There is a substantial body of research showing that teacher quality is an important determinant of student achievement in elementary and secondary schools, inspiring some states and districts to enact policies aimed at identifying and rewarding high-quality teachers. Yet relatively little is known about the impact of instructor effectiveness on student performance in higher education, where such insights could be particularly useful. Even more than leaders at K–12 schools, college administrators often have substantial discretion to determine which instructors receive teaching assignments.

This lack of research is largely the result of data and methodological challenges. Whereas K–12 schools administer standardized tests to most students in core academic subjects, there are few such common assessments at colleges—even among students taking the same course at the same campus. In addition, college students are able to choose their classes and thus, their instructors. Because college students have a great deal of flexibility compared to students in K–12, simply comparing their success rates across instructors is likely to be misleading.

In this study, we overcome these challenges by examining data on more than 2,000 algebra instructors at the University of Phoenix (UPX), a for-profit institution that is the largest university in the United States. UPX follows a unique instructional model based on common, standardized curricula and assessments, in both online and face-to-face classes. These assessments provide an objective outcome by which to measure instructor effectiveness, and we use them to examine two questions: How much does student performance vary across instructors? And is instructors’ effectiveness correlated with their teaching experience and salary?

We find substantial variation in student performance across instructors, both in the instructor’s class and in a subsequent class. Differences are substantial in both online and in-person courses, though they are larger for in-person classes. Notably, instructor effects on students’ future course performance are not significantly correlated with student end-of-course evaluations, the primary metric through which instructor effectiveness is currently judged. Our findings suggest that colleges could improve student outcomes by paying more attention to who is teaching their classes.

Examining a Higher-Ed Hybrid

We study instructor effectiveness at UPX in a required undergraduate mathematics course for BA-seeking students, College Mathematics I (Math I, known internally at UPX as MTH/208). As with most courses at UPX, Math I classes are five weeks long and taken one at a time. UPX students take Math I after completing about eight other classes, so enrollment signifies some level of commitment to persisting in their program. The course covers basic algebra, such as linear equations, graphing, and working with exponents, and is the prerequisite for a follow-on course, Math II, which covers quadratic equations and factoring polynomial expressions. Many students struggle in these core math courses, which are regarded by UPX staff as important obstacles to obtaining a degree.

Students can take Math I online or in person; UPX currently has campuses in 30 U.S. states, as well as Washington, D.C. and Puerto Rico. Math I course sections are split about evenly between the two modes. In the face-to-face sections, students attend four hours of standard in-class lectures per week, typically on a single evening. In addition, they are required to work with their peers roughly four hours per week on what are known as “learning team” modules, as well as spend 16 additional hours outside of class reading course materials, working on assignments, and studying for exams.

Online courses are asynchronous, so students can access course materials and complete their assignments at any time. Instructors provide guidance and feedback through online discussion forums, in which students are required to post substantive comments and questions six to eight times each week. There is no synchronous or face-to-face interaction with faculty in the traditional sense, but instructors engage with students via online discussion, including redirecting students to relevant materials when necessary.

There are differences between the online and in-person courses in terms of curriculum and grading flexibility. Both courses have standardized course curricula, assignments, and tests, and grading is performed automatically through the course software. However, while online instructors mainly use these elements as provided, in-person instructors are more likely to support students with their own learning tools, administer extra exams and homework, or add other components that are not part of the standard curriculum.

Both types of instructors experience similar hiring, training, and management practices, which are controlled by a central hiring committee at the Phoenix campus with input in some cases from staff at local campuses. Qualified candidates must pass a five-week standardized training course, including a mock lecture for in-person instructors and a mock online session for online instructors. An evaluator sits in on their first class or follows an instructor’s initial online course to ensure they meet university standards. Salaries are relatively fixed, but do vary modestly with respect to degree and tenure.

For online classes, the allocation of instructors is essentially random. UPX starts about 60 sections of Math I each week, and the instructor’s name is made available to students only two to three days before the course starts, at which point students are already typically enrolled. The only way for students to sidestep these teacher assignments is to drop the course and enroll in a different week.

For in-person sections, the assignment works differently. Most UPX campuses are too small to run multiple sections of the class at the same time or even one right after the other; instead, students may need to wait for a few months if they decide to take the next Math I section at that campus. This limits students’ ability to shop around for a better teacher. However, the assignment of students to in-person sections is likely to be less random than for online classes.

UPX tracks and evaluates its instructors’ performance through annual observations by an evaluator, ongoing review of students’ grade distributions and instructors’ responsiveness on the online platform by an in-house data analytics team, and additional evaluations if students file complaints about instructor performance.

