The state’s public retirement systems really are troubled and in need of serious reforms.

Now that we agree on the problem, what should we do about it?

For STRS at least, our report called for transitioning away from the defined-benefit plan to a “cash-balance” one, where the guaranteed rate of return is closer to the rate of risk-free Treasury bonds than the current eight-percent return rate that STRS hopes to see.

It’s not likely, nor would it be fair, that the state will make radical changes to the pension expectations of current teachers and retirees. But those pension promises rely heavily on the contributions of younger teachers. Thus, switching plans for incoming teachers would require a hefty influx of cash to the system to maintain the benefits already promised to current teachers and retirees.

Given the state’s current financial plight, it’s tough to imagine the state coming up with anything close to what it would take to right the STRS ship. Instead, the fundamental flaws that exist in the system are likely to persist, and employers and employees will have to pony up to keep the system afloat. Currently, teachers put 10 percent of their salaries into the pension fund, and local school districts contribute another 14 percent. STRS is seeking 2.5 percent more from each, along with changes to the average final salary calculations, minimum retirement age, and other aspects of the system.

In case you’ve been living in a cave, the 2010 FIFA World Cup is well underway, and Ohio Gadfly is a proud fan. (Did you know US head coach Bob Bradley is an Ohio University alum and the school’s former head soccer coach?) As we head into the final games, futbol aficionados and proud nationalists of every creed will cram into pubs so crowded that maximum occupancy signs lose all meaning. There will be endless speculation and commentary ranging from the evolution of the official World Cup ball and making fun of our own American cluelessness about the sport to why Americans are so behind the times when it comes to world’s most infamous game.

What you might miss amid the noisy vuvuzelas and the excitement of 91^st minute qualifying goals is commentary on the countries these players represent. Specifically, what are their education systems like? If Cristiano Ronaldo, Kaka, or Landon Donovan weren’t living their dreams playing for national teams, what might their lives be like given the schooling they received?

We took a look at the final 16 countries and tried to be impartial, unless the stat is as obvious as a handball in the box – for example, higher enrollment in school or gender parity in education are unequivocally positive. We’ll try our best to be good commentators; the only thing missing is a Scottish accent.

Take a look at World Cup finalists’ enrollment levels in pre-primary, primary, and secondary school, and tertiary education (higher education). Ghana is at the bottom of the pool with low enrollment, especially for secondary school (49 percent of boys and 45 percent of girls). The European countries have fairly high participation, though England’s 57 percent enrollment in higher education seems a bit low compared to other nations. Mexico has an interesting downward trend, with far more than expected pre-primary age children in school, but then enrollment drops steadily as students get older. The US, unsurprisingly, has the highest percentage enrolled in higher education, but has comparatively low enrollment in pre-primary. South Korea has the highest average school life expectancy (16.8 years) while Ghana has the lowest (9.7 years). 

Source: UNESCO Institute for Statistics (individual country profiles); dates in parentheses reflect older data.

*Gross enrollment ratio (GER) is the number of pupils enrolled in a given level of education regardless of age expressed, as a percentage of the population in the theoretical age group for that level of education (hence some numbers going above 100%). Net enrollment ratio (NER) is the number of pupils in the theoretical age group enrolled, expressed as a percentage of the same population.

**England’s data are generalized from United Kingdom data, the only data available from UNESCO.

***Chile and Germany have the most data missing, so that’s pretty lame. Automatic yellow cards for them.

Table 2 breaks down each nation’s public expenditure on education as a percentage of both GDP and overall government spending. Thanks to currency conversion calculators, it also looks at starting salaries for primary teachers in US dollars as well as the amount spent on salaries as a percentage of overall spending on primary and secondary. Notice that most countries have similar expenditures as percentages of their GDPs (hovering from four to six percent) except for Uruguay at 2.6 percent. As a percentage of overall government spending, Mexico spends a large proportion on education – one in four public dollars. Starting salaries for primary teachers don’t reflect the cost of living in various countries, so consider these figures with a grain of salt. Germany’s is high, especially compared to other European countries with potentially similar costs of living. In Ghana, less than three percent is spent on non-teacher expenditures, with 97.4 percent of education dollars going to teachers.

Source: UNESCO Institute for Statistics (individual country profiles). Salary data from OECD’s “Education at a Glance 2006” (except for Chile’s: obtained from “Teachers’ Salary Structure and Incentives in Chile”). Dates in parentheses reflect older data. 

