40 Years After ‘A Nation at Risk,’ Using Schools as Local Laboratories

The 74 is partnering with Stanford University’s Hoover Institution to commemorate the 40th anniversary of the ‘A Nation At Risk’ report. Hoover’s A Nation At Risk +40 research initiative spotlights insights and analysis from experts, educators and policymakers as to what evidence shows about the broader impact of 40 years of education reform and how America’s school system has (and hasn’t) changed since the groundbreaking 1983 report. Below is the project’s chapter on school finance and education funding priorities. (See our full series)

Education thrives on innovation. While core practices in education have slowly and incrementally changed over time, the innovation and creativity of teachers and administrators have facilitated improvements by modifying the organization and delivery of education. In calling for solutions, A Nation at Risk (ANAR) repeatedly called on local “political and educational  leaders to search for solutions” and to use the “ingenuity of our policymakers, scientists, State and local educators, and scholars in formulating solutions.” Indeed, ANAR lauded the “local laboratory” model that decentralized schooling affords to test new models that could scale and disseminate throughout the country. This chapter takes aim at understanding how innovation in the organization and delivery of education both within and outside the classroom has affected the quality of education.

I start by looking at large-scale initiatives that sought to modify the ways school districts orga nized themselves. For example, the Bill and Melinda Gates Foundation provided significant funding for the small schools or the “school-within-a-school” initiative. The initiative sought to divide large high schools into smaller sub–high schools that might provide a greater cohort experience and allow for greater interaction between teachers and groups of students. After reviewing small schools and other out-of-the-classroom innovations, I turn my attention to innovation within the classroom. Since ANAR, major evidence on class size and the timing of schooling, among other innovations, has influenced policy and practice. This has affected how states and local districts organize and deliver education to students inside schools and classrooms. I review evidence on these and other innovative practices.

Get stories like these delivered straight to your inbox. Sign up for The 74 Newsletter

It is impossible to really track forty years of innovation across thousands of school districts. Many innovations were never disseminated, scaled, or evaluated. Other innovations became so popular and widespread that the editors of this series chose to dedicate entire chapters to them. For example, the rapid changes in technology with computers and the integration of standards and accountability systems are two innovations that have altered every classroom across the United States. In this series, Tom Vander Ark’s chapter on technological innovations and Michael J. Petrilli’s chapter on standards and accountability review these innovations and how they have shaped school organization, classroom practice, and teacher and principal accountability. This chapter provides a useful companion to those chapters.

In this chapter, I focus on four sources of system innovation and two sources of classroom innovation. For system innovation, I review evidence on small schools, magnet schools, superstar superintendents, and innovation zones. For classroom innovation, I discuss evidence on class size and the duration of schooling. In choosing the specific innovations, I have conveniently sampled practices that have expanded beyond a single district or school and have thereby shaped and influenced education policy and practice. I also try to focus on innovations where rigorous research has provided some hint as to the causal impacts of these policies.

While many of the innovations seem to have strong evidentiary bases, they remain seemingly underutilized. In some cases, the costs of the interventions remain prohibitively large. In other cases, the modifications needed to establish the interventions at scale compromise the capacity of the interventions to affect outcomes. I discuss other impediments to expansion, and more generally I discuss obstacles that inhibit the use of the “school laboratory” model.

Freedom & Flexibility: To Scale School Innovation, Take the Risk out of Failure

Systems innovations

“Systems innovation” refers to new policies and practices that modify governance and the structure of school systems. The specific innovations upon which I focus are those in school organization, creation of specialized schools, school leadership, and innovation zones, which provide schools and districts latitude to implement new practices and policies.

School size and organization

Perhaps the most notable of the systemwide changes was the small schools movement. In the early 2000s, nearly twenty-six hundred small schools were created nationwide. One of the major forces behind this investment was the Gates Foundation, which began in the early 2000s to fund the decomposition of large high schools. Based on its interpretation of existing literature, the Gates Foundation encouraged schools to maintain a size of four hundred students. As Tom Vander Ark, who at that time was executive director of education at the Gates Foundation, said, “Young people who attend smaller schools that provide a rigorous, personalized education and enable close relationships with adults are more likely to graduate and continue their education.” The basic theory was that personalized education and deeper relationships with teachers would improve the quality of education and provide better role models and coaching as students began considering and then pursued subsequent education.