If these evaluation channels show the instructor has not met university standards, the instructor receives a warning. Instructors who have received a warning are monitored more closely in subsequent courses and may not be rehired if their performance does not improve.

Data and Methods

Data. We analyze university administrative records covering all sections of Math I and all instructors and students who have taken or taught it at least once between January 2001 and July 2014. The data include 339,844 students in 26,384 course sections, taught by 2,243 unique instructors. We also focused on a subset of data for which final exam scores are available, which includes 94,745 students in 7,232 Math I sections taught by 1,198 unique instructors. In addition, we analyzed student performance among students who pass Math I and enroll in the next required course in the sequence, Math II.

We reviewed performance data and demographic descriptors for instructors and students. For instructors, we reviewed their teaching history; self-reported information on ethnicity, gender, and residential zip code; and salary. Instructors in our study are majority white and male, have been at the university for an average of nearly five years, and earn $950 per class. They typically have taught more than 40 total course sections at UPX, including 15 sections of Math I. We calculate several experience measures for instructors, including the number of courses taught in the previous calendar year and total cumulative experience teaching, including Math I.

For students, we review course-taking histories, including grades and credits earned; demographics such as gender, age and residential zip code; and final exam scores when available. About two-thirds of students in our study are female and their average age is 35 years old. At the outset of the class, they typically have already taken 23 credits worth of classes at UPX, with an average GPA of 3.35. About 10 percent have taken (and failed to complete successfully) Math I before. Based on transcript data, we construct measures of student success, including course completion, cumulative grade point average (GPA), and cumulative credits earned, both before and after enrolling in Math I.

We also obtain end-of-course evaluations for sections between March 2010 and July 2014, in which students rate the instructor on a 10-point scale. However, student end-of-course evaluations are optional, and just 37 percent of Math I students completed them.

Methods. To measure instructor quality, we compare the outcomes of students who took the same course with different instructors, drawing on the sort of value-added model often used to measure teacher effectiveness in K–12 (see “Choosing the Right Growth Measure,” research, Spring 2014). These models take into account differences in student characteristics, and then compare the outcomes of similar students taught by different instructors on the same campus.

We account for student characteristics such as gender, age, GPA, credits earned at UPX prior to Math I, whether students have taken Math I before, the number of times they have taken the class, the years since they started their program, and the academic program in which the student was enrolled. We also control for total section enrollment, the averages of individual characteristics by section, and student zip code characteristics available through U.S. Census data, such as the local unemployment rate, median family income, percentage of families below the federal poverty line, and percentage of residents with a college degree. Finally, to account for any changes over time in unmeasured student characteristics or grading standards, we compare only students who take Math I in the same month.

Measuring instructor effectiveness accurately requires confronting three key issues: nonrandom assignment of students, variation in grading practices among instructors, and unmeasured differences across campuses.

First, could our estimates be biased by students seeking out especially skillful instructors? According to UPX administrators, there is no sorting at all in online courses, which is plausible given the very limited interaction students have with instructors in the initial meetings of the course. Some sorting in face-to-face courses is possible, but administrators believe this is minimal, as do we. On average, we find very few systematic correlations between student and instructor characteristics within individual campuses. Furthermore, the inclusion of the rich student characteristics mentioned above has little impact on our estimates of instructor value-added.

Second, student grades do reflect, at least in part, different grading practices by instructors. As a result, we also report results based on standardized final exams. This information is available for sections between July 2010 and March 2014, about 30 percent of the full sample, and we can identify final exam scores with high confidence for 88 percent of those students. We also examine future performance measures, such as students’ grades and cumulative GPAs six months and one year later. These outcomes are unlikely to be affected by individual instructors’ grading practices.

Finally, the UPX setting is unique in postsecondary education in that many instructors teach at multiple campuses, especially when the online campus is included. These campus “switchers” permit us to make performance comparisons between instructors across campuses while also controlling for cross-campus differences in unobserved student factors.

Effective Instruction Has Lasting Impacts

Our analysis shows that student outcomes vary considerably across instructors, both online and among those who teach the same course in person on the same campus. Students’ outcomes in Math I vary by their instructor in that class, as do their outcomes in the following Math II class. An increase of 1 standard deviation in instructor quality in Math I—equivalent to having a teacher at the 87th percentile of effectiveness rather than an average teacher—is associated with better outcomes for students, such as increases in students’ grades of 0.30 standard deviations in Math I and 0.20 standard deviations in Math II (see Figure 1). In other words, students with more-effective instructors usually get better grades, on the order of moving from a “B” to “B+.”

It’s a boost with staying power. Instructor effects on students’ grades in Math I and Math II are highly positively correlated in both formats, with a correlation coefficient of 0.60. This tells us that Math I instructors who successfully raise student performance in Math I also raise performance in follow-on courses. Instructors do not appear to be improving current performance in ways that harm longer-term learning.

