When former U.S. congressman and Obama administration chief of staf Rahm Emanuel marched triumphantly into the Chicago mayor’s office in 2011, he promised to revamp Chicago Public Schools (CPS) in ways that had barely been contemplated in 16 years of mayoral control over the city’s sprawling public-school system.

Longtime Chicago mayor Richard M. Daley had won control over the school system in 1995 and generally received accolades for rising scores on state tests; hard-charging superintendents, including Paul Vallas and Arne Duncan; tough accountability measures such as reduced social promotion; and a slew of new schools and shiny buildings. But the state, the city, and the schools were struggling financially. Chicago shed 200,000 citizens between 2000 and 2010, the only one of the nation’s biggest cities to lose population during that time period. Among big cities, Chicago emerged from the 2008 recession in worse financial shape than most. The 2007 teachers’ contract gave educators a 4 percent salary increase, boosting their compensation well above the state average.

And by the time Daley departed in 2011, the 400,000-student CPS had lost its reputation as a cutting-edge school district. Its NAEP (National Assessment of Educational Progress) scores were lackluster (see Figure 1).The district was still offering kids just 170 school days and just over five hours of instruction per day (compared to the national average of nearly seven hours). Chicago had never even made the finals for the Broad Foundation Prize, which is awarded annually to an urban district for marked improvement in student achievement. There were barely more than 100 charter school campuses, serving fewer than 40,000 students. As was the case statewide and elsewhere, the pension time bomb was ticking; pension obligations would soon require more than 10 cents of every dollar CPS spent. At the end of the 2010–11 school year, CPS faced a $1 billion budget deficit.

“There is no other school district in the nation that has as grotesquely underfunded a school system as Chicago,” said Laurence Msall of the Civic Federation.

To make a real difference, the impatient Emanuel would have to restore the city’s reputation as an education innovator, retain and attract more parents of school-age children, and confront a series of financial challenges. While no one knew it for sure at the time, there would be no federal Race to the Top funding, nor even, as it turned out, a federal waiver from No Child Left Behind that would free up existing federal dollars for new programs in Chicago. Emanuel and CPS were on their own.

 

Rushing through the First Year

In April 2011, Emanuel announced an education leadership team headed by Haitian American former Rochester, New York, superintendent Jean-Claude Brizard. There was no listening tour or even a pretense of considering a variety of policy options. (There wasn’t even time for Brizard to speak at the press conference announcing his appointment.) Emanuel was going to do bold, concrete things—enact a longer school day and year, implement principal performance bonuses, expand International Baccalaureate (IB) programs, and revamp teacher evaluations—and get them done as quickly and visibly as possible.

Along with the massive pension liabilities and a long-standing oversupply of school buildings, the district also faced a newly energized Chicago Teachers Union (CTU). The CTU had recently elected a rookie president named Karen Lewis straight out of her high-school classroom. She and Emanuel clashed almost immediately, and Lewis in particular found that her name-calling was effective in the media and with her members. (She claimed that Emanuel swore at her in private, too.)

It didn’t help Emanuel’s relationship with CTU that nearly the first thing he did was to declare that the district was in a “fiscal emergency” and to rescind the teachers’ 4 percent raise for 2011–12, the last year of the final Daley contract. The move was legal under the contract terms and was warranted, according to outside observers, including the Chicago Sun-Times editorial writers. Indeed, teacher salaries in Chicago were higher than in many other large U.S. cities (see Figure 2). At the time, Emanuel might not have been overly concerned that his actions would be extremely unpopular with teachers. Early indications were that Karen Lewis was an effective speaker but not a particularly effective negotiator. Lewis and her team had been steamrolled in Springfield just a few months before, when the state legislature passed a bill that limited Chicago teachers’ ability to strike.

Emanuel’s next move was to call for an immediate citywide extension of the school year and day for 2011–12. But such a move required teachers to waive the current contract, and teachers weren’t inclined to waive the contract just to please the mayor who was taking away their raises. When the dust settled, just 13 schools voluntarily extended the day that year, and the mayor’s rushed effort seemed to have backfired.

Things weren’t going very smoothly inside CPS headquarters, either, thanks in large part to the fact that Brizard had never run a system as large as Chicago’s, hadn’t picked his senior management team, and effectively shared control of the system with Emanuel’s City Hall deputy and his appointed board members. No one was really in charge.

“On any given day, there was a definite lack of clarity on where we were going and how we were going to get there,” recalled a former Brizard staffer, who did not wish to be named. Internal politics “consumed a lot of effort,” and verbal tirades from senior managers weren’t unusual. “It took a lot of work to get the work done,” she recalled. “You felt the turmoil every day.”

 

The Impossible Strike

In May 2012, a year into his administration, Emanuel appeared in San Francisco at the annual NewSchools Venture Fund summit, appealing to education funders, reformers, and high-performing charter management organizations to consider coming to the Windy City. Few would take him up on it. The reimbursement rate was too low. The political waters were too choppy. A handful of key CPS staff had already left. Rumors were circulating that Brizard would be forced out soon, and that the teachers might go out on strike.

In theory, striking was nearly impossible. The 2011 law dubbed Senate Bill 7 (SB7) specifically barred Chicago teachers from striking without 75 percent of members’ support, a threshold so high that no one imagined it would be possible to attain. Emanuel and his allies had put the provision into the law hoping to prevent a strike threat from blocking them from making tough moves in the contract and elsewhere. Stand for Children head Jonah Edelman had explained it all in what would become an infamous Aspen Ideas Festival panel appearance the summer before.

What no one seemed to have contemplated at the time was that CTU would open a strike vote well before negotiations had bogged down, but while teachers were still in the classroom, and keep the vote “open” until they got to the required 75 percent. They would essentially precertify a strike and have the members’ approval in their pocket during a long hot summer of negotiations. But that’s exactly what happened.

And so, when negotiations predictably bogged down at the end of the summer, Chicago teachers went out on strike for seven school days at the start of the 2012–13 school year. It was the first such interruption of classes in Chicago in more than 25 years. Neither New York City nor Los Angeles had experienced a strike in at least a decade.

“I remember how many people were on the streets on the strike during the Board meeting,” said special education advocate Rodney Estvan, a former classroom teacher. “The streets were literally filled with one group or another marching down, with their red shirts. It was stunning.”

According to some observers, including Sun-Times columnist Fran Spielman, SB7 “inadvertently” helped generate a strike vote against CPS. According to others, CTU was going to strike one way or the other. Instead of turning Chicago into school reform central, Emanuel seemed to be turning the city into organized labor central.

Who Won?

When the strike ended with a new contract, Chicago teachers generally seemed satisfied with the results Lewis had secured, voting to ratify the contract and basking in the national attention they had won. Lewis and her team would be handily reelected a few months later.

The union didn’t recoup the 4 percent raise that had been rescinded the year before; instead it won raises of 2 to 3 percent per year (or an average of a 17.6 percent pay raise over the period of the contract) on top of the traditional “step-and-lane” raises for experience and degrees. The new contract blocked merit pay. The “fiscal emergency” escape hatch was removed.

Among political observers, it was nearly unanimous that Emanuel “lost” the strike. The strike made for great theater, taking place in the president’s hometown just a few weeks before the 2012 elections, an absolute embarrassment for City Hall. Substantively, many thought that the mayor gave too much away, given the budget and pension problems that were looming. Raises in the new contract were “a significant cost added on without any plan for how it would be paid for,” observed budget watchdog Laurence Msall. Msall estimated that the first year of the new contract would cost CPS roughly $100 million.

Emanuel took the opposite view, airing commercials that listed his wins: a longer school day, principals’ freedom to hire the best teachers, parents’ freedom to choose the best schools, and a teacher evaluation system that would factor in student achievement. Starting immediately, school would run for seven hours for elementary schools, and seven and a half for high schools, and the year would last 180 days (up from 170). Teachers’ hourly rate (which many focused on as a barometer of their compensation) would be lower. There were few job protections for teachers against future layoffs or downsizing that were looming on the horizon. The union agreed to cancel its long-running lawsuits against layoffs that had taken place in 2010, and to continue to allow principals to hire and fire teachers, for the most part.

To start the 2012–13 school year, the union also “won” the resignation of Brizard, who would say in an August 2013 interview that Emanuel was a micromanager, and that he and CPS failed to take CTU and community and parent groups seriously enough. “We severely underestimated the ability of the Chicago Teachers Union to lead a massive grass-roots campaign,” Brizard said.

Indeed, the year leading up to the strike had proven to be a powerful opportunity for Chicago parents to bond with their children’s teachers and union organizers. The union made the case that it was striking for the schools and the students, not for themselves, and many parents seemed inclined to believe them. If parents had any complaints about the strike or the contract provisions won by the union, they were relatively few.

“From a parent’s perspective, I could care less about [the union’s rehire] priority list,” said Jill Wohl, who had helped launch a parent group called Raise Your Hand in 2010 to fight off threatened budget cuts. Otherwise, parent and community groups were generally aligned with the teachers.

Closing Schools

Brizard’s replacement, Barbara Byrd-Bennett, wasn’t as enthusiastic about new and charter schools as Brizard had been, but she was more experienced with larger districts. Most importantly, she was better suited to respond to the public attacks from the union and the near-constant media criticism that would come during 2012–13 as CPS moved to close schools for underenrollment (see Figure 3).

Initial reports were that 80 to 120 schools would be closed. Three hundred schools were on the list of possible candidates. Almost immediately questions swirled about which schools were on the list, and how the calculations were generated, and concerns about the safety of students having to travel to new schools. Incremental declines in total enrollment were just one part of the story: demographic shifts and the appeal of charter schools to African American families both contributed to half-empty buildings being clustered in a small number of neighborhoods.

In the end, the district decided to close fewer than 50 underenrolled elementary schools. Even those who were initially enthusiastic about Emanuel’s education ideas recoiled at the treatment parents received during the school-closing process.  “No one trusts you and for good reason,” complained Seth Lavin, a former Teach For America corps member and teacher whose neighborhood school was unexpectedly slated for possible closure (though later reprieved).

A few teachers blamed the contract Lewis had agreed to for the closings. Some defended the closing process as difficult but necessary, despite its flaws. But most everyone blamed CPS and City Hall. Sun-Times’ editorial page writer Kate Grossman termed the experience of parents and teachers, who for months feared having their school closed without knowing why, a “horrendous rigamarole.”

Emanuel continued to promote charters using the bully pulpit, and CPS was approving more charters even as the district was closing traditional schools. Concerned about the growing political pushback, CPS made two agreements that would likely prove incredibly hard to maintain: a five-year moratorium on any further enrollment-related closings and not to lease or rent empty buildings to charter schools. Schools receiving new students from closing schools were promised upgrades, including iPads for students and air conditioning.

Not surprisingly, given Chicago’s budget woes, the school closings were followed by layoffs and budget cuts. In June, nearly 900 teachers and support staff lost their jobs. In July, another 2,100 were laid off. The layoffs were still an unwelcome follow-up to the strike, and they were made worse by an ill-timed move to a student-based school funding system that put unhappy principals in charge of making layoff decisions. CPS made the budget-cutting process worse by insisting, despite 18 months of talk about a dire money problem, that budgets weren’t going down for 2013–14. Overall spending was up, but funding sent to schools went down 3.5 percent (roughly $68 million).

Since last spring, Lewis has kept up the pressure on Emanuel and CPS. In April 2013, she announced that she would register hundreds of thousands of new voters and run candidates for city council and the state legislature who supported the schools against those who didn’t. “If the mayor and his hand-picked corporate school board will not listen to us, we must find those who will,” declared Lewis. In a June speech at the City Club in downtown Chicago, Lewis asked her well-heeled audience, “When will we address the fact that rich white people think they know what’s in the best interest of children of African-Americans and Latinos, no matter what the parents’ income or education level?” Thanks to Chicago’s long history of youth violence and gang territory conflicts, there were concerns that sending so many students to so many new schools would create additional incidents before and after school.

Working Together to Win Resources

Chicago now has 681 public schools, including more than 130 charter campuses that serve 50,000 of the district’s 400,000 students. The 2013–14 school year opened generally smoothly, with students from the 50 closed elementary schools enrolling at new schools. Some l,600 part-time “Safe Passage” workers lined the streets to watch many of the students come and go from school. The chaos and violence that some had predicted for the start of the year didn’t take place. By the end of September, gun violence had claimed its first CPS student, a 14-year-old boy, but the first two months of the year were relatively quiet. Roughly 1,000 laid-off teachers were rehired. The anniversary of the 2012 strike came and went without any major upheavals.

Behind the scenes, there were some good things going on. A new, more-intensive teacher-evaluation program, developed and piloted largely without rancor during the previous year, was implemented citywide. It created a slower, tougher tenure process for teachers, and would base layoff and recall procedures on teacher evaluations rather than on seniority.

Common Core training was well under way across the city. And the district had finally created a single performance measure that would rate district and charter schools the same way. High school graduation rates have steadily improved.

Still, Chicago isn’t out of the woods yet. State test-score data show that in 2013 just over one-third of CPS 11th graders were proficient in reading, compared to 55 percent of their peers statewide. The district’s $5.6 billion operating budget is balanced, technically, but in so doing the school district has left itself extremely vulnerable to financial catastrophe. There will be another $1 billion deficit to face next year and hardly any reserves. Only about 20 percent of the $970 million in savings came from actual spending reductions. The city’s teachers’ pension was funded at just 55 percent of what it should be, leaving a whopping $8 billion shortfall (see Figure 4).