*England’s data are generalized from United Kingdom data, the only data available from UNESCO (except for starting teacher salary).

Table 3 looks at student-teacher ratios, literacy rates, and school calendars. Spain, Portugal, and Germany have low student-teacher ratios while Ghana, Chile, and Mexico have the highest. All nations except for Ghana have high rates of literacy among the total adult population, which is probably a reflection of an incredibly liberal definition of “literacy”—UNESCO says this is a “common definition – the ability to read and write at a specified age” – whatever that means. (We’d speculate that more than one percent of the US population is functionally illiterate.) Time spent in school varies widely across countries, and can even differ within countries; for example, Germany’s school calendar ranges anywhere from 188-208 days. Children in Ghana attend school for five days a week, nine months per year, but only attend for three hours a day. We’d also bet that various nations’ definition of “instructional hour” is loose (e.g., do Chilean children really go to school for 1,257 hours annually?)

Overall, Table 3 is informed by the widest variety of sources and reminds us of the international differences in terms of measuring instructional time, as well as the discrepancies in reporting and collecting data. In other words, the quality of some of this data differs about as much as international turf conditions, from Ghana’s dry soil to Germany’s lush green fields. 

Source: UNESCO Institute for Statistics (individual country profiles); dates in parentheses reflect older data. Literacy data found at CIA World Factbooks. School calendar data for OECD countries found at International Review of Curriculum and Assessment Frameworks Internet Archive found (www.inca.org.uk); for Latin American countries: “A View Inside Primary Schools: A World Education Indicators cross-national study” by UNESCO (2008); for Ghana, from “Achieving Universal Primary Education in Ghana by 2015: A Reality or Dream?” by UNICEF’s Division of Policy and Planning (2007); for Mexico, from “OECD Briefing Note for Mexico;” for Slovakia, from “Learning for Tomorrow’s World” (PISA 2003); for US, Netherlands, Japan, and Portugal – hourly estimate obtained from Ed Sector Analysis, “On the Clock.”

*To standardize the hourly estimates, we took an average of hours per day (averaging across grade levels) and then multiplied by the given number of days per year. Many of these numbers reflect averages and may not match the Latin American countries, where hours reflect the “mean number of hours of instruction in all schools, public and private.” 

**England’s data are generalized from United Kingdom data, the only data available from UNESCO (except school calendar data).

Finally, UNESCO collects some interesting data on how print-rich nations are (as measured by the total average circulation of daily newspapers), as well as investments in science and technology (measured by the number of researchers in the population and expenditures on R&D). South Korea and Japan blow everyone out of the water with investments in R&D of 3.47 and 3.45 percent of GPD, respectively, while the US and Germany hold their own. Paraguay and Uruguay have a dearth of researchers, while Latin American countries overall have fewer newspapers than their counterparts.

Source: UNESCO Institute for Statistics (individual country profiles); dates in parentheses reflect older data.

*England’s data are generalized from United Kingdom data, the only data available from UNESCO.

While these statistics might not be as scintillating as a header off a curving corner kick, or a lofty goal from the thirty-yard line, they’re interesting in their own right. If Cristiano Ronaldo weren’t flying down the field with his Euro-mullet, what can we say about him? Hailing from Portugal, he’s 77 percent likely to have attended secondary school; would have averaged 15.5 years in schooling, and been in classes with just 11 other kids in primary school. Kaka (Brazil) would have averaged 200 days per year in school but attended for only 14 years; his likelihood to enroll in secondary school is the same as Ronaldo’s. Landon Donovan would have done slightly better by some measures – 88 percent likely to enroll in high school, averaging 15.9 years over the course of his school career. (We’ve mentioned nothing about academic performance or discrepancies within nations; young fans of Donovan’s LA Galaxy MLS team attending some of Los Angeles’ public schools could attest to that.)

All in all, while most nations tout a commitment to educating their youth, there are many children who remain underserved. As we head into final games, here’s to raising a glass and hoping for a day when the world’s citizens demonstrate the same fervor for educational excellence and equality as they do for the beautiful game. 

Stay tuned for more World Cup education analysis from Ohio Gadfly on Flypaper, Fordham’s blog, as well as video commentary by Gadfly Studios.

The looming state budget crisis has become a political elephant in the room, with state leaders largely avoiding the topic ahead of November’s elections. Case in point, a state panel charged last September with recommending solutions to the crisis held its first meeting just yesterday.