We find that instructor quality is more variable among in-person sections than online. For grades in Math I, this difference is relatively modest: a one standard deviation increase in instructor quality improves grades by 0.32 standard deviations for in-person sections and by 0.25 standard deviations for online sections. However, there is a large difference by format when measuring student performance in the subsequent Math II class, where effectiveness varies by 0.24 standard deviations for in-person instructors and 0.04 standard deviations for online instructors.

There are a number of reasons that the impact of instructors may vary less in online courses than in face-to-face classes. Face-to-face instructors interact with students directly and have more discretion in amending and supplementing course materials. Online instructors, by contrast, operate under more centralized management and may not interact with their students in real time. Differences among the types of students and instructors who choose the online versus in-person setting may also be a factor.

Effects are different depending on which measure of student outcomes and course format is considered. For in-person classes, instructors’ effects on final exam scores vary more than those on course grades, with a standard deviation of 0.49 compared to 0.32. Effectiveness is highly positively correlated between current and follow-on course performance, at 0.60 for both grades and for test scores.

In online sections, both findings are reversed: the standard deviation of instructor effects on test scores is 0.14 for exam scores compared to 0.25 for grades. There also appears to be a weaker correlation with follow-on course performance for online sections for students’ final exam scores.

How big are the effects of instructor quality on students’ final exam scores? We compare them to outcome differences by student characteristics known to be important factors in their success: age and prior academic performance. On the standardized final exam, older students earn scores that are 0.15 standard deviations lower than classmates who are a decade younger. A 1-point difference in a student’s incoming GPA is associated with a difference of 0.46 standard deviations on his or her final exam score. So, having an instructor who is 1 standard deviation more effective produces a test score change that is larger than the gap between 25- and 35-year-old students, and comparable to the gap between students entering the class with a 3.0 compared to a 2.0 GPA. Instructor quality thus appears to be an important factor.

Finally, we examine instructor effects on measures of longer-term success at UPX, including students’ likelihood of taking Math II and the number of credits earned in the six months following Math I (see Figure 2). Students with more-effective instructors are more likely to continue their studies: we find an increase of 1 standard deviation in instructor quality in Math I is associated with an increase of 5 percentage points in the likelihood that students progress to Math II. The variability is twice as large for in-person sections as it is for online ones. An increase of 1 standard deviation in instructor quality is associated with an increase of 0.13 standard deviations in the number of credits earned in the six months after Math I. Similarly, we find twice as much variability for in-person instructors as for online instructors.

Does Better Teaching Mean Better Evaluations and Pay?

Although course grades and final exam performance provide two objective measures of student learning that can be used to assess instructor quality, end-of-course evaluations by students are the primary mechanism for assessing instructor quality at UPX and most other institutions. Our analysis considers a rating of eight or above by students as favorable, as these scores are considered “good” by the UPX administration.

We find substantial variability in student evaluation ratings across all instructors in our study, with one standard deviation of student-assessed effectiveness corresponding to a difference of 22 percentage points in the fraction of evaluations that are positive (higher than 8 on a 10-point scale). However, student evaluations of their instructor’s performance are most positively correlated with their grades in that class, suggesting that instructors may be rewarded through higher evaluations for high course grades. Correlations with subsequent course performance are much weaker. Collectively, this suggests that end-of-course evaluations by students are unlikely to capture much of the variation in instructor quality, especially for more distant or objective outcomes.

We also consider how instructors’ experience and pay correlates with their effectiveness. Are more-experienced instructors more effective? Are more-effective instructors paid more? For this analysis, we focus on the 18,409 sections taught by instructors hired since 2002 so that we can examine data on their full teaching history.

We find that experience matters somewhat, but that the pattern differs markedly across outcomes. Teaching Math I at least once before is associated with an increase in instructional effectiveness of 0.03 standard deviations, measured by students’ grades. However, teaching the class more than once in the past has very little additional impact. Experience has no impact on student performance in subsequent courses. And an instructor’s general experience teaching other subjects has little association with effectiveness in Math I. Consistent with the flat salary schedule, instructor salary is unrelated to measured effectiveness, varying only modestly with tenure.

Implications

We find that college instructors vary widely in their contributions to student performance—some consistently help their students succeed more than others do. And a skillful instructor provides benefits to students that last beyond a single class: those students are more likely to succeed in subsequent courses, earn more credits, and thus be better positioned to complete their degrees. Yet colleges largely ignore these important differences in performance, and instead rely on subjective measures like students’ end-of-course evaluations, which fail to truly differentiate between effective and ineffective instructors. In addition, instructor compensation is not linked to classroom performance in any direct way but rather is tied primarily to tenure and experience.