Enrollment in district schools was down again in 2013–14 in parts of the city’s South and West Sides. District revenues were down, too, for the first time in nearly 20 years. And thanks to teacher raises and benefits, among other things, operating costs are up.

Experts disagree about how much money CPS really needs to educate students, but there is no question that Illinois ranks near the bottom among states in its support for education. The state’s contribution was a measly 25 percent, and its targeting of funding (toward low-income, educationally disadvantaged children) was weak.

Chicago’s per-pupil spending in 2012–13 was $13,400, higher than the national average and what is spent in Los Angeles and Houston, for example, but far lower than what is spent in New York City or Boston. Some in Illinois nonetheless believe Chicago already receives more than its share. About one-third of the CPS budget comes from state coffers. Adjustments to the education funding formula made in 2000 have enabled Chicago to lower the property wealth used in the state’s funding calculations, an advantage that many outside the city believe is unfair. The city counts nearly all of its students as low-income for the purposes of state poverty-grant funding.

All that said, Chicago isn’t the only urban school district in the nation struggling with the demands of educating a large number of high-need students. In New York City, state contributions are arguably out of line with student needs. In Philadelphia, enrollment declines are creating enormous problems. In Detroit, the pension problem is even worse.

Chicago isn’t going to turn into Detroit anytime soon, though, thanks to a strong base of property owners. Emanuel nonetheless describes Detroit’s bankruptcy declaration as a “wake-up call.” City property-tax rates are comparatively low, any substantial increase in rates is unlikely to win enough public support to make it through the required referendum, and that hike still wouldn’t raise enough money.

Under normal circumstances, the city’s board of education and the unions would join forces to win more state funding and seek pension relief for Chicago schools. In September 2013, Emanuel made a public attempt to reconcile with Lewis for this very purpose, but it didn’t go far. Even when the Chicago Park District went to Springfield and won some significant pension reforms, CPS and CTU still held off joining forces. The city and union continued to put out wildly different ideas about how to increase funding for Chicago’s schools.

In early December, the state legislature convened and passed a controversial pension reform bill for the state as a whole. But the Chicago teachers’ pension was left out of the deal, and union president Lewis said she wouldn’t agree to anything similar.

For his part, Emanuel was still pressing for a delay to address the city’s pension situation, as unfunded pension liabilities for the city alone reached $19 billion. A $600 million payment would be due in 2014 unless the state gave the city the seven-year extension Emanuel was pushing for. “I’m going to turn this battleship around,” Emanuel told the Sun-Times, but “I’m not going to reverse 30 years of bad practices in just three years.”

Alexander Russo is a former teacher, researcher, and U.S. Senate education aide. He writes two education blogs, “This Week In Education” and “District 299,” and is the author of Stray Dogs, Saints, and Saviors: Fighting for the Soul of America’s Toughest High School.

While the competition is formidable, one case today best illustrates why the judiciary is ill suited to crafting education policy. Flores v. Huppenthal (formerly Horne v. Flores and Flores v. Arizona) has been in federal court since 1992 and could very well stagger on for several more years. Litigation, Flores shows, is time-consuming, costly, subject to political manipulation, and prone to prompting unpredictable policy reforms that even the plaintiffs may not wish for. When the Supreme Court heard the case in 2009, it cautioned against judicial adventures in reforming educational institutions. Recent developments give this warning some measure of vindication.

Four years ago, the Court held that Arizona should be allowed to argue that expenditures should not be considered the sole criterion for its English Language Learner (ELL) program. The plaintiffs had argued that Arizona was underfunding ELL instruction in violation of the federal Equal Educational Opportunities Act (EEOA). In response to the lawsuit, Arizona modestly increased its funding for ELL students and adopted a new policy requiring them to spend four hours a day in special language classes. Students would transition to regular instruction after testing proficient in English.

The plaintiffs replied that the funding increases were insufficient and opposed the instructional reforms. Four hours of language instruction amounted to segregation, they contended, and made ELL students fall behind in other subjects. The trial court and the Ninth Circuit agreed that both the funding and the reforms were inadequate, and even refused to let Arizona argue that the instructional changes were working and would bring it into compliance with the law.

The Supreme Court, however, remanded the case back to the Ninth Circuit and ruled that Arizona at the least should be able to argue that its instructional reforms were working before deciding whether more money was necessary. As well, the court worried that the case had become collusive because the governor, who was then a Democrat, Janet Napolitano, had refused to defend the state and thus was using the lawsuit as a political lever against a Republican legislature.

Despite winning before the Supreme Court, Arizona still was not out of the judicial woods. The same trial-court judge, Raner Collins, would hear the case on remand. Prior to the Court’s ruling, his attitude toward Arizona could best be described as hostile. He twice found the state in civil contempt and imposed $21 million in fines. Few would have been surprised if he had found that Arizona’s four-hour model was unlawful, leading to a fresh round of judicial interventions.

After concluding a four-month hearing in January 2011, he inexplicably waited to issue a judgment until March 2013. Even more surprising than his delay was his decision: he sided with the state. He quickly dismissed the argument that the program was discriminatory. He conceded that the state had adopted the four-hour model for the purpose of helping ELL students. Grouping students by English proficiency, like other groupings by ability, was not “segregation” and was in fact allowed under the EEOA. He then noted that ELL students had made progress since 2006, and that it was not his job to decide whether the four-hour program was “ideal” or whether the plaintiffs had an even better plan. He said the state has great latitude to set education policy, and with at least some hint at exhaustion, announced that “this lawsuit is no longer the vehicle to pursue the myriad of educational issues in this state.”

The plaintiffs naturally did not accept this bitter pill from their former ally and immediately appealed to the Ninth Circuit. It is not unreasonable for them to hope that Collins’s decision will be overturned there. But even under the best of circumstances, it would take several more years before any judicial action would change Arizona’s policy. Thus, after 21 years, most of which were spent before a compliant district and appellate court, the primary consequence of the litigation has been the adoption of an educational model that the plaintiffs oppose and under which at least an entire generation of students will be taught. Such results should give even the most hardened public-interest attorney pause before racing to the courthouse rather than the ballot box.

Joshua Dunn is associate professor of political science at the University of Colorado–Colorado Springs. Martha Derthick is professor emerita of government at the University of Virginia.

The Kalamazoo Promise provides college scholarships to graduates of Kalamazoo Public Schools (KPS), a midsized urban school district in Michigan that is racially and economically diverse. Anonymous donors promise to pay up to 100 percent of college tuition for any KPS graduate who attends a public college or university in Michigan. Scholarships start at 65 percent of college tuition for students who enrolled in KPS in 9th grade and stay until graduation, and increase to 100 percent for students who have attended since kindergarten. The scholarship is not limited to students with strong academic records or demonstrated financial need. Students must simply get into college and maintain a 2.0 grade-point average (GPA) while enrolled. Announced in 2005, the Kalamazoo Promise is unusual among scholarship programs in its universality and generosity.

As a model for revitalizing an urban school district and its community, the Promise, as it is called, has attracted much attention and many imitators. At least 24 other areas around the country have launched or are trying to launch Promise-style programs, with private or public funding. In part because of the Promise, President Obama delivered the commencement address to the graduating class of Kalamazoo Central High School in 2010.

Given the popularity and cost of the programs, there is a need for research to determine the size of the benefits of Promise-style efforts, or indeed, whether there are any discernible benefits. The tuition subsidies available through the Promise should create incentives for higher academic performance. Students who might otherwise attend Western Michigan University (WMU), the state campus located in Kalamazoo, may use the tuition subsidy to enroll at higher-ranked schools such as Michigan State University or the University of Michigan. Students who would otherwise have attended the local community college may use the subsidy to matriculate at WMU. And students who without the Promise might not have attended college at all may take advantage of the benefit to go to a community college. Previous research suggests that the Promise has in fact increased interest among eligible KPS graduates in enrolling at Michigan’s flagship schools.

Because admission to and graduation from more-demanding colleges requires students to have stronger academic preparation, the tuition benefits of the Promise also have the potential to encourage students to work harder and achieve more in high school. Yet despite the Promise’s incentives for improving performance, students may not respond as hoped. Some students may view the Promise’s benefits as too uncertain and too delayed, and students who want to improve their academic performance may not know how to do so.

Our study takes advantage of the unexpected announcement of the Kalamazoo Promise to study its effects on student achievement and behavior in high school. Specifically, we examine how the achievement and behavior of individual students eligible for a tuition subsidy changed after the program was launched. We find clear evidence that the Promise reduced student behavior problems. Our results on academic performance for all students are unfortunately not precise enough to draw strong conclusions. Even though our estimates of the program’s impact are not statistically significant, we cannot rule out the possibility of substantial positive effects. For African American students, however, the Promise both improved behavior and had a dramatic positive effect on high school GPAs.

Kalamazoo Public Schools

A predominantly urban school system, the Kalamazoo Public Schools have faced declining enrollment for many years. This reflects, in part, the modest pace of economic growth in Michigan and Kalamazoo. It also reflects the Kalamazoo district’s location in a center city that has more economic problems than the surrounding metropolitan area. For example, family poverty rates as of the 2000 census were 13.6 percent in the city of Kalamazoo and 6.5 percent in all of Kalamazoo County.

The Kalamazoo school district has historically had many low-income students and students from diverse ethnic backgrounds. In the years before the Promise was announced, although KPS retained a considerable percentage of white students and students who did not qualify for free or reduced-price lunch, the percentage of both groups of students was falling. Since the advent of the Kalamazoo Promise, KPS enrollment has been on the rise. Furthermore, enrollment has climbed proportionately for all ethnic groups, so the ethnic percentages have stabilized (see Figure 1). Previous studies indicate that these patterns reflect a large one-time increase in the number of students entering KPS the year following the Promise’s announcement, accompanied by a permanent decline in the exit rate among students already enrolled.

The Kalamazoo Promise was funded by anonymous donors who, according to the school district, had three primary goals: 1) to promote local economic and community development, in part by attracting parents and businesses to the Kalamazoo area; 2) to boost educational achievement and attainment; and 3) to increase confidence in KPS.

The Promise is available to all students who reside in the district, graduate from KPS, and have been KPS students for four years or longer. The scholarship must be used within 10 years of high school graduation. The size of the benefit depends on the length of continuous enrollment in the KPS system. The Promise covers 100 percent of all tuition costs and mandatory fees for up to four years for students who enrolled in KPS in kindergarten. For those who enrolled in grades 1 through 3, the scholarship covers 95 percent of those costs. The tuition benefit decreases by 5 percent per year as the number of years of continuous enrollment declines, covering 65 percent of costs for those enrolling in 9th grade. A student entering KPS in grade 10 or afterward is ineligible for Promise tuition benefits.

Other than onset of continuous enrollment, no aspect of a student’s K–12 experience or family background affects eligibility. Students do not have to demonstrate financial need, maintain any minimum GPA in high school, or take any particular mix of courses. Students obviously need to gain admission to a Michigan college, however, to receive Promise benefits.

The scholarship is available to all students who are admitted to and enroll at any public university or community college in the state of Michigan. The students must attend full time (defined as a minimum of 12 credit hours) and maintain a 2.0 GPA while in college. Students who fall below a 2.0 GPA can become eligible again for the Promise if they continue attending college on their own (or their family’s) dime and succeed in increasing their cumulative GPA above the 2.0 threshold. Students are eligible for Promise benefits for up to 130 credits of undergraduate college or university education, typically enough to complete a four-year degree program. The Promise’s benefits can be applied to certificate programs at community colleges, in addition to programs leading to an associate or bachelor’s degree.

The Kalamazoo Promise has been widely used. Between 80 and 90 percent of KPS graduates have been eligible for at least some Promise benefits. Of those eligible, between 82 and 85 percent at some point have taken advantage of their benefits. We estimate that the total value of the scholarship for a typical student over four years ranges from about $18,000 with the 65 percent benefit to about $27,000 at the 100 percent benefit. These averages include enrollment at colleges with very different prices, in the case of the 100 percent benefit, ranging from under $5,000 for four years of tuition at the least costly option (Kalamazoo’s local community college) to more than $55,000 at the most expensive (the University of Michigan).

 

Data and Methodology

Our data come from KPS administrative records. We focus our analysis on high school students for several reasons. First, it allows the analysis to include students who are ineligible for the Promise because they entered after 9th grade and can therefore serve as a comparison group for eligible students. Second, for high school students rather than younger students, the tuition benefits of the Kalamazoo Promise are closer in time. Third, high school students are more likely than younger students to believe that their achievement and behavior in school will affect their admission prospects at more-selective colleges.

We examine data on 9th- through 12th-grade students from the school years 2003–04 to 2007–08. This period includes two pre-Promise school years, the year the Promise scholarship was announced, and two post-Promise school years. The information we have on each student includes demographic characteristics, credits completed, grade-point averages, and disciplinary actions, including days of suspension and detention.