Passing the FY2010-11 operating budget was a contentious and drawn-out process, and enacting the next budget is likely to be much harder. Taking such realizations into consideration, The Center for Community Solutions has released a report offering a plan to tackle the crisis. Thinking the Unthinkable: Finding Common Ground for Resolving Ohio’s Fiscal Crisis urges lawmakers to act in a bipartisan fashion and offers a plan to close the budget gap through a series of tax increases and targeted spending cuts.

The Center’s calculations on the severity of the deficit confirm a $6 to $8 billion hole. The authors assert that a deficit of this size cannot be reduced by relying solely on taxation or spending cuts and will require legislators and the governor to seek a “balanced approach” that combines both practices. Toward this end, a three-part framework of tax increases, reduction in tax exemptions, and reduction in expenditures is presented.

Tax hikes are always a particularly bitter pill to swallow in an election year, and as such the report offers a variety of options to increase state revenue. These range from raising income taxes on the wealthiest Ohioans to increasing the state sales tax by half a cent. It also explores a number of reductions in tax expenditures. Tax expenditures (more commonly referred to as tax exemptions, tax breaks, or “loopholes”) add up to a significant portion of Ohio’s revenue and several methods to scale them back are provided.

The most notable part of the framework outlines ways to cut state spending. Virtually all of the recommended cuts come from state services – primarily Medicaid. Costs in Medicaid are described as spiraling out of control, and the report offers possible technical and legislative fixes to allow the state to slow Medicaid growth while still meeting federal spending mandates. Under the report’s plan, many other areas of state spending would receive a targeted reduction in funding of 10-20 percent, depending on the amount of the budget shortfall.

In regard to education, the report calls for preserving formula funding (the basic per-pupil amount that a school receives from the state) for K-12, early childhood, and higher education, meaning that despite the state and national fiscal woes, primary funding for most public schools would go untouched.

It’s particularly useful that this plan is scalable for different deficit amounts and budget priorities. This makes it an interesting template for state leadership to examine as they determine the best way forward. At a time when few are willing to publically discuss the budget crisis, any food for thought on the issue is certainly helpful.

Matthew Chingos
Harvard University, Program on Education Policy and Governance
May 2010 

This working paper by Harvard research fellow Matthew Chingos is a crystal ball for Ohio. Chingos examines Florida’s 2002 class-size reduction (CSR) policy – a universal mandate similar to Ohio’s mid-2009 policy enacted via the governor’s evidence-based model of school funding—and finds that it had no impact on student achievement.

The report outlines previous class size research and puts the infamous Tennessee STAR experiment (frequently cited by CSR defenders) in context. Other research has since contradicted STAR’s findings, and minority children benefited more than non-minorities in that study – a finding that should caution against universal CSRs and encourage more targeted reductions, if anything.

Using student-level data from the Florida Department of Education, Chingos examines deviations from prior trends in student achievement at districts and schools required to reduce class size, as well as deviations among those districts and schools that weren’t require to do so. Simply put, the comparison group included districts and schools which had already reduced class size and so didn’t have to comply with the 2002 mandate. The treatment group focused on those that did have to reduce classes by at least two students per year. (Chingos explains clearly why looking at achievement before and after the policy isn’t appropriate since Florida also had all sorts of other policy changes at that time that may have contributed to rising achievement.)

To add insult to injury to CSR defenders, Chingo also found no positive outcomes on other indicators like absenteeism, suspension, or crime and violence. In exposing Florida’s class size disappointment, the report points out the poor assumptions underlying Ohio’s policy: it presupposes “that resources provided to reduce class size will have a larger impact on student outcomes than resources that districts can spend as they see fit.” Rather than requiring all Ohio schools to reduce class sizes dramatically, the state would be better off targeting CSRs to the students it would help the most – especially considering the exorbitant costs the mandate imposes on ailing school budgets.

Read the study here.

Center for Reinventing Public Education
Robin Lake, Brianna Dusseault, Melissa Bowen, Allison Demeritt, Paul Hill
June 2010

This publication contains interim findings from a nationwide study of Charter Management Organization (CMO) effectiveness. Started in May 2008 and expected to conclude summer 2011,  the work was commissioned by New Schools Venture Fund and is being conducted jointly by Mathematica Policy Research, Inc., and The University of Washington’s Center on Reinventing Public Education. The point of the project is to examine the effectiveness of CMOs (non-profit organizations that manage “more than one school with a unified management team responsible for delivering the educational program and supervising school leaders”) by investigating their operations, internal structures, growth strategies, missions, and challenges.