Although our analysis is based on a single university and enabled by its nontraditional format, we believe it could have broad applicability among other institutions of higher education. Our findings imply that personnel decisions and policies that attract, develop, allocate, motivate, and retain effective faculty are a potentially important tool for improving student success and institutional productivity. While these factors may not generalize to all sectors of higher education, they could be relevant at for-profit colleges, as well as at less-selective four-year and two-year schools, such as community colleges. In these institutions, which are focused on teaching rather than research, personnel policies that focus on meaningful differences in effectiveness among instructors could theoretically be adopted, to the benefit of students.

The for-profit sector, in particular, may be ripe for innovation. For-profit institutions like UPX have come under greater scrutiny over the last decade, and several high-profile closures have made improving student outcomes a first-order concern. New regulations and mandated transparency measures have come online, such as Gainful Employment and the College Scorecard.

While our study does not directly speak to the specific strategies that institutions should use to identify and improve instructor effectiveness, it does indicate the potential of such practices. For-profit institutions like UPX and other nonselective institutions, or any institution looking to improve graduation rates, should consider the important variations in performance among their instructors. By putting more effective instructors at the front of the class, they can improve the outcomes of college students—particularly those least likely to succeed.

Pieter De Vlieger is a graduate research assistant at the University of Michigan. Brian A. Jacob is co-director of the Education Policy Initiative (EPI) at the University of Michigan, where he is a professor of education policy, economics, and education. Kevin Stange is assistant professor of public policy at the University of Michigan.

Over the past decade, Hispanic students have graduated high school and entered college in growing numbers. Yet the rate of Hispanic college completion has remained persistently lower than that of whites and other ethnic groups in the United States: only 23 percent of Hispanic adults hold any postsecondary degree compared to 42 percent of all adults. Helping raise the Hispanic college graduation rate is an urgent goal, given the persistently high rate of poverty among Hispanic families, growth of the Hispanic population to account for one in five college-age Americans, and mounting concerns about racial and economic inequality.

The question is, how?

One potential strategy involves helping high school students broaden the set of colleges to which they apply and enroll. Hispanic students may be more constrained in their college-selection process than other groups, and are far more likely to attend two-year colleges, which typically have far lower graduation rates than four-year institutions. Just 56 percent of Hispanic college students enroll at four-year institutions compared to 72 percent of non-Hispanic white students. Hispanics are also less likely than members of other ethnic groups to earn a bachelor’s degree: 15 percent of Hispanics have a bachelor’s degree, compared to 33 percent of whites, 54 percent of Asians, and 22 percent of African Americans (see Figure 1).

We examine an intervention designed to expand Hispanic students’ college exposure: the National Hispanic Recognition Program (NHRP), a College Board initiative that identifies top-performing Hispanic students based on their 11th-grade Preliminary SAT/National Merit Scholarship Qualifying Test (PSAT/NMSQT) scores. NHRP changes two key features of their high-school experience. First, the College Board notifies students and school staff, such as school counselors, about this prestigious award. Second, with the student’s permission, the College Board shares lists of NHRP honorees with postsecondary institutions looking to recruit Hispanic students. We measure the impact of this early, pre-application recognition on students’ enrollment decisions, as well as their college persistence and degree attainment rates.

We find evidence that the program induces students to apply to and attend more elite institutions, shifting students from two-year to four-year institutions as well as to out-of-state and public flagship colleges, all areas where Hispanic attendance has lagged. Overall, NHRP recipients are 1.5 percentage points more likely to enroll at a four-year institution, 5 percentage points more likely to attend both an out-of-state college and a recruiting institution, and 3 percentage points more likely to attend a public flagship institution. The program’s impact on college completion is generally positive but statistically insignificant; however, we find sizable increases in bachelor’s degree completion among students who otherwise were at the highest risk for dropping out of college.

Together, these findings demonstrate that college outreach can have substantial impacts on the enrollment choices of Hispanic students and can serve as a lever for institutions looking to draw underrepresented, academically talented students.

An Elite Honor for Hispanic High-School Students

NHRP was founded in 1983 by the College Board, a nonprofit that advocates for expanded access to higher education and administers college-level exams such as the SAT. Similar in spirit to the National Merit Scholarship Program, an annual academic scholarship competition conducted by the National Merit Scholarship Corporation, the NHRP was designed to recognize outstanding Hispanic high-school students and encourage them to enroll in college. The program identifies the top 2.5 percent of Hispanic students each year based on their performance on the 11th-grade PSAT/NMSQT, which assesses skills in math, critical reading, and writing. About 5,000 students are recognized by the NHRP annually, with California and Texas having the largest numbers of NHRP scholars.