In each year, we calculate for each student what the Promise subsidy would have been had the Promise been in effect for that year and had the student continued attending KPS until graduation. We call this the student’s “virtual Promise benefit,” which in 2003–04 and 2004–05 was indeed virtual in the sense that the student was unaware of it, as the Promise was not announced until November 2005. Therefore, we would assume that any effect of this simulated Promise benefit before 2005 reflects consequences that are associated with the grade level in which the student entered KPS rather than the effect of a Promise benefit of which the student had no knowledge.

On November 10, 2005, students became aware of the potential Promise benefits that would accrue to them given their enrollment in KPS to date. It is possible that Promise benefits had some effect on student achievement and behavior starting on that date. However, it would also be reasonable to assume that it took some time for students to understand and respond to the incentives of the Promise. By November 2005, students had already made certain decisions about that academic year, such as what courses to enroll in for fall 2005. When the first full post-Promise year began in fall 2006, students may have more fully understood what the program might mean for their future.

We measure the effect of the Promise by comparing the achievement and behavior of students before and after they became aware of their benefits. For example, students who were enrolled in KPS since kindergarten found themselves eligible for a 100 percent tuition scholarship after the Promise was announced. We examine whether their achievement and behavior improved more after they learned of the scholarship than that of students who did not become eligible for a scholarship because they had enrolled in KPS much later. This comparison amounts to a natural experiment, because whether students became eligible for a scholarship was determined by enrollment decisions their families made without knowledge of the Promise.

Because of their enrollment status in KPS, the overwhelming majority of students who are eligible for any benefit are eligible for a benefit of 80 percent or more of college tuition. This suggests that it would be difficult to make fine distinctions for students in different tuition-subsidy groups. In addition, it is unclear whether the difference between a 65 percent subsidy and greater subsidies is salient for most high-school students. We therefore focus on differences in academic achievement and behavior between Promise-eligible and -ineligible students, before and after the Promise was announced.

By comparing changes in the behavior and achievement of eligible and ineligible students, our analysis is likely to understate the overall effects of the Promise because it will not capture any effects on ineligible students. For example, if eligible students improve their achievement and behavior, ineligible students may be positively affected. In addition, the Promise triggered efforts by the school district to increase overall academic standards and college focus among all students. Because of the Promise, teachers and parents may also have increased their expectations in ways that benefited all students. Our comparison of changes from before and after the Promise for eligible versus ineligible students will only capture the narrow effects of the Promise’s monetary offer. It will not fully capture the Promise’s effects on overall school climate through changes in the attitudes and actions of administrators, teachers, and parents.

We focus our analysis on four primary outcomes that reflect possible student responses to the Promise: days suspended, days in detention, whether the student completed any course credits in the school year, and GPA. Our data reveal low pre-Promise levels of behavior and achievement, leaving plenty of room for improvement. Prior to the Promise, more than 20 percent of students received an out-of-school suspension each year, and the average GPA was around 2.0 (a C average).

Results

Our results indicate that the Kalamazoo Promise increased by 9 percentage points the chances that a student earned any high school credits in 2007–08, the third year after the program’s announcement, but did not show increases in the two previous years. We also find that the Promise decreased the number of days a student was suspended by 1.3 days in the second year after the announcement and by 1.8 days in the third year. Our estimates of the Promise’s effect on GPA and days in detention are in the expected direction, with eligible students earning higher grades and spending less time in detention, but fall short of statistical significance. In general, our results are consistent with the notion that the Promise had little effect in the first year, when it was not announced until November, but some noteworthy effects in the second and third years (see Figure 2).

Previous research on educational interventions often finds different effects for students from different racial groups. The diverse nature of KPS allows us to analyze student outcomes by race. Our results for African American students are striking. The data suggest that the GPA for this subgroup increased sharply following the announcement of the Promise and continues to improve for Promise-eligible African American students. These effects are quite large; for example, GPAs for African American students increased by 0.2 points the year the program was announced, by 0.3 points the second year, and by a remarkable 0.7 points in year three. The latter effect amounts to an increase of 63 percent of a standard deviation.

One might wonder why the Promise effects on the GPAs of African American students continue to increase over time rather than showing a one-time jump. We would expect to find such a continuing increase if following the Promise there are virtuous cycles that translate into higher performance; for example, higher effort and performance in one school year could lead to still higher performance the next school year.

For African American students in KPS, we also find impacts on behavior, specifically, a decrease of two days of suspension in the first full post-Promise year (2006–07) and a three-day decrease in 2007–08. A possible explanation for the difference in results between African American students and students overall is that, on average, African American students in KPS have lower GPAs and spend more days in suspension than their white peers. The decrease in the number of days spent in suspension might have shifted past some “tipping point” beyond which more presence in the classroom leads to higher grades, while leaving the white students less affected.

Conclusion

This study uses the large change in expected college tuition costs caused by the surprise announcement of the Kalamazoo Promise’s tuition subsidies to measure the Promise’s short-term effects on student achievement and behavior. The structure of the Kalamazoo Promise benefit formula creates a natural experiment for evaluating the impact of the scholarship on Promise-eligible students. We find positive effects for credits earned and a decrease in days spent in suspension for all students, but we find significant effects on GPA only for African Americans.

It should be noted that the effects of the Promise may have increased further in subsequent years. In addition, our paper can only examine individual effects on eligible students of the Kalamazoo Promise due to the monetary incentives. Promise effects that stem from changes in the school district’s atmosphere or morale or from peer effects cannot be captured by our methodology. More research is needed to analyze these broader consequences.

Our overall results are consistent with the idea that students do not fully understand how to change their behavior to obtain better outcomes in some areas, such as course grades. If this is the case, even strong incentives for higher academic performance may not produce notable improvements in key outcomes. In our study, students may have understood that the opportunities presented by the Promise depend on displaying better behavior in school, and therefore reacted to the Promise in ways that resulted in fewer students spending time in suspension. Yet the students may simply not know how to achieve a higher GPA.

If this hypothesis is correct, our findings suggest that Promise-style policies, and other policies focused on making higher education more affordable, may be usefully supplemented by helping students better understand how their behavior affects their future. Subsidies for higher education may have a greater impact on student achievement and behavior if students understand the link between their behavior and work habits and their GPA, and the link between their GPA and the future rewards offered by programs like the Promise.

Timothy J. Bartik is senior economist and Marta Lachowska is economist at the W.E. Upjohn Institute for Employment Research. This article is based on a 2013 study published in Research in Labor Economics.

The cost of college has been rising at an unsustainable rate. The federal government has tried to soften the impact of these increases on families and students by providing more assistance in the form of loans, grants, and tax credits. For the academic year 2011–12, a total of $173 billion was spent for these purposes, according to the College Board’s Trends in Student Aid. The largest items were spending on loans ($105 billion) and grants ($49 billion), with most of the latter going to Pell Grants. Another $18 billion was devoted to tax subsidies. With the nation facing severe fiscal constraints, it’s only a matter of time before these amounts face greater congressional scrutiny. In addition to their effect on deficits and debt, we should be asking what exactly we are getting for the large sums being spent. And are there more effective ways to spend the money?

Right now a college degree is very valuable in the market. The premium that graduates earn is substantial, comparing favorably to what one could earn on most alternative investments. Not surprisingly, then, enrollment rates in higher education have risen sharply over the past few decades. That’s the good news. The bad news is that college graduation rates have hardly budged. Overall, only about 60 percent of those who enroll full time in a four-year school graduate within six years. Success rates at community colleges are even lower, and roughly half of all enrollments are in open-access community colleges (40 percent) or private for-profit institutions. Students drop out for many reasons, including the difficulty of combining college with a job or family responsibilities. Still, very high dropout rates raise questions about whether these students are ready for college. Another indicator of this lack of preparation is the large amount spent on remedial education, especially at the community-college level, where such spending is estimated at $2 billion annually. Frustration at having to spend time and money on remedial classes is also likely a factor in a student’s decision to drop out of college. Perhaps it’s time to make financial aid a little more conditional on a student’s readiness to go to college. Doing so would have a number of positive benefits, including 1) making sure that the taxpayer dollars devoted to this purpose are being spent on those most able to benefit, 2) encouraging students to work harder during high school to prepare themselves for college, and 3) increasing what students actually learn as opposed to the amount of seat time they acquire.

These potential benefits should not be exaggerated. Moreover, they have a downside: depending on how they are structured, they might screen out a large number of less-advantaged and low-performing students who deserve a chance to go to college.

Whatever one’s views on these matters, it is time to ask some fundamental questions about federal financial aid and its purposes. To get the conversation started, let me put forth an idea. Suppose we devote a larger proportion of federal aid to those who otherwise wouldn’t have an opportunity to go to college but also made it more conditional on performance. Specifically, we could shift some of the federal aid budget to performance-based grants and make the assistance a student receives more conditional on both income and performance. Families with incomes above $100,000 a year, for example, who now receive tax subsidies to offset the cost of higher education, might be asked to forgo some of this assistance. At the same time, students whose performance on the ACT, the SAT, or the new Common Core State Standards is below a certain level might be denied assistance, and those who do particularly well might get extra help.

According to 2009 National Assessment of Educational Progress (NAEP) data, only a small fraction of high school seniors are at or above proficiency in math and reading: 26 percent and 28 percent, respectively. This lack of preparation makes it difficult for them to do college-level work. For example, of younger students enrolling in college in 2003–04 with a high school grade-point average (GPA) below 2.0, only 16 percent had received a degree six years later, while 84 percent had not. The question we need to ask is whether taxpayers should foot the bill for students whose odds of success are so low.

The President’s New Proposals

President Barack Obama has already spoken about the need to tie financial aid to performance. Compared to other countries, the U.S. is in the middle of the pack in terms of both how our students perform on various tests and the proportion who graduate from college. (We are currently number 12 among 24 Organization for Economic Co-operation and Development (OECD) member countries in the proportion of 25- to 34-year-olds with college degrees.) The president’s goal is to once again make the United States number one in college graduation rates by 2020. To achieve the president’s goal will require that students who need it be provided financial aid. But it will also require that they be prepared to do college-level work.

The president, along with many other Americans, sees a college education as the way to improve social mobility. As he said in a speech in Buffalo, New York, on August 22, 2013, “higher education is still the best ticket to upward mobility in America.” Unfortunately, education in the United States has not been a mobility-enhancing enterprise. Quite the opposite. Test-score gaps between high- and low-income students have been growing, as has the gap between the college enrollment rates of children from more- and less-advantaged families. Indeed, as documented in several recent studies, a well-qualified student from a low-income family has a lower chance of going to college than a poorly qualified student from a high-income family.

Improving social mobility implies maintaining or even increasing access to higher education. But students who drop out, many of whom are burdened by large amounts of debt, are not helped by this process. Nor are taxpayer dollars being spent in as cost-effective a way as they might be. Finally, there is already some evidence that employers are discounting the value of a college degree from an open-access school. Unless that degree is valued by employers and associated with an enhanced ability to do the job, it will do little to improve productivity and earnings over time. Given these conflicting imperatives between maintaining access and improving performance, what should we do?

In his speech, the president talked about creating a new rating system for colleges that would be unveiled in 2015. It would provide students and their parents with better information on what different colleges have to offer. In addition, starting in 2018, the rating system would affect how much aid a college received from the federal government. Right now that assistance is based on enrollments and not the number who graduate, much less on what students learn. States, which have been cutting back on aid to higher education, would be encouraged to make the same kind of performance-based decisions. Students attending higher-performing colleges would receive larger Pell Grants and more-affordable student loans. Performance ratings would be based on outcomes (such as graduation rates and graduates’ earnings) as well as on access (e.g., the proportion of the student body receiving Pell Grants) and affordability (tuition net of scholarship aid). In my view, these are all good and welcome ideas from the president. He plans to propose them as part of the reauthorization of the Higher Education Act. But they are just a start.

What Does Federal Financial Aid Accomplish?

As noted above, most of the $173 billion we currently spend on federal financial aid is in the form of loans and tax credits. Under current budget-scoring rules, loans impose no costs on the taxpayer, but grants and tax subsidies do. Tax credits are available to families with an adjusted gross income (AGI) of up to $90,000 ($180,000 if married filing jointly). They do little to help those at the bottom of the income distribution, since they are not refundable. Subsidized loans are targeted similarly, with about 60 percent of the loans to dependent undergraduates going to families with an AGI under $60,000. Pell Grants are less than one-third of the total amount of federal aid, accounting for about $35 billion in spending. They are the most targeted form of assistance, focused on students with family incomes of less than about $60,000, with the bulk of the dollars going to people with family incomes of around $30,000. For the 2013–14 school year, the maximum award is $5,645. Pell Grant spending has grown rapidly in recent years but mostly because of increased enrollment and not because the grants are particularly generous. Indeed, the awards have declined in value relative to the costs of college.

Unlike Pell Grants, there is no evidence that loans and tax credits have increased enrollment. In fact, they may simply be raising tuition levels. Granted that higher education is expensive, but does it make sense in an era of fiscal constraints to be spending so much on the middle and upper-middle classes? While the political imperative to do so is completely understandable, it may not be good policy. Some funds could be moved into Pell Grants by tightening up on tax subsidies that mainly benefit the relatively well-off. The idea would be to provide more money for Pell Grants but gradually tie them more closely to academic performance. More counseling would be needed as part of any such scheme since, according to researchers Caroline Hoxby and Sarah Turner, many lower-income but academically qualified students don’t know what they are eligible for and don’t apply to selective schools (see “Expanding College Opportunities,” research, Fall 2013).