The findings come from 43 of the 82 known CMOs (as of 2007) and highlight educational programs, school cultures, teacher accountability systems, and the economics of starting and sustaining CMO operations. Most interesting among the takeaways is that while many of the CMOs firmly believe that their particular educational approach is critical to the success of their programs, their approaches differ widely. This confirms what, anecdotally, we’ve seen from an authorizer perspective: the school leadership and people at the front of the classroom matter most. It also gives credence to another anecdotal observation: all individual schools have different needs – even schools under the umbrella of the same management organization - and individual schools must have the ability to adjust quickly to meet internal (e.g., human resource) needs and the needs of students. Although this study focuses on charter schools, the latter observation likely applies to non-charter public schools as well.

The report also notes the challenges CMOs face with growth, particularly the fine line between developing internal procedures to efficiently manage operations and layering on rules and procedures, thereby becoming the very bureaucracy many of these CMOs sought to avoid. Interestingly, the larger the CMO, the more control the central office is likely to exert over the individual schools and their operations. 

The authors also set forth next steps in their research, which include four issues that definitely make this work worth revisiting in 2011 in the final report: 1) how differences between CMOs and districts relate to outcomes, 2) the costs and benefits of increased instructional time, 3) whether there is an optimal CMO size, and 4) an analysis of CMO cost structures and sustainability. Read it here.

Editorial Projects in Education
June 2010

Did you know that just 2 percent of Americans earned high school diplomas in 1870? That’s just one of the tidbits you’ll find in this year’s Diplomas Count (find 2009, 2008, 2007, and 2006 here). In addition to the usual graduation statistics update, this edition attempts to chr onicle “data in action,” i.e., how the smart use of information can raise graduation numbers, mainly by identifying students at risk for dropping out. As usual the news isn’t good: The graduation rate hovers around 70 percent, having actually declined slightly from 2005 to 2007 (the most recent year of available data). But that’s somewhat misleading, because while the overall graduation rate has fallen, rates for every racial group have improved. This is Simpson’s Paradox in action, as the lower overall number can be attributed to the fact that the population of students most at risk of dropping out—minorities, especially Latinos—is composing an increasing percentage of the overall student body. Perhaps most interesting is that just twenty-five of the nations’ 11,000 school districts account for a whopping 20 percent of all dropouts, or 250,000 students. New York City and Los Angeles are the worst offenders, each failing to graduate more than 40,000 students every year. Admittedly, these districts are the nation’s largest and would statistically have more dropouts than smaller ones, but it’s also a lesson in how much difference data systems could make by being used well in just a handful of places. The accompanying journalistic pieces are heartwarming: School leaders responding to data with tutors, coaches, and mentors; teachers tracking down missing students and helping them enroll in summer school; and district and state leaders improving the use of data by tracking across borders, even state lines, to make sure no child falls through the cracks. The bottom line is this: Dropouts remain a huge problem, but robust and consistent use of data is a promising solution. Read it here and check out the accompanying web materials here.

Mathematica Policy Research, Inc.
Christina Clark Tuttle, Bing-ru Teh, Ira Nichols-Barrer, Brian P. Gill, & Philip Gleason
June 2010 

This is the interim report of an ongoing (until 2014) longitudinal study of achievement in a quarter of KIPP's eighty-two schools. Though KIPP schools have been the focus of previous research (see here and here, for example), this is by far the largest and most rigorous study to date. And the results are encouraging. Using matched student achievement data from twenty-two middle schools that had been open since at least 2005-06, Mathematica analysts found statistically significant impacts on reading in fifteen of the twenty-two, and on math in eighteen. Conversely, just two schools had a significantly negative impact on reading, while one school had a significantly negative impact on math (in year 1), which actually reversed into a positive impact by year three. These positive effects are sizable, especially in math. After three years in a KIPP school, a student will have made on average 4.2 years of growth in math and 3.9 years of growth in reading. This was true even though KIPP included in its treatment group all students who were ever enrolled in a KIPP school during the study, including those who spent just one year at KIPP and subsequently left, as well as the results for two schools that lost their KIPP affiliation during the study and subsequently closed. That means these results are probably conservative in terms of students who remained enrolled at KIPP all four years of middle school because they hold KIPP accountable for students who actually were not at KIPP for the majority of their middle school years. Though KIPP surely deserves praise for these results, it should also be applauded for subjecting itself to such a rigorous assessment. Read it here.