Unfortunately, the short-term results did not generate the anticipated results. Over nine years, the Gates Foundation invested more than $2 billion in creating small schools, and in 2009, it “refined” its strategy. As Bill Gates wrote: “[M]any schools had higher attendance and graduation rates than their peers. While we were pleased with these improvements, we are trying to raise college-ready graduation rates, and in most cases, we fell short.”

Indeed, the short-term evidence on small schools was a bit pessimistic. At the time that Gates was making allocative decisions, the evidence was not strong. However, the tide of positive evidence was soon to come. Studies by Bloom et al., Schwartz et al., Barrow et al., and Abdulkadiroğlu et al. found positive, long-run impacts. The study by Abdulkadiroğlu et al., the only one that took advantage of randomized admis sion lotteries, reported that college attendance rates in New York improved by seven per centage points, with additional improvements in math and English scores on the state’s High School Regents Examinations. Additional work by Schwartz et al. argued that small schools improved the performance of most New York high schools, including schools that were not small schools.

Yet despite the positive, long-run evidence on small schools, they are no longer receiving the same public and philanthropic support they did in the past. Strong evidence has not revitalized the initiative or the funding streams. While existing small schools have remained, few additional small schools have been added. Why the lack of continued or renewed support?

40 Years After ‘A Nation At Risk,’ Fixing Our Classrooms Through School Funding

First, there was a perception that lessons could be applied and scaled up in other settings that did not require small schools. For example, one theory as to the success of small schools relied on the notion that students developed rich, meaningful, personalized relationships with faculty and counselors. However, such relationships may be possible in other settings as well. As Robert Hughes, director of K–12 Education at the Gates Foundation, explained, “[W]ith some work, you can really build structures that enable kids to be known and to get the kind of support they need to be successful [even] in larger schools.” Also, small schools cost more per student than traditional schools. If traditional schools can replicate the counseling and other relationship-based mechanisms, then expanding small schools, a far more expensive and involved intervention, may not be necessary. Indeed, the fact that all schools in a school district, including large schools, benefit from the presence of small schools suggests that small schools increase the visibility of some mechanisms that can be transferred to larger schools.

Second, there have been concerns about the capacity of institutions to scale small schools. One concern is that the cost of staffing might be prohibitively large. Another concern is that small schools raise the demand for both teachers and principals, and it is unclear that the supply of new teachers and qualified principals can satisfy the demand.

While small schools remain somewhat stalled, there are several aspects of the small schools movement that have had a lasting impact. For example, it was one of the largest, highest-profile experiments in innovating the structure and design of schools. This gave some momentum to districtwide interventions and experimentation, expanding the scope of the laboratories envisioned in ANAR. Also, as mentioned above, it provided significant information about reforming underperforming schools that could be applied in other settings. And there was also important heterogeneity in the impacts that provided additional policy lessons about the implications of certain types of schools — namely charter schools, which generated significant impacts. For example, the KIPP schools were cited by Gates in the 2009 announcement that the Gates Foundation was refining its investment strategy. Subsequent work such as Abdulkadiroğlu et al. demonstrated that there were specific strategies (e.g., high accountability) employed by charter schools that may have led to the greater impacts observed in some small schools as opposed to others. Each of these lessons reinforced the notion that school districts could be local laboratories for innovation.

Specialized schools

A second strategy that altered the ways districts organized schools had to do with school choice and the underlying supply of schools. In this series, the chapter by John D. Singleton focuses extensively on school choice, so I defer any discussion of charter schools and vouchers to that chapter. My discussion here centers on magnet schools. While many charter schools behave like magnet schools and vice versa, the administration of magnet schools typically continues under the direction of school district offices, whereas the administration of charter schools often moves outside the district’s purview.

Magnet schools existed long before ANAR. As districts grappled with how to desegregate schools, many created specialist or alternative schools to give parents additional options. The first magnet schools began in the late 1960s. The first large-scale experiment occurred in 1970 in Minneapolis, and the first specialized high school, focused on career themes, opened in Dallas in 1971.