Should We Condition Aid on Performance?

Currently Pell Grants are not at all conditioned on readiness to do college-level work. All one needs is a high school diploma or to pass a General Educational Development test (GED). Contrast this system with those in many other countries, where access to college is heavily subsidized but also tied more closely to achievement. Granted there are many thorny questions about how to measure performance (SAT, ACT, proficiency on Common Core, high school GPA, etc.). I defer to those in the education field to design a sensible set of academic standards that would balance the need to motivate students to work hard in high school with the need to preserve access for those who succeed. Moreover, it would be best to test students no later than the beginning of high school so they can be counseled on what they need to achieve to be prepared by the time they apply for college. Test-score gaps open up early, so sending a message to students that readiness for college requires studying hard and mastering certain material has to be part of the program. Such a program would combine high expectations with a very concrete reward for achieving them.

The goals would be to encourage students to work harder during their high school years and to put more emphasis on achieving national benchmarks such as the Common Core State Standards. Another goal would be to make our tax dollars go further by reallocating existing funding toward those students most likely to benefit from college-level work. Any such policy should be introduced gradually so that students already far along in their school careers would not be penalized. In the early years of the program, there would be carrots (bigger Pell Grants) for those whose academic achievement in high school met certain standards and, over a period of years, a gradual denial of assistance to, say, the bottom-scoring 20 percent of applicants. However, because such a policy is likely to be controversial in a country dedicated to open access, and might have unintended effects, it would be best to test it out in a small-scale program, under a state waiver as allowed by the president’s proposed Race to the Top Fund for higher education. In the early years it would take the form of a significant bonus on top of the usual Pell Grant, paid for out of savings from current tax subsidies for higher education.

These ideas can be improved, I’m sure. My purpose in writing about them is to encourage more creative thinking and debate about how we use our limited resources to advance the twin goals of equity and excellence.

Isabel Sawhill is co-director of the Center on Children and Families and the Budgeting for National Priorities Project at the Brookings Institution.

This article is part of a forum on Federal Pell Grants. For another take, please see “Conditional Pell Dollars Miss Students Who Need Them Most,” by Sara Goldrick-Rab.

Education reform is a well-intentioned effort to improve outcomes for all students that is undercut by a misguided focus on achieving those goals on the cheap. The proposal I’ve been asked to discuss, which would condition Pell Grants on “college readiness,” is just the latest example. This idea is redundant and expensive, and will decrease the Pell’s cost-effectiveness by exacerbating an existing trend toward retargeting aid to students who are less affected by it.

Let’s start with the facts. As it stands today, the federal Pell Grant program requires students to complete secondary school; no student can receive the Pell unless she has a high school diploma. Until July 2012 there was an alternative way to obtain the Pell via a standardized “ability to benefit” test. That no longer exists. In this sense, the college-readiness standards for Pell are stronger than ever and will be even stronger if the Common Core State Standards initiative has its way with the high school diploma. Moreover, the Pell Grant also includes a satisfactory academic progress standard that ensures that students who are failing to make decent grades in college do not keep the award. Typically set at a C average (2.0), close to the mean grade-point average (GPA) of Pell recipients, that standard revokes funding from tens of thousands of students every year.

Whether or not increasing standards is a positive step in the right direction depends on one’s perspective. If you are concerned about the American dream, aiming to ensure that hard work and talent, rather than family background, determine children’s opportunities, this is a move in the wrong direction. The chances of obtaining a high school degree (and thus being eligible for the Pell Grant) remain highly unequal based on family income and wealth, and ending any second-chance alternative pathways to federal financial aid reinforces that stratification. But it’s already been done.

In today’s world, we must be concerned with what we get for our money, and Pell dollars are no exception. For this reason, it is instructive to return to the creation of the Pell Grant and its explicit purpose: “the right of every youngster, regardless of his family’s financial circumstances, to obtain a postsecondary education.” Note the focus on “obtaining” the education, not merely “accessing” it. In this critical sense, the Pell Grant is meant to support students as they move from college entry to college completion, and yet for the last 30 years, that purpose has been all but forgotten.

Moving students from low-income families from initial college entry to the completion of degrees requires that Pell Grants effectively reduce the costs of attendance so that students are able to work less and study more, and can overcome financial obstacles in their way. Its purchasing power has declined to the point that it does not do this. Moreover, research indicates that low-income students who obtain a high school diploma but are not in the top echelons of their school are the most likely to need and benefit from financial assistance: they require extra time to devote to schoolwork and yet are more likely to have unmet financial need, since so-called “merit” aid rarely flows to them.

When need-based financial aid programs like the Pell Grant are evaluated, researchers tend to find that students reformers might not deem “college ready” obtain the greatest returns from the effort. For example, Mark Schneider of the American Enterprise Institute found that in both Texas and Louisiana aid worked best for the students who faced a lot of unmet need partly because they did not qualify for aid based on their academic profiles. These students, however, are not the focus of programs in their states. Similarly, economists Bridget Long and Ben Castleman found that Florida’s Student Access Grant boosted college attainment the most for students who graduated in the top 25 percent of their high school graduating class but did not qualify for Florida’s Bright Futures merit aid program. Raising the bar on academic requirements for the Pell Grant would thus seem to reduce program effectiveness, not increase it.

One of the greatest challenges facing our nation is the increasing stronghold that family income has in determining college attainment. University of Michigan scholars Martha Bailey and Susan Dynarski found that among people born in the early 1960s, there was a 31-percentage-point income gap in the chances of bachelor’s degree attainment, with just 5 percent of those from poor families completing college compared to 36 percent of wealthy students. Over the next 20 years, the gap grew to 45 percentage points primarily because the attainment of the wealthiest Americans raced ahead (up to 54 percent completing college), while Americans from more modest means made far smaller gains (increasing their chances to just 9 percent). The consequences are a reduction in America’s labor force productivity, future increases in spending on the social safety net, and a loss of tax revenue.

Restricting the federal Pell Grant to students who are well prepared for college would make this bad situation worse. The K–12 system remains overwhelmingly unequal, and chaining Pell eligibility to it even further ensures that both ends of the educational process remain unequally distributed. It transforms the Pell Grant from a policy aimed at transforming lives to one that simply rewards students lucky enough to be born into situations where their families are able to seize good high-school educations for them.

Let’s be honest: adding such merit criteria to Pell Grants will merely make the program cheaper. We can look to history to tell us this. Consider the last merit-based Pell Grant, the Academic Competitiveness Grant (ACG). When implemented in 2006, the ACG restricted the provision of this supplemental Pell Grant to students who took rigorous high-school coursework, as assessed by a rubric that had to be manually verified and followed by financial aid officers. The lessons from the ACG are instructive: it served far fewer students than expected, added substantially to administrative costs, and was widely viewed as a failure. The program was ended in 2011 during a period of budget cuts.

Clearly, if the ACG criteria were imported into the main Pell Grant program, far fewer students would be served, and costs would fall. But so would its cost-effectiveness. In fact, improving the Pell’s cost-effectiveness requires actions in exactly the opposite direction. America needs more of its low-income families to send their children on for postsecondary education so that they can join our workforce, pay taxes, and build healthy families for future children. Focusing only on the “talented tenth” of Pell recipients will be a disaster, leaving millions of other students behind. That is a financial risk we cannot afford to take.

The data suggest that removing the Pell Grant from less-prepared students will not compel many of them to forgo college. Instead, they will enroll, and without grant aid, they will take on debt, even more than they already do. Their debt will become our debt, as they fail to repay, go into default, and become part of underground economies that do not pay taxes and exact substantial strain on our neighborhoods and communities. There is no escaping the cost of educating these students at the postsecondary level in today’s economy; there is merely the question of whether to pay now, or pay later.

If the goal is to increase the cost-effectiveness of the Pell Grant program rather than simply cheapening it, policymakers should refocus their sights on the real problem: we spend a lot on financial aid but spending alone is insufficient to make college truly affordable. The purchasing power of the Pell Grant has been devastated. When created, the Pell Grant covered nearly 90 percent of the costs of attending a public college or university, but today it covers barely 30 percent. Students from working poor families, earning an average of $16,000 a year, are asked to fork over as much as $12,000 a year—after taking grant aid into account—in order to finance attendance at a public bachelor’s degree–granting institution. Is it any wonder that some drop out? The real culprit is not a lack of academic preparation, but instead the actions of state legislatures, colleges, and universities that hike up the costs of attendance, underinvest in need-based financial grant aid, and spend the least on support services at the schools where students possess the greatest economic and academic needs.

Creating a cost-effective Pell Grant that promotes both access and completion requires adding state and institutional accountability and increasing spending on the Pell and the federal work-study program. In sharp contrast to K–12 education, where the federal government contributes at most 10 percent of revenue yet has strong accountability demands, in the postsecondary arena colleges and universities receive up to 90 percent of their support from student financial aid and yet are asked to do very little in return. We need to restore the purchasing power of the Pell Grant by bringing states and institutions to the table and driving down college costs. Financial aid will never keep pace with uncontrolled costs of attendance, and it should not have to. We must provide incentives for states to move toward providing two years of community or technical college at no cost to families. The federal government should match these commitments by expanding the federal work-study program, especially at community colleges. Raise the bar by ensuring that every Pell Grant recipient has access to a minimum of 20 hours per week of on-campus employment: they will gladly work to earn the support, and increased college contact has positive benefits for their academic progress as well. Require schools to provide all students with supportive staff to help them construct realistic schedules and financial plans, and ensure that they are screened for eligibility for all forms of financial aid and public benefits each year to support their college attendance. Finally, adjust the calculation of need so that it is possible for the expected family contribution to drop below $0 for the most severely poor students; this will allow them to accept as much financial aid (and subsidized loans) as they need to ensure their college costs are covered. Only then can we expect students to really focus, work hard, and finish their degrees. Certainly, these kinds of changes are much more difficult than simply cutting students out of the program, and they will take more time to achieve, but they will also lead to the creation of the cost-effective Pell Grant program that the nation deserves.

Sara Goldrick-Rab is an associate professor of educational policy studies and sociology at the University of Wisconsin.

This article is part of a forum on Federal Pell Grants. For another take, please see “Target Aid to Students Most Likely to Succeed,” by Isabel Sawhill.

Ensuring college access amid rising tuition costs has hovered near the top of the nation’s education agenda for years. Facing heightened concern about the student debt burden and disappointing college graduation rates, policymakers are hungry for fresh ideas. Federal Pell Grants, the primary means for subsidizing college tuition for low-income students, meanwhile, cover only a small fraction of students’ costs at most institutions. In this forum, Isabel Sawhill, co-director of the Center on Children and Families and the Budgeting for National Priorities Project at the Brookings Institution, calls for conditioning Pell Grants on both financial need and the likelihood of college completion. Offering a critical take on Sawhill’s proposal is Sara Goldrick-Rab, associate professor of educational policy studies and sociology at the University of Wisconsin.

• Isabel Sawhill: Target Aid to Students Most Likely to Succeed

• Sara Goldrick-Rab: Conditional Pell Dollars Miss Students Who Need Them Most

Scott Hamilton is the Forrest Gump of education reform, although with a lot more IQ points and fewer chocolates. He worked for Bill Bennett in the U.S. Department of Education and for Benno Schmidt at the Edison Project. He authorized charter schools in Massachusetts, co-founded the KIPP network, quadrupled the size of Teach For America (TFA), and introduced blended learning at urban Catholic schools. He’s been around.

He’s proud of what he’s done. “There are 40,000 kids in KIPP schools who will have a very different life,” Hamilton says. “Thousands of TFA teachers are having a terrific impact.”

But that’s not enough to “move the needle” on American education, he says. He wants to do more.

Now 47, he’s started a new initiative called Circumventure, based in San Francisco. Through surveys, focus groups, field tests, and interviews, Circumventure is asking fundamental questions: Do people want what schools are offering? If not, what do they want? Can technology make it happen?

Hamilton’s “aha moment” came in a conversation with a friend who worked for the Gates Foundation’s global health initiative.

“We’ve spent billions on K–12 with little to show for it,” Hamilton said. “What should we in education learn from global health people?”

“What’s the demand?” is the first question, the global health specialist said. “Do people want what you’re trying to provide?”

“Of course they do!” said Hamilton. Then he thought, “Do they?”

Bill Gates encouraged Hamilton to investigate whether low demand for schooling was causing America’s mediocre academic results. He created Circumventure.

Market Research at the Mall

Since starting Circumventure, Hamilton has become a “mall jumper.” He drives to Bay Area shopping malls, buttonholes parents, and asks what they want for their children. Then he asks the kids what they want from school.

Low-income parents want good schools, he says. But demand for what schools are offering weakens in working-class families and diminishes even more for middle-class parents. One or two steps up the economic ladder, parents say, “I just want my kids to be happy.”

Kids have their own perspective. “The ‘why’ behind education is not clear to kids,” he says. Everyone wants to go to college, but not to get an education. “College is like a bar mitzvah. It’s a rite of passage.”

Being a “good learner” is valued. Being “well educated” is not. “Students think they can learn on their own,” he’s found. “Teens think, ‘I’ll never use geometry. If I need it, I’ll learn it then.’