To establish eligibility requirements, the College Board analyzes PSAT/NMSQT performance among Hispanic students within six different geographic regions each year, and then identifies the score that separates the top 2.5 percent of test-takers from other students in that region. The eligibility cutoffs range from the low 180s to the mid-190s out of a maximum of 240 points. Students with qualifying scores are invited to join the program. To be formally accepted, they must then verify that they are at least one-quarter Hispanic, and their high school must document that the student’s junior year cumulative GPA is at least 3.5. Almost all students who are initially deemed eligible for the program are able to satisfy both requirements.

Students recognized by NHRP are more likely to live in cities and attend large high schools with significantly more low-income and Hispanic students, compared to white students with similarly strong PSAT/NMSQT scores. They are also about four times as likely to have families with incomes below $50,000 and to have parents who did not graduate from high school. Compared to their white peers, they take and pass fewer Advanced Placement (AP) exams, and are less likely to attend four-year or out-of-state colleges.

The program does not provide any direct financial reward to students but has the potential to influence their college preparation and application decisions in other ways. The College Board sends a letter directly to students that congratulates them for being in the top 2.5 percent of Hispanic students nationwide—a clear and perhaps surprising recognition of academic ability. Students are encouraged to participate in the program and list it on college, scholarship, internship, and job applications. The organization also informs school counselors of newly recognized students, and asks that the staff members help the students complete the necessary paperwork and apply to top universities as well as honor them through some type of school recognition.

In addition, with the student’s permission, the College Board shares a list of NHRP recipients with about 200 four-year postsecondary institutions hoping to recruit academically exceptional Hispanic students. These recruiting institutions, as measured by rankings, graduation rates, and average SAT scores published by Barron’s Profiles of American Colleges, are more competitive, on average, than other, non-recruiting four-year institutions attended by Hispanic students who fall just short of qualifying for the NHRP program. They are also slightly larger and more expensive, though the percentages of enrolled students identifying as Hispanic are comparable at recruiting and non-recruiting schools.

Thus, NHRP status can affect students’ college-going decisions and degree completion through two primary mechanisms. First, research has shown that many high-performing students do not apply to elite institutions even though they are eligible, and that informing them about college options can shift where students apply and ultimately enroll. Second, recognized students often receive targeted outreach and financial incentives from colleges that actively recruit NHRP scholars as part of their efforts to increase diversity among their students.

Data and Method

To conduct our analysis, we constructed a national sample of all Hispanic 11th-grade PSAT/NMSQT takers in the U.S. from the graduating high-school classes of 2004 to 2010. We then linked individual-level records to a number of auxiliary data sources from the College Board and the National Student Clearinghouse, including basic demographic information, the high school they attended, eventual college enrollment, their history of SAT attempts, the institutions to which they sent their SAT scores, and AP participation. We also included information from the Common Core of Data (CCD) and the Private School Universe Survey (PSS), which include data about the size, demographics, and geographic location of the high school attended when each student took the PSAT/NMSQT, as well as the Integrated Postsecondary Education Data System (IPEDS), which includes the characteristics of postsecondary institutions.

We study the impact of NHRP on student outcomes using what is known as a regression discontinuity design. This approach takes advantage of the program’s initial eligibility requirement: a numeric cutoff. Because students are unable to precisely manipulate their PSAT/NMSQT scores, those scoring just below the cutoff should be essentially identical to those whose scores are just above. Furthermore, because the eligibility cutoffs vary by year and are unknown in advance, and because students have only one opportunity to take the PSAT/NMSQT in 11th grade, there is no way for students to “game” the system. We restrict our analysis to students whose PSAT/NMSQT scores are within 15 points on either side of the cutoff, to produce a data set of approximately 58,000 students across all years.

Our preferred results adjust for the region and year in which each student took the PSAT/NMSQT, as the eligibility cutoff varies by region and year. This adjustment also accounts for unmeasured differences in high school and college policies, such as state spending on higher education, changes in high school curricula, and the relative competitiveness of college admissions in a given year. We also confirm that we obtain similar results when we control for student characteristics measured at or before the PSAT/NMSQT, including sex, parental education, family income level, whether a student took the PSAT/NMSQT in 10th grade and his or her previous score, indicators for ethnic background (for example, Mexican, Cuban), and controls for the type of high school attended, including affiliation (public or private), urbanicity (that is, city, suburban, rural), size, and concentration of Hispanic students.