As of 2016, there were 4,340 magnet schools across the United States, with the most common theming centered on science, technology, engineering, and mathematics (STEM); fine and performing arts; international baccalaureate; career and technical education; and world languages. Whereas before ANAR, magnet schools were often considered as a way to encourage desegregation, the major expansion and specialization of magnet schools took place after ANAR with greater emphasis on school choice.

The magnet school expansion happened for a variety of reasons. Some of it was driven by parents as they tried to find ways to improve their children’s educational performance by building on specific skills. Some of it came as magnet schools, particularly vocation-oriented magnet schools, demonstrated that they could have positive impacts on students who were struggling in mainstream classrooms.

To date, the evidence on magnet schools is largely positive. Gamoran (1996) uses the National Education Longitudinal Study to compare test scores of students at magnet, public, and private schools. Gamoran found that students at magnet schools score higher in science, reading, and social sciences. Crain et al. (1998) use oversubscription lotteries to measure the impact of career-oriented magnet schools. The researchers provide mixed evidence. On the one hand, career magnet schools had lower graduation rates than comprehensive schools.

This was the result of greater emphasis on career and vocational curricula. On the other hand, students who attended magnet schools reported fewer “reckless adolescence behavior” at age twenty. Kemple and Snipes provide evidence that career magnet schools are an effective way to reduce dropout rates among those at the highest risk not to graduate. A synthesis of the literature suggests that impacts are “generally positive.”

Magnet schools were refined in the laboratory of public schooling, and new iterations of magnet schools build upon the principles of school choice discussed by Singleton and on principles of innovation. Career education in particular has become much more central to education policy. As the returns to high school education have stagnated, emphasis on employability and skills has fueled much of the advance in career education. This concern not only has been present in the United States but also has become increasingly popular as a policy tool in developing countries. The evidence from magnet schools provided significant lessons that shaped early attempts to strengthen vocational education.

Magnet schools also raise the question as to whether education should be similar across students. In many ways, the traditional school model presents very little variety across the types of skills that students develop; however, magnets exist to allow some students to specialize beyond what a traditional school might allow. Heterogeneous students might need more heterogeneous offerings than traditional schools can provide. Magnet schools might be a way to improve the efficacy of education for a subset of students, and there may be limits to the degree of differentiation among students and schools that are possible. Hence, the gains could be large but diminishing as differentiation expands.

In the short run, magnet schools’ enrollment and presence will continue to expand. Whereas they were at one point an answer to desegregation and integration of schools, they are increasingly a means for parents to express their preferences in terms of students’ education opportunities. Their growth continues, and while charter school enrollment remains larger, magnet schools remain a viable way to provide differentiated education opportunities. Moreover, as charter schools become more specialized (e.g., STEM or vocation focused), the line dividing magnet and charter schools will continue to blur.

School leadership

Another trend that has taken place since ANAR is the increased emphasis on high-profile school district superintendents. In many large school districts, superintendents have become chief executive officers with greater power and salaries. For example, Barbara Byrd-Bennett served as the CEO of both Cleveland Public Schools and, later, Chicago Public Schools. She previously served in a leadership capacity for New York City schools. Her compensation in both Cleveland and Chicago was controversial, given the size of the packages. Other CEOs grabbed headlines and made national news as well, such as Michelle Rhee in Washington, DC, and Arne Duncan in Chicago Public Schools. In districts with more than twenty-five thousand students enrolled, superintendent compensation ranges from $140,000 to almost $400,000. To put this in perspective, the median base salary for a beginning teacher in districts of the same size is $44,150.

Prioritizing hiring high-profile CEOs with extensive experience and increasing their compen sation has been a prevailing societal trend. Just as CEOs’ track records were believed to have an impact on a company’s performance, schools sought to enhance their quality by appointing elite superintendents who would bring about substantial improvements and elevate the institutions they served. Hence, an increasing number of school districts applied corporate strategies to superintendents. In the corporate world, there was a perception that the supply of such leaders was finite, leading to bidding wars and large compensation. The emphasis on superstar superintendents faced the same competition and compensation.

However, as Chingos et al. demonstrated, superintendents who bring about significant, statistically reliable changes in student achievement within their districts, while controlling for other factors that affect academic performance, are indeed rare. They found that superintendents account for only a tiny fraction (0.3 percent) of student differences in achievement, significantly less than other factors such as student characteristics, teachers, schools, and districts. They further indicated that student achievement does not improve with longer superintendent service, and hiring a new superintendent does not lead to immediate gains in student achievement.