“Today’s kids are very entrepreneurial,” says Hamilton. “One-third of high schoolers say they’ll start their own business.”

In focus groups, he has met a Philadelphia girl who plans to open her own strip club and a boy who wants to run a saltwater aquarium store.

Kids are focused on “creating their digital brand,” Hamilton says. “How they’re seen online is what matters.” He remembers when people would talk about “discovering who I am.” Now teens say, “I’m creating my brand.” He shakes his head. What used to be exploration is now marketing.

In the past, the family sat down at the dinner table together and ate the same meal, Hamilton recalls. It was meat loaf, mashed potatoes, and peas for everyone. Now dinner is “hyper-personalized.” Everyone wants something different.

People expect education to fit their personal tastes and interests, too. “They like learning what they want to learn and expect to do it online.”

But technology doesn’t work if the motivation isn’t there. And it can’t do everything. Children learn teamwork and citizenship by interacting face to face.

“Young Millennials and their Generation Z siblings” think they don’t need school to learn new things. They’ll do it all themselves—if and when they feel like it. If kids don’t see a direct connection to their own goals, they don’t want to do the work.

“If they’re playing a computer game, they’re willing to dig for 100 treasure chests to win the Golden Sword of Awesomeness,” Hamilton says. But they’re not as willing to work hard in school because the reward isn’t clear to them.

Circumventure is pursuing ways to put learning directly into the hands of children. Digital learning can be very effective, Hamilton believes. “Students like learning privately in their own time and their own space. If you need to watch the Khan [Academy] video five times no one knows or thinks you’re dumb.”

He lent mini iPads to 18 early elementary school kids and gave them incentives for completing learning pathways in curated math apps. The children averaged eight months of growth in math in just four weeks.

Schools won’t vanish, Hamilton predicts, if only because “people want a safe place for kids in the workday and a place for kids to be with their peers.”

What if today’s kids are not as smart as they think they are?

Hamilton wonders how to persuade the future strip-club owner that she needs to master math and science.

“I don’t yet have a list of solutions,” says Hamilton. Just questions.

From Plato to KIPP

Growing up in Colorado, Hamilton attended “good but not great” public schools. His high school had 3,800 students. “If you wanted a good education, you could get it,” he recalls. If you didn’t care, you could settle for mediocrity.

Interested in politics, he postponed college to work on a senatorial campaign. The campaign manager, who had studied Latin in college, turned him on to classics. Hamilton majored in ancient Greek at Penn. He wrote his senior thesis on the education ideas in Plato’s Republic. “It’s been education ever since,” he says.

Hamilton interned for Secretary of Education Bill Bennett, and then worked as a policy analyst for the department.

He left Washington to work for former Yale president Benno Schmidt, who was starting the Edison Project. Edison, which tried to run public schools at a profit, faltered. But it was there that Hamilton met Stacey Boyd, who became his wife.

Hamilton became Massachusetts’s first associate commissioner of education for charter schools, recommending which schools would get, keep, or lose their charters.

Boyd earned an MBA and started a charter school in Boston. (Hamilton recused himself on all matters concerning his wife’s school.) The Academy of the Pacific Rim seeks to combine “the best of the East—high standards, discipline and character education—with the best of the West—a commitment to individualism, creativity and diversity.” Now a high-performing school serving grades 5 to 12, the academy enrolls primarily black and Latino students.

The family left Boston for San Francisco in 1999 when Hamilton went to work for Don and Doris Fisher’s Pisces Foundation. The Fishers, who had started the Gap clothing stores, wanted to “reduce the gap” in achievement between the haves and have-nots, says Hamilton. “They thought education was the way to do that.”

The Fishers’ goal wasn’t to start something new. They told Hamilton to “find what works and make it bigger.”

Boyd suggested Hamilton look into KIPP. She’d met KIPP co-founder Mike Feinberg after she started the Academy of the Pacific Rim in Boston. Hamilton visited KIPP’s two schools in Houston and New York City.

“It was working,” he concluded. He sold the Fishers on expanding KIPP. “Don and Doris said, ‘This looks great. Here’s $15 million. Go do it.’”

Hamilton decided KIPP would not run KIPP schools. “We wanted entrepreneurs to start schools, not administrators.”

Hamilton had seen how his wife used her business school training to build a leadership team and create a culture of learning at the Academy of the Pacific Rim. He asked University of California-Berkeley’s business school to design an intensive leadership training program for people who wanted to start their own KIPP school.

Hamilton became CEO of KIPP and the KIPP Foundation, while continuing to work at Pisces.

In his search for fundable education ideas, Hamilton kept discovering Teach For America alumni who had become innovators and leaders. “The Fishers and I agreed that TFA was addressing a human capital problem for American education. So I went to Wendy Kopp [TFA founder and board chair] and asked if she could grow the size of TFA.” With the Fishers’ help, TFA doubled the size of its teaching corps.

Hamilton served on the TFA board for several years and helped with its reorganization.

“Four years later, we went back to Wendy and asked if she could double the size of the corps again, and she said yes. But this time, before committing millions more, we got Wendy to agree to place a good number of TFA corps members at charter schools and not just district public schools.”

Blend and Save

Hamilton left Pisces, now the Doris & Donald Fisher Fund, in 2007. Then, as a distinguished visiting fellow at the Thomas B. Fordham Institute, he worked on a report exploring the question, “Who Will Save America’s Urban Catholic Schools?”

A friend said: “Why not you?” So he co-founded Seton Education Partners to develop a new financial model for urban Catholic schools, which faced rising costs and competition from tuition-free charter schools.

“Catholic schools never really recovered from losing the nuns,” says Hamilton. “It used to be one nun teaching 50 kids. Now there’s a lay teacher who makes close to a public school salary teaching 20 kids.”

By this time, Hamilton and Boyd were parents. Boyd now runs the Schoola web site, which creates fundraising tools for schools. The couple’s daughters, a kindergartner and a 3rd grader, attend private schools. “Most anyone in San Francisco who can afford it sends their kids to private school,” says Hamilton.

When their older daughter was four, she had taught herself to read using software. Hamilton was intrigued.

Seton worked with Mission Dolores Academy, an independent K–8 Catholic school in San Francisco, to introduce “blended” learning. Students split their time between mastering lessons on computers and working with the teacher. Class size rose, lowering costs, but the rotation model let teachers “work with small groups and build a closer relationship with kids,” says Hamilton.

Seton has expanded the blending learning model to three other Catholic schools.

But Hamilton decided he needed to do more. “I can change a thousand Catholic schools’ economics and results, but that won’t move the needle.

“The focus on closing the achievement gap in the last 10 to 20 years has shown results,” he says. “But I’m very worried about the future of America and our inability to improve what American kids know and can do.”

Suburban and private schools aren’t doing well compared to America’s competitors overseas, he explains. “Everyone likes to think their own kids’ school is fine. It’s those other schools that are a problem. We’ve gotten too fat and happy.”

Americans think “creativity and entrepreneurship have made our country great” and will keep our economy strong in the future, Hamilton says. His work with Circumventure confirms this: Generation Z is creative, enterprising, and very confident.

But do they know math?

Joanne Jacobs, a former San Jose Mercury News editorial writer and columnist, writes about K–12 education and community colleges at joannejacobs.com and ccspotlight.org.

State education agencies and school districts are increasingly using measures based on student test-score growth in their systems for evaluating school and teacher performance. In many cases, these systems inform high-stakes decisions such as which schools to close and which teachers to retain. Performance metrics tied directly to student test-score growth are appealing because although schools and teachers differ dramatically in their effects on student achievement, researchers have had great difficulty linking these performance differences to characteristics that are easily observed and measured.

The question of how best to measure student test-score growth for the purpose of school and teacher evaluation has fueled lively debates nationwide. This study examines three competing approaches to measuring growth in student achievement. The first approach, which is typical of systems using the popular student growth percentile (SGP) framework, eschews all controls for differences in student backgrounds and schooling environments. The second approach, typically associated with value-added models (VAM), controls for student background characteristics and under some conditions can be used to identify the causal effects of schools and teachers on student achievement. The third approach is also VAM-based, but fully levels the playing field between schools and teachers by eliminating any association between school- and teacher-level measures of test-score growth and student characteristics.

We examine the appeal of these three approaches in the context of a system for evaluating schools, although the substance of our findings also applies to evaluations of teachers and districts. We conclude that the third approach is preferable in the context of educational evaluations for several reasons: it encourages educators in all schools to work hard; it provides performance data useful for improving instruction system-wide; and it avoids exacerbating labor-market inequities between schools serving advantaged and disadvantaged students. The key distinguishing feature of our preferred approach, and the reason we advocate for its use in evaluation systems, is that it ensures that the comparisons used to measure performance are between schools and teachers that are in similar circumstances. Similarly circumstanced comparisons are well suited to address the policy goals listed above, and in an evaluation context this is a more important consideration than perfectly capturing the school’s or teacher’s true causal effect on student achievement. Simply put, comparisons among similarly circumstanced schools send more useful performance signals to educators and local decisionmakers than the alternatives.

Student Growth Measures

The three approaches we examine in this article represent the range of options that are available to policymakers. The first approach, based on aggregated student growth percentiles, has been adopted for use in evaluation systems in several states. SGPs calculate how a student’s performance on a standardized test compares to the performance of all students who received the same score in the previous year (or who have the same score history in cases with multiple years of data). For example, an SGP of 67 for a 4th-grade student would indicate that the student performed better than two-thirds of students with the same 3rd-grade score. An SGP of 25 would indicate that the student performed better than only one-quarter of students with the same 3rd-grade score.

To produce a growth measure for a district, school, or teacher, the SGPs for individual students are combined, usually by calculating the median SGP for all students in the relevant unit. The number of years of student-level data used to calculate median SGPs can vary. In our analysis, we use the median SGP of students enrolled in a given school over five years.

A key feature of the SGP approach is that it does not take into account student characteristics, such as race and poverty status, or schooling environments. Advocates of SGPs, and of “sparse” growth models more generally, view this as an advantage; they worry that methods that do take into account student or school-level demographic characteristics effectively set lower expectations for disadvantaged students. Critics of SGP-type metrics counter that not taking these differences into account may in fact penalize schools that serve disadvantaged students, which tend to have lower rates of test-score growth for reasons that may be at least partly out of their control.

A second approach, by far the most common among researchers studying school and teacher effects, is a one-step value-added model. Many versions of the value-added approach exist. The version we use takes into account student background characteristics and schooling environment factors, including students’ socioeconomic status (SES), while simultaneously calculating school-average student test-score growth. Specifically, we calculate growth for schools based on math scores while taking into account students’ prior performance in both math and communication arts; characteristics that include race, gender, free or reduced-price lunch eligibility (FRL), English-language-learner status, special education status, mobility status, and grade level; and school-wide averages of these student characteristics.

Researchers have gravitated toward the value-added approach because, under some assumptions, it provides accurate information on the causal effects of individual schools or individual teachers on student performance. But interpreting growth measures based on the one-step value-added approach in this way requires assuming that the available measures of student and school SES, and the specific methods used to adjust for differences in SES, are both adequate. If the measures are insufficient and the academic growth of disadvantaged students is lower than that of more advantaged students in ways not captured by the model, the one-step value-added approach will be biased in favor of high-SES schools at the expense of low-SES schools.

The third approach we consider is also based on value-added but is carried out in two steps instead of one in order to force comparisons between schools and teachers serving students with similar characteristics. In the first step, we measure the relationship between student achievement and student and school characteristics. In the second step, we calculate a growth measure for each school using test-score data that have been adjusted for student and school characteristics in the first step.

By design, this third approach fully adjusts student test scores for differences in student and school characteristics. In fact, it may overadjust for the role of such differences. For example, suppose that students eligible for free or reduced-price lunch attend schools that are truly inferior in quality, on average, to the schools attended by ineligible students. The average gap in school quality between these groups would be eliminated in the first step of the two-step value-added procedure, and thus would not carry over to the estimated growth measures. Consequently, it is important to interpret the results using this approach accurately, as they do not necessarily reflect differences in the causal effects of schools and teachers on student performance. We argue below, however, that this approach is still the best choice for use in an evaluation system aimed at increasing student achievement.

Comparing Results from the Three Approaches

We calculate growth measures in mathematics for 1,846 Missouri schools serving grades 4 to 8 using each of the three approaches. The data available for our study are from the Missouri Assessment Program (MAP) test results, linked over time for individual students. They include nearly 1.6 million test-score growth records for students (where a growth record consists of a linked current and prior score) covering the five-year time span from 2007 to 2011 (2006 scores are used as prior-year scores for the 2007 cohort).

For both the SGP and one-step value-added approaches, we find a clear relationship between the school growth measures and the socioeconomic status of the student body, as measured by the percentage of FRL students. Figures 1a and 1b show that schools with more FRL students tend to have lower growth measures. In the case of the SGP approach, this reflects the fact that low-SES students make less progress, on average, than high-SES students, even after conditioning on prior test performance. The one-step value-added approach corrects for SES effects to some degree but a relationship still remains.

Figure 1c shows the same data for the two-step approach. Because of how it is constructed, this approach ensures that there is essentially no relationship between the growth measures and aggregate measures of student poverty. As a result, high- and low-poverty schools are roughly evenly represented throughout the school rankings. A notable feature of the flat-line figure is that there are still considerable differences in the growth measures within any vertical slice in the graph. In other words, even when schools are compared to other schools with similar student bodies, large differences in test-score growth are clearly visible.