Results

NHRP eligibility has a significant effect on college attendance patterns. Eligible students are 5 percentage points—almost 16 percent—more likely to attend colleges that recruit Hispanic students through NHRP (see Figure 2). They are also more likely to attend out-of-state and public flagship institutions by roughly 5 and 3 percentage points, respectively, with much of that effect due to students opting to attend recruiting institutions. In addition, students are approximately 1.5 percentage points more likely to enroll at a four-year institution, with about two-thirds of this effect driven by movement away from the two-year sector (see Figure 3).

Closer inspection reveals a group of seven universities that account for virtually all of the shift toward NHRP-recruiting schools, with eligible students 5.8 percentage points more likely to attend one of these schools than their non-NHRP peers. Given a baseline of 4.1 percentage points, this means students were more than twice as likely to attend one of these seven core recruiting institutions as a result of the program.

For privacy reasons, we cannot identify the particular colleges, all of which are large public institutions located outside of California and Texas. Three are included on Barron’s “Highly Competitive” list, four serve as state flagships, and all seven offer substantial financial aid for NHRP scholars and make this information easily available to prospective students on their websites.

Given the sizable shift of students toward recruiting institutions, the primary mechanism by which NHRP affects students’ initial college choices appears to be a combination of direct college outreach and financial aid. An alternate, or perhaps complementary, theory is that NHRP induces students to improve their academic preparation in high school, making them more attractive candidates to these colleges. We investigate this possibility, however, and find no evidence that NHRP scholars improve their academic performance during their senior year; they do not take or pass more AP exams, score higher on the SAT, or retake the SAT more often than non-NHRP students.

We also examine whether NHRP-eligible students are more likely to earn a bachelor’s degree within four years. Our overall results are positive though statistically imprecise, with NHRP eligibility increasing degree completion by 1.3 percentage points. We also examine the effects of the program on six-year graduation rates. The results are noisier due to a smaller sample size, but follow a similar positive pattern. These positive but imprecisely estimated completion impacts suggest that, at the very least, high-performing Hispanic students are not harmed by enrolling at colleges that they may not have ordinarily considered.

The fairly small effect on overall completion conceals much larger effects on where NHRP-eligible students earn their degrees. The percentage of students earning bachelor’s degrees in four years at recruiting institutions increases by 4 percentage points. Moreover, NHRP eligibility increases the share of students who earn degrees outside their home state by 2.8 percentage points, or 16 percent, suggesting that NHRP helps to disperse high-achieving Hispanic students by geographic region. We estimate that nearly 60 percent of students induced to attend out-of-state colleges ultimately earned bachelor’s degrees in four years. When the time frame is extended to six years, this jumps to nearly 80 percent. This evidence casts doubt on the prevailing sentiment that high-performing Hispanic students may struggle to flourish in unfamiliar or uncomfortable postsecondary environments, or when separated from immediate family structures.

Who Benefits the Most?

Given the variation in high school environments, college quality, state policies, and proximity of recruiting colleges across the United States, we should expect a range of responses to NHRP eligibility depending on where students reside. When it comes to college enrollment, we do find strong regional differences, with students in the western and southwestern regions driving the overall results. There is almost no impact on attendance patterns for students in the other four regions, except for an increase of 2.4 percentage points in the likelihood of attending one of the seven core recruiting institutions. That figure, over a baseline of 0.2 percent, indicates the strong effects of college recruitment even in the presence of distance. Perhaps surprisingly, given the cross-regional differences in enrollment patterns, the effects of NHRP on degree attainment are relatively equal across regions, measuring 1.1, 1.4, and 1.6 percentage points in the West, Southwest, and other regions, respectively.

Are there other student characteristics that coincide with varied effects of the program? We found the outcomes are driven entirely by students at public high schools. The strongest effects are among students whose high schools have the highest concentrations of Hispanic students or are in more rural environments (see Figure 4). These students experience the largest increases in four-year college enrollment and out-of-state college enrollment, as well as an increase of 2 to 6 percentage points in the likelihood of attending a college in the “Most Competitive” category. We also find sizable increases in four-year degree completion, ranging from 4 to 8 percentage points among the types of disadvantaged students who typically have the lowest completion rates.

We also examine differences in program impacts based on student academic performance, as measured by a student’s SAT score. We find the largest impact for students whose SAT scores are in the bottom third of NHRP-eligible students—those traditionally at the highest risk for dropping out of college. These students’ four-year bachelor’s degree completion rates increase by more than 4 percentage points, or 10 percent. In addition, their college attendance increases by 10 percentage points at recruiting colleges, and by 4 percentage points at both four-year institutions and “Most Competitive” colleges. By contrast, the impacts for students whose SAT scores were in the highest third of all NHRP-eligible students were either considerably smaller in magnitude or indistinguishable from zero.