Increased emphasis on superintendents may not directly yield higher test scores, but nevertheless it remains an area of continual research. For instance, Hart et al. proved that encouraging superintendent longevity can support student achievement, as those with more in-state experience possess a comprehensive understanding of the state’s curriculum, testing programs, and the organizational stability required for effective leadership. Mitigating superintendent turnover, as suggested by Grissom and Mitani, could involve considering salary increases, particularly in smaller and rural districts and those with lower student achievement, as this would help retain superintendents who are often lured by higher-paying positions in larger, more urban districts with better academic performance.

The hiring and reliance on superstar superintendents is very much an experiment in progress. While some districts have moved away from the strategy of hiring CEO-like superintendents in favor of other approaches, there are still districts that continue to explore this path. Ongoing research and the findings regarding superintendent longevity and compensation emphasize the importance of considering contextual circumstances being faced or the necessity of exploring alternative strategies.

Innovation zones

While the principles of innovation zones may have been part of the policies dating back to the early 1990s, states and school districts began implementing legislation in the mid 2000s. Innovation zones are schools or districts to which states or districts grant greater autonomy over curriculum, budgeting, and staffing. Typically, states and districts grant innovation zones additional relief from other state and local regulations. While schools are free to enact policies and practices that differ from the norm, the schools are held accountable for improvements in student outcomes. After some early experimentation, innovation zones began expanding, and by 2017, they covered more than 108 schools and 63,000 students.

Innovation zones presuppose that regulation and centralization impair the ability of a district to try new and innovative practices. By providing greater autonomy, schools and districts can explore new practices in finance, governance, curriculum, and staffing. This greater autonomy comes at the cost of higher accountability for student outcomes, and schools and districts can lose the autonomy if student outcomes do not improve. More generally, innovation zones are just one category of school turnaround. Under the Race to the Top (RttT) legislation, the federal government funded school turnaround strategies that included other variants such as school improvement grants or No Child Left Behind (NCLB) waivers. These school turnaround programs also allowed school districts to have more autonomy in some aspects of staffing, management, and curriculum.

Many studies of innovation zones are emerging. Zimmer et al. (2017), for example, examine the innovation zones established in Memphis, Nashville, and Chattanooga, Tennessee. They showed that innovation zones significantly outperformed other public schools and other alter native methods of changing governance. Math scores, for example, increased by 0.20 standard deviations in innovation zones relative to other schools. Science and reading scores also increased. Zimmer et al. (2017) argue that one of the largest mediating factors in the innovation zone was the retention of experienced, successful teachers. Innovation zones in Tennessee generally offered significant raises for teachers who transferred to innovation zone schools. Teachers who previously had significant value added in the classroom were more likely to shift into these innovation zone schools. While the competition for high-achieving teachers may be a zero-sum game in the short run, the responsiveness to compensation incentives alongside the added autonomy may strengthen the overall workforce by inducing the retention and recruitment of top teachers.

The use of innovation zones and other strategies aimed at strengthening school autonomy remains a hot topic. In this series, Michael T. Hartney, for example, explores other innovations in governance and how they have played out. While the current scale of innovation zones is low, the case of innovation zones is interesting as the initial results have encouraged continued expansion, with at least twenty-five states having adopted policies encouraging innovation zones in districts that were previously classified as failing and more considering legislation to allow innovation zones.

The expansion of innovation zones raises questions about teacher supply. If, as Zimmer et al. (2017) argue, the mechanism by which innovation zones improve outcomes is through attracting top teachers at the cost of having other schools lose top teachers, then innovation zones might lead to a continued division between high-value-added teachers and others. If the higher wages and reduced legislation in innovation zones serve to attract more (and better) teachers to the profession, then innovation zones might generate momentum toward improving the overall teaching pool. However, if this does not happen, then the competition for teachers is a zero-sum game in which the available teachers for underprivileged schools will be disadvantaged relative to those who want to attend innovation zone schools.