As we would expect, schools serving disadvantaged students are ranked higher by the two-step approach than in either of the other models. For example, high-poverty schools (those with at least 80 percent FRL students) make up just 4 percent of schools in the top one-quarter of all schools based on the SGP approach and 10 percent of top-quarter schools based on the one-step value-added approach. Using the two-step approach, however, high-poverty schools represent 15 percent of the top one-quarter (see Figure 2).

Choosing an Approach

Using the SGP and one-step value-added approaches, low-SES schools are ranked lower, on average, than high-SES schools. If this is entirely the result of bias in these two approaches, then the two-step method is an attractive alternative. But what if high-SES schools truly are more effective, on average, than low-SES schools? Even if this is the case, we argue that the two-step approach is still the most appropriate for use in evaluation systems for three reasons.

First, the two-step approach is best equipped to encourage public school employees to work hard. Research on how individuals respond to incentives shows that different signals often need to be sent to competitors in different circumstances. These signals need not be direct measures of absolute performance; instead, they should be indicators of performance relative to peers in similar circumstances. The logic is that if advantaged competitors are competing directly with disadvantaged competitors, neither group will try as hard as they would if all competitors were evenly matched. The two-step method encourages optimal effort by leveling the playing field. In contrast, the SGP and one-step value-added approaches do not result in balanced comparisons across school types and, in fact, favor the advantaged group, which runs counter to the goal of eliciting maximum effort.

Second, the two-step approach creates the kind of information that is most likely to help schools improve instruction. Measures of student achievement growth can improve instruction in K–12 schools by reinforcing positive educational practices and discouraging negative ones. For example, a positive performance signal might encourage a school to continue to pursue and augment existing instructional strategies. Alternatively, a negative signal can provide a point of departure for instructional change or outside intervention.

Information signals throughout the system can also be used to identify productive learning opportunities. Low- and high-poverty schools differ along many dimensions that likely influence what constitutes effective educational practice, including curriculum choice and implementation, instructional methods, personnel policies, and all the other day-to-day decisions that combine to create the educational environment. The two-step approach sends signals to schools about how they are doing relative to other schools in similar circumstances, rather than relative to all schools, many of which operate in quite different contexts. By doing so, this approach can help school leaders to identify those peer institutions that are performing well and are most likely to be a source of relevant lessons. Even if the two-step results conceal differences in absolute performance across schools in different contexts, they still facilitate comparisons among schools in similar contexts, which is sufficient to give schools performance signals that can be used to improve instruction.

Third, the two-step approach best avoids degrading the already-weak ability of high-poverty schools to recruit and retain teachers. It is well known that schools serving disadvantaged students are at a competitive disadvantage in the labor market (see “The Revolving Door,” research, Winter 2004). As stakes become attached to school rankings based on growth measures, systems that disproportionately identify high-poverty schools as “losers” will make positions at these schools even less desirable to prospective educators. Policymakers should proceed cautiously with implementing an evaluation system that could worsen the working conditions in challenging educational environments. An important benefit of the two-step method is that the “winners” and “losers” from the evaluation will be broadly representative of the system as a whole.

The two-step approach is preferable for each of these reasons, but a remaining concern about leveling the playing field across schools is that it will “hide” inferior performance at high-poverty schools. In our view, the best way to address this concern is to report the results from the two-step approach along with information on test-score levels. In fact, state- and district-level evaluation systems that incorporate test-score growth also typically report test-score levels and include them in schools’ overall ratings.

Reporting test-score levels will allow policymakers to clearly see absolute differences in achievement across schools, regardless of which growth measure is adopted. Reporting results from growth measures that level the playing field in conjunction with information about absolute achievement levels is desirable because it allows for the transmission of useful instructional signals. For example, a low-SES school that is performing well can be encouraged to continue to refine and improve an already-effective instructional strategy (in terms of raising test scores compared to similar schools) but still be reminded that the students are not scoring sufficiently high relative to an absolute benchmark. The latter information need not disappear in any evaluation system that includes information on achievement growth.

A related concern is that the two-step approach will lower expectations for students in high-poverty schools. However, it is important to recognize that setting expectations for individual students is not the purpose of an evaluation system. Philosophically, policymakers may not want to lower expectations for disadvantaged students. If this is the case, then the proper approach to student-level evaluation is to set fixed performance benchmarks for all students and evaluate progress toward those benchmarks. None of the three approaches to measuring student growth that we consider here is designed to achieve this objective. For example, even SGPs allow for different growth targets for different types of students by taking into account individual prior achievement. Leveling the playing field between schools is a desirable property of a student-growth measure couched within the context of an educational evaluation system. Metrics used for other purposes may need to be designed differently.

Conclusion

We examine three broad approaches to measuring student test-score growth: aggregated student growth percentiles, a one-step value-added model, and a two-step value-added model. These approaches reflect the spectrum of choices available to policymakers as they design evaluation systems for schools and teachers. All three approaches produce growth measures that are highly correlated, but the high correlations mask an important difference. Only the two-step approach levels the playing field across schools so that “winners” and “losers” are representative of the system as a whole. Although the other approaches to measuring student growth may have benefits in other contexts, when one considers the key policy objectives of evaluation systems in education, the “leveled playing field” property of the two-step approach is highly desirable.

Some states are considering using, or are already using, aggregated SGPs as part of their evaluation systems. Policymakers in these states may not have carefully considered the issues associated with applying the SGP approach, or more generally, any “sparse” growth model, in the context of an evaluation system. A likely consequence is that schools and teachers serving disadvantaged student populations will be disproportionately counted as underperforming. At a minimum, states using SGPs in their evaluation systems should consider setting up “league tables” so that performance in high-poverty schools is not compared against performance in low-poverty schools.

Mark Ehlert is research associate professor of economics at the University of Missouri, Columbia, where Cory Koedel is assistant professor of economics, Eric Parsons is research assistant professor of economics, and Michael Podgursky is professor of economics.

The full version of the working paper, “Select Growth Models for School and Teacher Evaluations,” is available here.

Entrepreneurs are having a heyday. Mark Zuckerberg of Facebook graced the cover of Time magazine as Person of the Year, Ashton Kutcher played Steve Jobs of Apple in a recent biopic, and Amazon’s Jeff Bezos even bought the venerable Washington Post. Our culture is enamored with the idea that a visionary individual can create a brand-new business that not only makes it big, but also makes a big difference in the way we live and work.

Meanwhile, few sectors are more desperate for new ideas than the $600 billion system of K–12 public education. Many intrepid entrepreneurs have waded into the waters, hoping to improve outcomes for students. And many have failed or faltered in attempting to address an institution that “alienates creative problem solvers while erecting bureaucratic barriers against those who would devise new solutions,” as Frederick Hess and Chester Finn wrote in these pages a few years back (“What Innovators Can, and Cannot, Do,” forum, Spring 2007).

Education entrepreneurs create either a for-profit or nonprofit enterprise, based on their fundraising needs, the revenue model that will suit their product or service, and the employees they hope to entice. Those who take the for-profit route face mistrust on the part of policymakers and many parents, and for-profit ventures have consequently been prevented from participating in federal grant programs like Investing in Innovation (i3) and barred from operating charter schools in some states.

Despite a surge in education entrepreneurship over the last several decades, for-profit education ventures have received far lower levels of investment than those in telecommunications, medicine, and energy. When they need new products, school and district administrators often choose to develop solutions in-house or buy from one of the big publishing companies rather than take a chance on a new, possibly untested, innovation. Not surprisingly, it has been difficult for entrepreneurs to persuade individual angel investors and venture capital firms to back their ideas with funding. Willing investors give a company money in exchange for equity (a share of ownership in the company), figuring that if the venture does well they’ll recover their investment at a premium once the company is either sold or sells its shares on the public market (known as an “exit”). But there is often little patience among investors for the slow growth required to create a high-quality education product and to develop trust among school, district, and parent customers (and earn revenues).

In the late 1990s, investments in e-learning companies and for-profit school management firms surged along with those in Internet companies. The public stock market saw 11 initial public offerings (IPOs) of education companies in 2000 alone. Most education companies founded and funded during that time went bust (although a few of those left standing eventually became success stories, as I’ll show below), leaving entrepreneurs and investors alike gun-shy about venturing into the sector for most of the last decade.

By most accounts, however, the economics of education investing are changing. Schools are now wired and have accountability incentives to invest in technology to boost student achievement, while teachers are ready to experiment with new tools. For start-ups, hardware costs have come down and software is cheaper than ever to develop. Longtime education banker Michael Moe of GSV Capital says a higher quality of entrepreneurs is entering the space. Consequently, education technology companies raised $1.1 billion in funding from venture capitalists in 2012, more than double the amount raised the prior year and nearly 10 times as much as a decade earlier. Today’s education entrepreneurs and the investors who back them believe they can avoid the mistakes of their predecessors and find their place among the 20 or so percent of companies that succeed.

In this article, I look at three entrepreneurs who have recently succeeded, with an eye toward understanding what made them successful and what that might tell us about the future of innovation in education. These companies have all exited in the last five years, bringing in hundreds of millions of dollars and earning sizable returns for their investors and their founding teams (see Table 1). What can we learn from them about what it takes to have a significant impact on K–12 education?

The Data Guys: Larry Berger and Jonathan Harber

As the disciples of innovation scholar Clayton Christensen know, market-leading companies are rarely the ones that invent the future. Although large technology companies like IBM often work with school systems, the real pioneers in the use of technology and data in education have been start-ups. The founders of Wireless Generation and SchoolNet, respectively, took into account the ways in which teachers and schools wanted to work with data (and, crucially, the ways in which school systems were able to pay for them) rather than bolting existing technologies onto the desks of teachers and administrators.

Most people trying to incorporate technology into schooling in the 1990s were focused on instruction, recalls Larry Berger, who started Wireless Generation with Greg Gunn, a graduate-school friend. But in other fields, he observes, “much of the success technology had had was in automating the workflow behind the scenes.” Berger and Gunn worked together at a web development company and also created an online presence for actor Paul Newman’s Hole in the Wall Gang camp for children with serious illnesses. After observing that new Internet-connected mobile devices might be better suited than desktop computers to the needs of teachers, they began in 1998 to pursue their longtime dream of starting an education company together. On nights and weekends, they worked on the business plan, eventually raising $17 million from individual investors and Seavest Capital, as well as from the W. K. Kellogg Foundation. Their very first check was from then Qualcomm CEO Irwin Jacobs, who a few days after a brief pitch meeting with Berger sent a check for $250,000 along with a Post-it note: “Let me know what percent of your company I bought.”

Berger says he and Gunn planned to develop mobile-friendly applications that would harness the power of the Internet to change the way teachers worked in the classroom. “We liked the idea of creating tools that would make teachers feel more empowered rather than cost them a lot of time and somehow feel less empowered,” he says. Initially, the founders set out to distribute their applications through existing channels, such as publishers. Their first product, mCLASS, automated the burdensome paper-based “running record” that teachers use to capture students’ early-learning progress. When demand for the products emerged faster than did satisfactory deals with publishers, Wireless Generation shifted its strategy and built a sales team.

They quickly found that in order to succeed in public education, they would need to be attentive to public policies that would drive demand and funding for their products. Wireless Generation employees pored over the No Child Left Behind Act (NCLB) the night it was signed, looking for potential opportunities, and found that the act’s Reading First component “created an unusual amount of liquidity centralized at the state level (about $200 million per year) that did not already have a bureaucracy trained to spend it,” as Berger has written. Wireless Generation went on to secure at least 18 state contracts, including those of New Mexico and Ohio.

By 2010, the company had grown to 350 employees and an estimated $60 million in annual revenue, including a sizable contract with the New York City Department of Education to build its Achievement Reporting and Innovation System (ARIS) as an IBM subcontractor. “What impressed us with Larry was his largeness of vision,” says Tony Berkley, who spearheaded a $5 million convertible loan from the W. K. Kellogg Foundation’s Mission-Driven Investments portfolio in mid-2010. “He had built a company that was trying to tackle teacher effectiveness and accountability and innovation in the curriculum market—pretty much all the big challenges in education.”

That year, several publishers as well as media conglomerate News Corporation approached Wireless Generation with acquisition offers. The highest bidder was News Corporation, with a $360 million deal that allowed Berger, COO Josh Reibel, and chief product officer Laurence Holt to collectively maintain a 10 percent stake. (Gunn left the company prior to the acquisition for an entrepreneur-in-residence position at double-bottom-line venture firm City Light Capital.) “We figured [the purchase] would let us do bigger and better and more exciting things,” says Berger, noting that News Corporation’s offer was the only one that kept the company and its technologies intact rather than integrating them into existing products.

The reaction in the education entrepreneurship community was pleasant surprise, though some were skeptical about News Corporation’s entry into the education sector, including the state of New York, which pulled a planned $27 million state data-system contract away from Wireless Generation in 2011. Today, Wireless Generation’s technology underpins News Corporation’s Amplify division, which is run by former New York City schools chancellor Joel Klein (who joined the company as CEO of the division a few weeks before the acquisition, though the deal was already underway at the time). Amplify delivers not just software but also customized tablet computers to schools, along with Common Core–aligned content developed by the Core Knowledge Foundation. Amplify’s large tablet rollout, in Guilford County, North Carolina, hit the skids in late 2013: the district planned to spend half of its $30 million Race to the Top grant to buy 20,000 of Amplify’s tablet computers, but suspended the effort in October due to persistent hardware malfunctions.