Implications

The National Hispanic Recognition Program has a meaningful impact on the college-going decisions of high-achieving Hispanic students. However, simply offering recognition for academic performance is insufficient for achieving the college enrollment and completion effects documented in this study. Rather, our results suggest that college enrollment and completion can be positively influenced by higher-touch efforts on the part of colleges, such as the targeted recruitment and financial aid offers that accompany NHRP recognition.

Such active outreach and financial support are particularly important for students and high schools with fewer resources, and stepping up efforts to publicize these opportunities among high-achieving underrepresented students may be an effective way of increasing their postsecondary success. Our results also confirm that colleges actively seeking to enroll high-performing Hispanics can accomplish those goals, which is consistent with a mission of creating a diverse and academically strong student body.

Taken together, our results reveal several straightforward lessons. First, inducing students to attend out-of-state colleges likely requires large financial incentives, given that all the observed out-of-state shifting in this study is driven by colleges known to provide large student aid packages. Second, this program allows colleges to effectively recruit Hispanic students attending a wide variety of high schools. The consistency of effects suggests that NHRP serves as an effective recruitment tool for Hispanic students from diverse backgrounds. Third, student responses are regionally specific, emphasizing the role of the higher-education marketplace and local options in selecting which colleges to attend.

Perhaps most important, the results of this program show that high-achieving Hispanic students succeed at selective post-secondary institutions, and that any culture shocks from unfamiliar environments are not impeding completion. We find the opposite, as NHRP scholars from predominantly Hispanic high schools actually experience large increases in four-year bachelor’s degree completion rates. These findings challenge the narrative that unique cultural circumstances and intense family and community ties shared by Hispanic students might impede their success at colleges far from home.

Oded Gurantz is an Institute of Education Sciences fellow in the Stanford University Graduate School of Education. Michael Hurwitz is senior director at the College Board. Jonathan Smith is an assistant professor of economics at Georgia State University. A more detailed account of this investigation can be found in the Winter 2017 issue of the Journal of Policy Analysis and Management.

“Scarcely any political question arises in the United States that is not resolved, sooner or later, into a judicial question,” observed Alexis de Tocqueville in 1835. The battle over school choice in American education has been no exception.

In its 2002 decision in Zelman v. Simons-Harris, the U.S. Supreme Court erased all doubt as to whether the use of government funds to send children to religious schools violates the First Amendment’s ban on the “establishment of religion.” It does not, the court said in a 5–4 decision. Yet the controversy has continued to swirl, in part because many state constitutions include more-specific provisions barring aid to religious organizations. The uncertainty has left some state legislators reluctant to consider school voucher proposals and subjected the few state-authorized voucher programs in existence to a seemingly unending series of legal challenges.

The Supreme Court has a new opportunity to clarify matters in a case scheduled for oral argument on April 19, just days after Justice Neil Gorsuch’s arrival on the bench. Trinity Lutheran Church v. Pauley involves a Missouri church that was turned down by a state program that provides grants to nonprofits to resurface their playgrounds with rubber from recycled scrap tires. The state justified denying the benefits on language in its constitution stating that “no money shall be taken from the public treasury, directly or indirectly, in aid of any church, sect or denomination of religion.”

The question now facing the court is whether such provisions, if used to exclude organizations from aid programs based solely on the organization’s religious status, violate the federal Constitution’s guarantee of free religious exercise. If so, voucher opponents will find it harder to argue that the religion clauses of state constitutions are a barrier to the creation of school voucher programs. Indeed, while Zelman established that states enacting voucher programs may include religious schools, a decision in favor of Trinity Lutheran could imply that they must.

The court should not shy away from that conclusion. Thirty-seven states have constitutional provisions similar to Missouri’s. Known as Blaine Amendments, these provisions are named for the 19th-century presidential aspirant James G. Blaine who, as Speaker of the House, sought to exploit nativist, anti-Catholic sentiment by offering a constitutional amendment barring the government from funding religious schools. Blaine’s proposal to amend the federal Constitution narrowly failed in the Senate, but 22 states adopted similar provisions over the next half century. Missouri adopted its Blaine Amendment in 1875, one year before Congress rejected the concept.

The court passed on a similar opportunity to strike down Blaine Amendments in 2004. In Locke v. Davey, it upheld Washington State’s decision to prevent students pursuing theology degrees from receiving a state-funded scholarship. But it was careful to limit its decision to the narrow question of whether states can withhold funding for the training of clergy, which it noted was historically “one of the hallmarks of an ‘established’ religion.”