Finally, the continued expansion of innovation zones has two implications for the future. First, the continued expansion of legislation allowing innovation zones suggests that these zones will become increasingly visible in the future. Second, given that the emphasis on innovation zones is both deregulation and expanded accountability, it also suggests a growing discontent with the existing regulations in traditional school districts. Innovation zones are a means of circumventing some regulations. If innovation zones eventually create momentum around deregulation, then deregulation might displace (or potentially devalue) innovation zones.

Classroom innovation

I next turn to classroom innovation. I focus on two separate innovations — class size and the  timing of schooling. As before, these are only a fraction of the possible innovations that I could use; however, these are two areas where significant experimentation and subsequent policy implementation have happened since ANAR.

Class size

Scholars from all disciplines have long postulated that class size affects student outcomes. The underlying theory suggested that teachers can give more attention to students in smaller classes and that this extra attention might provide a boost in students’ education outcomes. Lazear (1999), for example, uses a model to demonstrate that the probability of classroom disruptions likely increases as class size goes up.

However, in the mid-1980s, some doubt emerged on the relationship. In a series of papers, Hanushek (e.g., 1986, 1999) showed that estimates of the effects of class size were ambiguous. Perhaps students were not as sensitive to class size as they might have been to other inputs, or perhaps teachers used different technologies as class size changed. Nonetheless, the relationship between class size and academic achievement has been hotly contested in the education literature.

In 1985, then governor Lamar Alexander sponsored the Tennessee STAR experiment. The experiment created small classes in kindergarten through third grade and implemented the intervention with a school-based randomized controlled trial. The results of the experiment were stunning. Education test scores improved by roughly 0.25 standard deviations, roughly one grade level higher than students in regular classrooms. Subsequent research suggested that the impacts endured through primary, secondary, and tertiary schooling.

Some criticisms have been made of the Tennessee STAR experiment. For example, Hoxby discusses the possibility that the results are exaggerated. She argues that they may be the result of the Hawthorne effect arising from teachers performing differently than they would have otherwise as a result of participating in a high-profile experiment. Others have refuted this characterization, calling the Tennessee STAR experiment the “Barbary steed” of the class size literature. Nonetheless, Tennessee STAR influenced policymakers. States including California, Florida, and Texas established class size limits. The policy in California in particular provided extensive financial incentives to schools that implemented class size limits.

While the policy debate has leaned heavily in recent years toward reductions in class size, there have not been significant studies to date documenting whether state policies around class size have generated close to the same effects as observed in Tennessee STAR. In fact, there is some evidence that the emphasis on class size has come at the expense of other inputs. For example, Sims shows that California schools largely achieved class size reductions up to grade three by increasing class sizes in subsequent grades. He shows that the increase in test scores after grade three may have reversed potential positive impacts of class size.

The class size debates are not over and will likely continue for the foreseeable future. Until evidence can definitely show that the expansion of reduced class size through state policies leads to sustained improvements in student achievement, the debate over class size will continue. Even if Tennessee STAR’s evidence shows improvements in student outcomes as a result of class size reductions, it does not mean that class size reductions can produce impacts in scaled-up policies. In the Tennessee STAR experiment, Tennessee allocated additional funds. In scaled-up versions at the state level, the cost is likely prohibitive. States have to reallocate funds toward increasing the number of teachers and away from other inputs. In the case of California, it allocated funds to increase the number of early elementary school teachers, yet the cost of this was a decrease in the funds to hire teachers in other grades, and hence, higher class sizes resulted in those other grades. This could counteract any positive impacts from class size in early grades. Indeed, there is no evidence to date that California’s aggressive class size policy has led to any improvement in outcomes. The literature on class size largely focuses only on class size, but in a scaled-up policy, the improvements from class size must be weighed against the costs of reduced educational inputs elsewhere. As long as costs remain prohibitive, it is unclear whether any state can produce a class size policy that can replicate the gains from the Tennessee STAR experiment.

While ANAR did not necessarily take on the issue of class size, its call for local experiments to identify promising solutions resonates with the issue of class size. In considering changes in classroom practice, the debates on class size have led to significant investigations throughout the United States and beyond—not just in primary schooling but also in higher education. However, a limitation of experimentation can be its ability to understand how the impacts would change as scaled-up versions of the policy reverberated throughout the education landscape. The formulation of policy around class size has largely proceeded without finding a solution for the costs of reduced class size, and states have sacrificed other inputs in order to accommodate class size. While innovation is present in the case of class size, pushing innovation forward without considering the costs of scaling may never generate the promised impacts.