Berger is now president of Amplify Learning, the piece of the business focused on curriculum, which may suit his taste for digging into teachers’ needs. “I found Larry Berger to be a very quick study in grasping the cognitive science behind our work,” says E. D. Hirsch, founder of the Core Knowledge Foundation. “His flexibility and imagination were a breath of fresh air for me.”

Berkley of the Kellogg Foundation, which has already recovered its original loan at a premium, believes Wireless Generation’s best days are ahead, as its team acts as an “innovation engine” for News Corporation. “Their first innovations aren’t the story; it’s the ability of the team to keep innovating over time,” he says.

Akin to Berger and Gunn, SchoolNet’s Jonathan Harber recalls that he has also been fascinated by technology and education, particularly “how computers can both mimic how people learn and think, and how can they assist in peoples’ learning and thinking.” After a summer job in high school researching the synapses of giant sea slugs, Harber studied (human) cognitive science at Wesleyan University and at the Massachusetts Institute of Technology before starting several educational software and video-game companies.

But when Harber began to investigate applying technology to the needs of low-income schools and districts, the first consultant he hired had a significantly different background: Denis Doyle was a former federal government official and think tank scholar who had co-written books with business leaders about the need for education reform (including Reinventing Education, with IBM CEO Lou Gerstner). As he and Doyle worked—like Berger and Gunn, outside of their day jobs—to develop the plan for SchoolNet, they met with superintendents to learn about their needs.

“I was amazed to hear [superintendents] talk about the achievement gap,” recalls Harber. “Some would say, ‘I have no idea how many students I have, let alone what they’re achieving,’ and others would say, ‘achievement isn’t my job.’ They had a ton of data, but it was designed for compliance purposes, not achievement.”

SchoolNet opened for business officially when it responded to and won a request for proposals from the tiny school district of Beaufort, South Carolina, to create “an enterprise resource planning (ERP) system for curriculum and academics.” SchoolNet’s first funding came soon after, from friends and angel investors; over time, the company raised $30 million from investors, including Seavest Capital, which also backed Wireless Generation. In its early days, Harber found his fledgling company competing against large business-software providers that built systems to hold basic student and school data, and against publishing and assessment companies that delivered year-end high-stakes tests. The “gaping hole” Harber saw was for systems that could capture, analyze, and report formative data quickly to allow teachers and principals to make instructional changes accordingly, a gap that grew even wider when NCLB began to shine a spotlight on the dismal progress of student subgroups and put pressure on schools to improve their performance.

SchoolNet’s data platform soon evolved to include formative assessments (“the quality of the data is only as good as the information you’re gathering,” Harber explains), professional development information, and frameworks for teacher observation and evaluation. In 2009 and 2010, states competing for Race to the Top grants began looking for “instructional improvement systems” that would earn them points against the grant program’s criteria for providing teachers, principals, and administrators “with the information and resources they need to inform and improve their instructional practices, decision-making, and overall effectiveness.” (Interestingly, two states that didn’t win Race to the Top grants signed on with SchoolNet first: Idaho and Kentucky.) By 2011, SchoolNet was earning an estimated $75 million in annual revenue, and its products contained data on 5 million of the nation’s 50 million public-school students.

“Jonathan was very good at developing creative means to reach the market, including finding ways to work with both school districts and educational foundations supportive of SchoolNet’s mission and products,” says Brian Hayhurst of the Carlyle Group, which led a $13 million Series D round of funding in 2009.

The growing appetite among school districts and states for data systems like SchoolNet’s caught the attention of education publishers. In April 2011, just a few months after News Corporation bought Wireless Generation, Pearson scooped up SchoolNet’s software, contracts, and team for about $230 million. Harber has become Pearson’s head of K–12 technology, where he says he is creating a “unified technology stack” out of Pearson’s inventions and acquisitions, which also include student information system PowerSchool; together, SchoolNet and PowerSchool now reach more than 20 million students.

Like Amplify, Pearson has run into challenges in extending its newer technologies into schools. A sizable contract to provide Pearson-powered iPads to the Los Angeles Unified School District came under fire by teachers and school board members concerned that the curriculum was released into schools before it had been fully developed. Harber is undeterred and believes that the data platform he began creating at SchoolNet will help Pearson “lay the tracks” for the future of learning. “When all kids have computers, and learning moves from all print-based or all teacher-delivered to being more digital and blended, data will be a key component to learning at scale,” he says.

The School Guy: Ron Packard

For-profit schools are the red-headed stepchildren of the education entrepreneurship movement, more objectionable to many than other for-profit ventures in education. It’s one thing if a profit-seeking entrepreneur wants to offer food, computers, transportation, or gym equipment, but woe to the entrepreneur who considers operating schools at a profit.

“Private involvement in public schools pushes people’s buttons. For whatever reason, it’s a sensitive topic and arouses strong feelings,” says Steven Wilson, CEO of nonprofit charter management organization Ascend Learning and former CEO of for-profit education management company Advantage Schools.

It irks some stakeholder groups that the people starting for-profit schools are rarely career educators and come into education from fields like media and banking.

For example, a Vanity Fair profile of Edison Schools founder Chris Whittle wondered, “Is Chris Whittle the Devil?” after he used the proceeds from a series of successful media businesses to create Channel One Communications, which placed free televisions in America’s classrooms that carried not only educational news programming but also advertising. So the ground was laid for mistrust when Whittle forged into for-profit schooling. First imagined as a conglomerate of 1,000 private schools known as the Edison Project, Whittle changed the name to Edison Schools and the business model to managing existing schools under contracts with school districts and operating charter schools. Despite fierce opposition by teachers unions in many communities, the company grew quickly, raised more than $200 million in private equity, and went public in 1999. But after a series of setbacks, Whittle and outside investors bought back the company in 2003 for $1.76 a share. Now called EdisonLearning, the company offers tutoring and assessments in addition to managing schools, although Whittle himself departed in 2007 to start Avenues: the World School, a private international school.

Just as Edison’s star began to fall, in 2000, former investment banker Ron Packard founded virtual school operator K12. Packard left banking and consulting in 1997 to join junk-bond king Michael Milken’s education conglomerate Knowledge Universe, attracted by the fact that “education was at a point where technology and private companies would be able to have a positive impact on affordability and effectiveness.” With a strong focus on revenue growth in emerging education markets, he led investments in for-profit education managers like Charter Schools USA and LearnNow (later sold to Edison) and ran the firm’s daycare, preschool, and afterschool businesses.

During that time, he was in search of a complete online math course to supplement his daughter’s learning and was disappointed by the lack of offerings. “I started to think we should build them, and in fact build an entire school online,” he says. Knowledge Universe kicked in $10 million, former secretary of education Bill Bennett agreed to be chairman, and Packard became CEO. “I’ve always believed the most important thing is delivering something that customers want, and there were always people out there who wanted a more individualized approach,” Packard says. “We had a version that was as good or better and could offer it at a lower per-pupil funding level.” K12 raised $90 million in private funding between 2000 and 2007 from Knowledge Universe and individuals, including Andrew Tisch of Loews Corporation and Larry Ellison of Oracle. The funding enabled the company to invest heavily in what many view as a strong curriculum, led by former Core Knowledge curriculum developer John Holdren, but also to spend millions on lobbying efforts to put in place policies and to secure the necessary approvals to educate students virtually.

Growth has been a necessity for K12. Although the costs of operating an online education business are somewhat lower than what Edison faced, K12 must spread the benefits of its design and implementation costs over more schools and students in order to recover its substantial upfront development costs. “Implementing this vision is expensive—or will be until the productivity gains of a fully integrated online education system are realized,” Packard emphasizes in his book Education Transformation. “When a course can be used by fifty to one hundred million students, however, it becomes affordable. That’s why scale is so necessary.” Unlike Edison, K12 turned a profit before it went public, with revenues skyrocketing from $141 million in 2007 to $848 million last year, drawn mostly from its management contracts with states and districts to operate virtual and blended-learning schools, but also from operation of three private online schools paid for by parents and from direct sales of its courses to schools and districts. Packard himself has also earned a healthy chunk of change, with annual compensation of $670,000 and millions more in stock options.

Does the company deliver for students? It’s hard to say for sure. For years, K12 has reported student progress using the results of a Scantron assessment students take at home with no supervision, saying it wants to use a more reliable measure than state tests. “The 33 states in which we manage public schools each measure academic performance using different methods,” noted K12’s 2013 annual report. “Academic success defined by using grade-level, static proficiency tests is even more problematic given high enrollment growth rates, high student mobility and a high percentage of students who enter behind grade level.” But K12 students’ academic performance on state assessments has begun to raise concern. Shortly after K12 went public, Packard acknowledges, the company’s student demographics began to shift from average students in search of a different school experience (such as home schoolers, rural students, or frequent travelers like athletes and actors) to include more struggling students performing below grade level, for whom online education is a last resort.

“They’re serving students that the product was not originally built to serve,” says Michael Horn of the Clayton Christensen Institute for Disruptive Innovation. “I think K12, in its effort to grow, has fallen into that stereotypical trap of trying to be a one-size-fits-all intervention.”

Packard maintains that K12 made significant adjustments to the instructional model to address a wider diversity of students and that value-added scores have risen in some schools. Some are still skeptical, including a group of investors that filed a class action lawsuit in 2012 alleging that K12 had misled them about student retention. (The lawsuit was settled for $6.75 million in early 2013.)

“K12’s aggressive student recruitment has led to dismal academic results by students and sky-high dropout rates,” activist investor Whitney Tilson noted in a presentation to other investors earlier this year, explaining why the stock is the largest “short” position (an option held with the expectation that the stock price will go down) in his portfolio, despite its strong performance to date.

“After the IPO, I got discouraged because the company’s priority seemed to shift from academics to growth—it wasn’t so much about academic achievement but on delivering the promised enrollment numbers to shareholders,” former principal of Ohio Virtual Academy (operated by K12) Jeff Shaw told Tilson.

Packard says K12 has no interest in recruiting students who don’t stay, nor can he turn away those who might not be a fit, since public charter schools are not legally permitted to be selective. He also appears unconcerned about the mounting criticism. “I’m not sure a lot of people could take the levels of attacks that some of us have gone through, but it energizes me, it means we’re successful,” says Packard. “It’s our moral obligation to make sure that anyone who wants [online education] can access it.”

Making good on that obligation going forward will fall to board chair Nate Davis. In January 2014, Packard turned over the CEO reins to Davis so Packard could run a new e-learning company that K12 is launching with a syndicate of global investors led by Safanad Limited.

The Takeaway

Despite sizable stumbling blocks and the increased public scrutiny that have come with scale and success, each of these entrepreneurs and their companies have had an undeniable impact on the daily operations of millions of schools across the country. They have not only dreamed big, but patiently built their teams and businesses over a long period of time, adjusted course regularly, skated toward favorable policies (and the resulting available capital), and experimented with alternatives to the traditional district sales process that has frustrated so many others. Along the way, they’ve taken the heat that comes with pioneering new ways of working and educating (much as Microsoft and Facebook have done over the years). A few lessons can be gleaned from their experiences.

First, start with (at least) one geek. Both SchoolNet and Wireless Generation were started by technologists intrigued with the needs of education. Wireless Generation’s Berger and Gunn compared their early ideas for what handheld computers could do in education to the Star Trek tricorder that could scan geological, meteorological, and biological data. K12’s Ron Packard spent some time in college writing an image-processing language but also loves financial modeling.

“Successful entrepreneurs have a fascination for a particular kind of intellectual problem and a relentless, unstoppable, endlessly inventive, and improvisational effort to solve that problem,” argues Matt Greenfield of Rethink Education, who backed Wireless Generation as an angel investor. “They do not start with a vague desire to be an entrepreneur and a quest to figure out something someone will pay for.”

Build a utility team. These successful entrepreneurs surrounded themselves with teams that straddle a range of disciplines, from curriculum and assessment, to policy and evaluation, to product development and sales. SchoolNet’s Harber brought in Denis Doyle and his policy and business connections, much as K12 earned instant connections and credibility with Bill Bennett as its first chairman, along with curriculum developer Holdren’s education experience at the Core Knowledge Foundation.

“A key determinant of success is the ability to hire, retain, and excite great people,” says Greenfield. Education banker Moe says the prospects for such “hybrid” teams are only improving as the entrepreneurial education ecosystem matures, with new start-ups able to hire managers who have ascended the ranks of Teach For America or KIPP, received leadership and management grooming from the Broad Center or Education Pioneers, or run even larger operations at Blackboard or Kaplan.

Get rich slowly. The entrepreneurs profiled above spent at least seven years with the original business rather than jumping ship. That shows commitment. “There are lots of easier ways to make money,” acknowledges Packard. “You have to love what you’re doing, because education is not the fastest-moving or simplest market,” agrees Berger.

It’s critical to raise money from people who are also willing to get rich slowly, whether that’s a bunch of deep-pocketed individuals like those that Larry Berger attracted to Wireless Generation or a cash-rich parent company like Knowledge Universe, which backed K12.

Pull an end run around schools themselves. Technologies with the greatest wide-scale success tend to sell to individuals rather than businesses (or, heaven forbid, public agencies). While many of K12’s revenues come from states and districts today, the company’s scale came through addressing the demand among home schoolers for a strong curriculum. Even Wireless Generation and SchoolNet credit some of their success to pursuing bigger fish than schools and districts, whether that’s the state-level contracts that SchoolNet secured via Race to the Top or the development contract that the Bill & Melinda Gates Foundation awarded Wireless Generation (now Amplify) for work on its inBloom student-data framework.

Push (and pay) for policy change. Every successful education company has managed to tap a significant new revenue stream made possible by public policy. In the case of K12, which relied more than others on state-level policies, that translated into significant spending on lobbying and government relations, although Wireless Generation and SchoolNet executives also spent time testifying in front of Congress in support of various policies.

Finally, fly low—or be ready for bullets. There are a variety of successful entrepreneurial personalities out there. Several observers characterize Berger and Harber as easygoing and cooperative, which has suited them in their behind-the-scenes work building data platforms that did not thrust them into the public eye. Eventually, though, both systems reached significant enough scale to raise the hackles of privacy advocates (as with the inBloom project that Amplify is working on) and other local activists (like those who questioned whether cities should spend millions on SchoolNet systems amid budget struggles). Meanwhile, some describe Packard as more confrontational, perhaps a crucial attribute for a for-profit school operator. “K12 has been willing to take a lot of bullets on choice and virtual and digital…but that also allowed them to grow fast,” says Horn.

Julie Petersen is a writer, editor, and communications strategist based in Petaluma, California.

Today’s education technology holds immense promise, but what matters more than the tools themselves are how they are used in schools and in classrooms. In Breakthrough Leadership in the Digital Age, Frederick M. Hess and Bror Saxberg argue that educators have tended to think of adopting technology as a way to “reform” or “fix” schools. The would-be reformers have poured tablets and online learning software into classrooms, presuming that magic would eventually happen. But schools are complex and hard to move, while these efforts have been correspondingly unfocused. The more promising way forward involves tapping learning science to determine where the familiar schoolhouse falls short on providing deliberate practice, timely and copious feedback, and extensive opportunities to build mastery—and how new tools can help us do better. Following are three excerpts from the book that convey the core of their argument.

Why educators should start thinking like learning engineers.

Engineers are the world’s most creative and effective problem solvers. That’s because they combine imaginative thinking with an appreciation for how the world really works. Engineers in any field operate by identifying problems to be solved, designing smart solutions consistent with relevant science, and figuring out how to make those solutions feasible. It’s our contention that education has suffered for its dearth of engineers. In fact, we think, this is why so many of our debates seem to go nowhere. Engineers can bring fresh workable approaches to stubborn problems. As Donald Coduto, civil engineer and professor at California State Polytechnic University, Pomona, once wrote, “Some say the cup is half empty, while others say it is half full. However, in my opinion both are wrong. The real problem is the cup is too big. Sometimes all we need is a new perspective on an old problem.”

When it comes to realizing the promise of digital technology, educators need to start approaching classroom challenges as learning engineers. While such a label may sound unfamiliar at first, stick with us for a moment. The fact is that learning engineering is what tech-savvy education leaders—and more than a few who aren’t so tech-savvy—already do every day (whether they know it or not). These educators ask what problems need to be solved for students, turn to research to identify solutions, and devise smarter, better ways to promote terrific teaching and learning. What is education technology’s role in all of this? Learning engineers see this technology as a tool, not a solution.

In fact, just about everything in a school, classroom, or learning environment should be regarded this way. And we mean everything. For example, a team at Kaplan, Inc., an education provider that serves more than a million students a year, was frustrated that classroom furniture was compromising teaching and learning. While classroom furniture isn’t usually regarded as an education technology, it should be. Tables and chairs are tools for learning, just like books, pencils, whiteboards, and laptops. In fact, a 2013 study from England’s University of Salford found that classroom architecture and design factors like classroom orientation, natural light, noise, and use of space had a measurable impact on student learning.

Yet traditional classroom desks and chairs often make it tough for students to collaborate or rotate through activities without wasting time and disrupting instructional flow. In response, Kaplan started piloting new furniture designs from Steelcase, an office furniture company. The new chairs and tables can be easily reconfigured from a lecture-style class setting, to paired or group arrangements, and back again. Kaplan is now collecting data to see how the new configurations are actually working and what modifications are needed. This is learning engineering, applied to furniture.

We’d be surprised if Ford, General Electric, or Apple adopted new technological products without carefully scrutinizing what works, what doesn’t, what’s annoying, and why this is so. Think about how rapidly Internet firms like Amazon or Facebook modify their offerings in response to complaints and feedback. Indeed, designers take care to consider what science and research can teach them about aerodynamic efficiency, convenience, and the rest—and then they try it out, see what happens, try to rapidly fix mistakes, and keep going.

How developments in learning science inform the role of learning engineer.

What we’re talking about is not revolutionary. It’s really nothing more than educators using learning science and technology to solve practical challenges. In a nutshell, that’s what engineers do.

Sure, the familiar caricature of engineers is that they love differential equations and aren’t much fun at parties. The reality is different (well, we make no claims about the parties). Engineers solve real-world problems. Lots of them. Repeatedly and creatively. They do this by taking scientific knowledge, applying it to the problem (whether that involves the strength of steel or the speed of computers), and designing the smartest solution they can find within financial and practical constraints. The best engineers seek optimal solutions that are easy to use well, that serve user needs, and that directly and practically solve the problem at hand. Think of those folks at Apple who cooked up the iPod, the iPhone, and the iPad—imagine if they were immersed in learning instead of semiconductor device science and consumer device usability. They’re what we’ve got in mind.

Unlike scientists, engineers have to account for real-world complications. Ken Koedinger, codirector of the Pittsburgh Science of Learning Center, says of the difference between a scientist and an engineer, “Let me use physics and mechanical engineering as an analogy. Physics comes up with principles. But when engineers have to employ those principles, they have to deal with [practical questions] like the fact that the coefficient of friction depends on the surface that you’re using. Engineering is a different game. Good learning design is not just about principles regarding how learning works, but about learning to apply those principles in a particular context.”

Here’s an easier example: Think of physicians as the engineers of biological science. Biologists explore how cells work, organs function, and molecules interact with each other and with bugs, but it’s physicians who apply this knowledge to solve health problems for real patients. That winds up including asking questions like, “How do I keep this medicine refrigerated in the Serengeti?” Physicians know that what works in a lab is still a long way from working “out there” for a patient—and they pilot, train, and monitor accordingly.

The same is true for learning science and learning engineering. Learning science, for example, teaches that learning has to work through our limited, verbal, conscious minds (“working memory”) before it becomes integrated into our rapid, parallel-processing, nonverbal long-term memory. Moreover, we’ve learned that this is only possible through lots of deliberate practice with targeted feedback. Unfortunately, this science does not reveal how to organize the work of teachers and students to actually do this, so, like any engineer, we still have to apply judgment to the application of the science.

Many factors can contribute to learning difficulties, like awful curricula, mediocre instruction, a lack of timely assessment and student feedback, a lack of discipline, or a student’s inability to find time outside school to do what’s asked. In many of these cases, technology may have a restorative role to play. But technology isn’t always helpful. Indeed, sometimes it can be a distraction. Some readers may recall the early era of desktop publishing, when the earliest Macintosh computers allowed users to select fonts as they liked. Some of us (no names!) created documents festooned with fonts of different kinds, drunk on the opportunity to shift with every paragraph, every page, and every topic. Needless to say, that did not help with the clarity or accessibility of the writing.

Similarly, while we can now view a page from a biochemistry textbook on our iPhone, or even write a term paper on one, this does not mean it’s wise to do so. The question is not what can be done with technology, but what technology can do to promote learning. When a principal bragged to us that a student had written a term paper on his smartphone, we could only wonder whether the principal would have bragged that the student wrote a term paper without using his smartphone. If not (and we presume the answer is “no”), then the principal was fixated on the technology rather than the learning. And that’s a recipe for disappointment.

Case study of the Carpe Diem Collegiate High School and Middle School in Yuma, Arizona, where the idea of thinking like a learning engineer has been put into practice.

Founded in 2002 by Rick Ogston, former president of the Arizona Charter School Association, Carpe Diem explored a “blended” learning model, combining computer-assisted instruction with redesigned classrooms. Carpe Diem offers a radical twist on the familiar schoolhouse: more than 200 personal cubicles, each containing a computer, fill the single room that makes up one floor of the school. Hechinger Report journalist Nick Pandolfo has noted that it “looks more like an office or call center than a school.” Students split their time between individualized online instruction on their personal computers and teacher-led collaborative workshops.

At the heart of the redesign was a desire to help students learn better. As Ogston explains, “Now, the chalkboard model may seem more ‘efficient,’ since you have a lot of people doing the same thing at the same time, but it’s not very good for learning. I was looking to see how we could provide more individualized and personalized instruction for kids. And, at this point in time, the best way to do that is to leverage technology. Now if you can find a better way, great.”

The key, Ogston says, is to understand technology as a tool rather than some kind of secret sauce. He says, “When you’re leveraging technology like we are, people want to look at us in terms of technology. But the secret sauce is not the technology, it’s the relationships. The downstairs is the collaborative social learning environment, and the upstairs looks like a call center. People say this looks impersonal. I say, ‘Well, as opposed to a classroom with rows of desks and a teacher lecturing?’ You know, that doesn’t strike me as real personal, either. The upstairs offers space that’s separate and non-social so that students can focus on their own path and the downstairs is a place for collaboration. That’s the blend that we find works.”

The Carpe Diem design required “the entire ecosystem to change, including the role of teacher, parent, student, and administrator,” Ogston notes. The school has five teachers and four teacher aides for 226 students with each teacher teaching all the students in their subject, regardless of grade level.

So, for instance, Carpe Diem has one math teacher for all students in grades 6–12. It’s not about economy, Ogston explains, but about cultivating relationships that can span across grade levels and reducing the learning curve required for teachers and students to start from scratch each year. Ogston says that teachers get to know their students, their career goals, their family, and their learning needs.

Ogston came to education not as a career educator but after earlier stints in the Marines, business, as a family therapist, and as a church pastor. When he entered schooling, he eventually earned a Master’s in Education and found himself running two charter schools. In that role, Ogston recalls, “Just like I had feared, I realized one day that we had re-created the traditional system.” Ogston’s eclectic background may have helped equip him to think differently about technology and school design. Nearly a decade ago, he sat down with a consultant and a computer to explore what was possible. He remembers, “I didn’t know the terms and all the pedagogy…. I ended up creating the model just by asking questions. Then I put that on the shelf. A year and a half later, the building we were leasing was being sold out from under us and we needed a new home. With this new model, I didn’t need all these individual classrooms, I just needed one large space—a call center would be fine. We found a space and did a conversion in three months.”

Carpe Diem has delivered some promising results, while serving a student population that was 46 percent low-income in 2011–12: Carpe Diem ranks among Arizona’s 10 highest-performing charter schools, outperforming Arizona’s statewide four-year graduation rate five of the six years between 2007 and 2012 (with a 96 percent graduation rate in 2011), and regularly exceeding the Arizona average at every grade level on the statewide assessment.

The Carpe Diem model is also cost-effective. It requires fewer teachers per student than a traditional school, so Carpe Diem has achieved those results with only about $5,300 of the $6,300 per pupil allocation, according to Ryan Hackman, the school’s chief operating officer. Most of the balance goes toward paying off the bond on the $2.6 million facility or to procuring technology (the school sets aside 2.5 percent of its budget, or about $50,000 a year, for this purpose) [see N. Pandolfo, “Education Nation,” The Hechinger Report, 2012]. Ogston says of the model, “One of the wonderful benefits of what we do is that it becomes more economical, even though it wasn’t designed with that in mind. Now, teacher unions are making accusations that computers will replace the teachers. My usual retort is, ‘If a computer can replace you, then it probably should.’ They tend to just gasp and stare at me. And I go on to explain that a really good teacher could never be replaced by a computer. And so, if you think you could be, then, maybe you should evaluate what kind of a teacher you are.”

Ogston notes there are plenty of stumbling blocks and pitfalls for learning engineers. He wryly acknowledges, “I’ve been stupid. When I started this, I didn’t know anything. And I didn’t know what I didn’t know.” And, he says, “I think we’re still like 10 years behind where we should be. But technology and financial limitations tend to impinge upon my happy exploration.”

For others pursuing tech-enabled redesign, he advises: “The most important thing is the vision of what it is you’re going to do. Do your research on technology. Are you simply trying to overlay the technology over the existing system? Or is it a transformational initiative to truly personalize education? Once you’ve got a vision, there are various kinds of support that are needed in terms of curriculum and infrastructure. Trying to backfill technology into existing systems can be difficult.”

Radical? We suppose. But what we find most interesting is that in using technology to rethink the schoolhouse, Carpe Diem has focused on the latter, not the former.

Frederick M. Hess is director of education policy studies at American Enterprise Institute and executive editor at Education Next. Bror Saxberg is chief learning officer at Kaplan, Inc.