Timing of schooling

One input that was specifically mentioned in ANAR was the length of the school day and year. ANAR’s authors lamented that the United States had shorter school days and school years than its competitors. The ANAR authors strongly recommended a seven-hour school day and a school year of two hundred to two hundred twenty days.

For at least a decade after ANAR, there was very little movement or experimentation with the length of the school day. In 1997, Arizona became the first state to increase the length of the school year, requiring at least two hundred days of instruction rather than one hundred eighty. By 1998, fourteen states were considering changes to the school calendar; however, outside of a few districts in Arizona, few changes were happening at scale.

Since 2000, though, there have been significant changes in the time allocated for school ing. Some of these have come in response to the charter school movement. For example, from 2000 to 2012, the average length of the school day nationally increased by 0.2 hours; by contrast, the average length of the school day in charter schools increased by 0.4 hours. As Farbman noted, multiple studies of charter schools and other school turnaround efforts have attributed the impacts of charter schools, in part, to the length of the school day.

Additional evidence has come from outside the United States. For example, Germany increased weekly education instruction by two hours, thereby improving outcomes, particularly for high-achieving students. Studies in Chile, Israel, Italy, Brazil, and Latin America more generally have shown similarly positive impacts of increasing instructional time. These other studies have found greater benefits for both low- and high-performing students.

Within the United States, RttT grants for school improvement often targeted limited exper iments in the length of the school day and evidence to date. Some schools implemented changes in the length of the school day in response to these grants. More generally, the largest policy shifts have been in Chicago and Boston. In 2012, Chicago moved from a 5.75-hour school day to a seven-hour school day, and in 2015, Boston Public Schools approved a forty minute extension of the school day.

While the length of the school day has been the subject of both policy changes and exper imentation, there are few studies on lengthening the school year. The average number of school days has shown almost no change nationally and remains around one hundred eighty days.

What does the future hold? The mounting evidence on the impact of increased instructional time will likely increase pressure to consider policy options, particularly for students who are struggling. The continued expansion of charter schools, which have longer school days on average, will continue to put upward pressure on the length of the school day. Not only do they contribute to the increased average school day, but they also put pressure on districts to examine the amount of instruction time they offer. Areas where charter schools provide greater competition to local schools are likely to face greater pressures to increase instruc tion time. In terms of increasing the average length of the school year, there appears to be little momentum.

Other systemic shifts

While I have highlighted systematic changes that have focused on improving education quality, there are other systemic shifts that have occurred. Many of these have less to do with school inputs and more to do with the changing context of education. I briefly consider three examples.

Learning disabilities

Learning disabilities have become more prevalent over time. Zablotsky et al., for example, report a significant increase from 2009 to 2017 in the percentage of children diagnosed with any developmental disorder, attention deficit disorder, and autism. Special education enrollment rates continue to rise. Moreover, the COVID-19 pandemic increased attention on issues of mental health among students.

These changes in health have impacts on classrooms. Students with disabilities have renewed protection and have increased access to accommodations as a result of the Americans with Disabilities Act and the expansion of “504” plans. Students with disabilities are often more expensive to educate, costing as much as thirteen times that of the average student, and the increased incidence of documented disabilities puts financial pressures on schools. While schools receive additional funds for students with disabilities, the marginal cost of educating a student with disabilities is likely higher than the increased allotment. Indeed, Bergman and McFarlin showed that charter schools actively discriminate against students with disabilities in the way that they encourage (or discourage) enrollment. The reason for this discrimination is likely the disparity between the cost and revenue associated with a student with disabilities.

While the increase in disability diagnosis and treatment will certainly improve education quality for those with disabilities, the additional education expenditure required to teach students with disabilities inevitably leads to reductions in expenditures elsewhere. Increased expansion of charter schools, if indeed charter schools discriminate against students with disabilities, could exacerbate existing inequalities by segregating students by costs. Improvements in our ability to diagnose and treat learning disabilities can reduce the costs of educating students with disabilities and reduce the fiduciary burden.

School safety

School shootings have become more commonplace, and a frequent motivation for students to pursue charter or private schools is often school safety. Since ANAR, the presence of police, metal detectors, and other security enhancements has shifted the ways schools behave. While the prevalence, particularly in the wake of violent shootings, seems high, in truth there has been a decline in the rate of victimization and threats to teachers from 1994 to 2016.

School safety, including policies and procedures to ensure safety, continues to be a hot topic in state and federal legislation. Each school must maintain a plan for ensuring safety and for dealing with school violence. As in the case of increased disability rates, a focus on school safety requires resources and attention. Governments have been reluctant to increase funding to fully cover the costs of such expenditures. The resulting policies create more pressure on schools to cut expenditures in other ways. Moreover, to date, there has been little experimentation in ways that can help identify cost-effective strategies for improving school safety. Using schools as laboratories across the United States could provide greater opportunities to learn best practices.

Parental inputs

Education scholars have often posited that parental inputs are a significant part of students’ academic achievement. While the correlation between parental characteristics is extremely strong, particularly in the case of the mother’s education, few papers have established causal relationships between parental inputs and student outcomes. Many localized experiments have attempted to increase parental involvement; however, none of these efforts have scaled in any meaningful way.

There are a couple of notable exceptions in more recent years. While not occurring in the United States, the Oportunidades conditional cash transfer program was a major randomized controlled experiment in Mexico that targeted parents and students. Parents received a subsidy conditional on student attendance and student health visits. These programs had a demonstrable impact on student attendance and attainment. While parents are clearly involved in the treatment, it is not clear if the effects came because of their vigilance or other factors (attendance, health, or increased family income).

Recent randomized controlled experiments have aimed at a more novel approach to encouraging parental involvement. With the expansion of texting capabilities, researchers have used text messages to try to engage parents. York and Loeb did this for literacy among low-income parents. Through a series of text messages, York and Loeb coached parents of young children how to teach literacy skills. They found that students arrived at kindergarten with improved literacy as a result of parental engagement.

Bergman and Bettinger et al. tested interventions that focused on communicating with parents about students’ academic behaviors. Parents received notes about students’ truancy and assignment completion. These notes reshaped parents’ beliefs and led to improvements in attendance and academic achievement. Bettinger et al. demonstrate that the saliency of the messages in informing parents of important behaviors that they should be monitoring was likely the mechanism by which this impacted student achievement. In both cases, the cost of the intervention was small relative to the benefits.

New and innovative research designs are extending more and more interventions to parents, and this remains fertile ground for the laboratory of public schools. The recent text message interventions are particularly cost-effective and may have more potential to scale.

Conclusion and discussion

When I was invited to write this chapter, the charge was to document how education systems changed after ANAR. How does one capture forty years of trial and error, of innovation and failure? I have chosen to identify a handful of innovations within districts and within schools. These examples — using small schools, specialized schools, school leadership, and innovation zones as cases of education system innovations and class size and using the time spent in schools as cases of classroom innovation — are just a sampling of innovations that have changed, at least incrementally, the way in which education is delivered. One only needs to browse the research-related web pages of organizations such as the American Institute for Research, MDRC, Mathematica, and RAND Corporation to learn about the breadth of continued experimentation and innovation in schools.

Perhaps the most disappointing aspect of many of these interventions is their failure to be scaled up or to generate impacts when scaled up. The key problem in many cases is the cost of scaling. Oftentimes, scaled versions lack the same features as the original laboratory experiment, and in many cases, funding the scaled version requires sacrifices in other areas. Moreover, in many cases, we lack the supply of personnel or funding to move forward. Perhaps the great challenge of the next forty years will be learning how to create cost effective versions of the innovations that laboratories produce.

ANAR envisioned an education ecosystem where experimentation and learning from the laboratories of local schooling provided lessons and accelerated the process of change. While one can debate the relative quality of education over time, education systems of experimentation and learning across organizations have greatly increased, especially with the advent of the internet and the role of social media in drawing attention to innovation and to evidence.

The What Works Clearinghouse and other formal and informal collections of evidence on innovative practice and policy only accelerate the role of education institutions as laboratories in identifying promising practices and moving them to scale. However, finding and scaling the products of these laboratories remains the next challenge to be solved if the vision of schools as laboratories is to yield long-run improvements in the quality of education.

See the full Hoover Institution initiative: A Nation At Risk +40.