How will the justices rule this time? Court watchers note that they accepted the case in January 2016, which ordinarily would have led them to hear it later that year. It therefore seems likely that the justices delayed the case to avoid a 4–4 split after Antonin Scalia’s unexpected death. In short, the outcome would appear to turn on the views of its newest member—a product of Catholic schools who, as Arizona Supreme Court justice Clint Bolick explains in this issue (see “Gorsuch, the Judicious Judge,” features), has interpreted the establishment clause narrowly during his tenure as an appellate judge.

A decision in favor of Trinity Lutheran would hardly end the legal conflict over school vouchers. Depending on the breadth of the court’s ruling, critics may be able to argue that subsidizing school attendance advances religion in more significant ways than playground modernization does. And voucher opponents have been creative in identifying a wide variety of constitutional provisions, having nothing to do with religion, under which to challenge school choice programs.

Yet such a decision would bring the most persistent and pervasive aspect of that conflict nearer to a close; it is no wonder, then, that the National Education Association has filed an amicus brief in support of the state.

Furthermore, it would help stamp out the remaining legacy of one of the most infamous politicians in U.S. history. It is ironic that this prospect may depend on the thinking of a justice appointed by a president whose rise to power was fueled by nativist appeals.

— Martin R. West

On March 22, the U.S. Supreme Court decided the most significant special-education case in 35 years. In Endrew F. v. Douglas County School District, the justices unanimously ruled that, under the Individuals with Disabilities Education Act (IDEA), public school students with disabilities are entitled to greater benefits than some lower courts had determined.

Endrew F. (Drew), an autistic child in Douglas County, Colorado, showed severe and increasing behavioral problems from preschool through 4th grade. Dissatisfied with his lack of progress under his Individualized Education Program (IEP), his parents withdrew him from public school in 2010 and enrolled him in a private school specializing in serving autistic students. He made significant gains in the new school and is now a 17-year-old high-school student learning vocational skills.

Drew’s parents believed that under the Individuals with Disabilities Education Act (IDEA), they were entitled to reimbursement from the Douglas County School District for the cost—$70,000 per year—of Drew’s private education. The school district, they argued, was not providing Drew with the “free appropriate public education” required by IDEA, thus qualifying him for placement in a private program.

The school district disagreed. While Drew was not making significant progress, he was making some progress and that was all the law required. The district pointed to a 1982 Supreme Court decision, Board of Education v. Rowley, which held that schools must merely provide “some educational benefit” for children with disabilities. As long as an IEP was “reasonably calculated to enable the child to receive educational benefits,” the school district had complied with the law. Thus, the district contended, anything more than a trivial benefit was sufficient.

In response, Drew’s parents argued that revisions to IDEA since Rowley showed that Congress intended an IEP to provide opportunities that are “substantially equal to the opportunities afforded children without disabilities.” As well, while several federal circuits had followed Rowley, other courts had imposed a higher standard and still others had produced conflicting precedents.

Drew lost his case before an administrative law judge and before the Tenth Circuit Court of Appeals, leading to the Supreme Court showdown. At oral argument in January, the justices were torn. Even as they indicated that IDEA must surely require more than just “some benefit,” they seemed to have no idea how to articulate a clear alternative standard. They seemed equally troubled by the potential consequences of changing course, worrying that a higher standard would impose unreasonable costs on school districts. Justice Breyer expressed concern that judges, who don’t know much about education policy, would be called upon to second-guess the judgment of experts and, in doing so, would encourage still more litigation, saying, “I foresee taking the money that ought to go to children and spending it on lawsuits and lawyers and all kinds of things that are extraneous.”

Those concerns did not prevail, though they clearly influenced the court’s reasoning. While the justices unanimously overturned the Tenth Circuit decision, the standard they articulated required that the educational benefits provided to students with disabilities be meaningful but not necessarily equal to those provided to other students.

Writing for the court, Chief Justice John Roberts noted that the circumstances in Rowley were different. That case involved a child who could be integrated into a regular classroom and, with appropriate support, make grade-level progress. But that was not possible for Drew. For children like him, Roberts said, an IEP must be reasonably calculated “to make progress appropriate in light of the child’s circumstances.” Thus, an IEP must provide more than some benefit but schools cannot be expected to provide grade-level progress for students with severe disabilities. Roberts explicitly declined to provide a “bright-line rule” or to “elaborate on what ‘appropriate’ progress will look like from case to case.” Roberts further wrote that courts should give “deference” to the “expertise and exercise of judgment by school authorities.” Because of this lack of clarity, there is in fact no guarantee that Drew’s parents will ultimately prevail. The court simply remanded the case back to the Tenth Circuit to be reconsidered in light of the higher standard.

In short, thanks to this decision, we now know that IDEA requires meaningful benefits. We just don’t know what “meaningful” means.

Joshua Dunn is professor of political science at the University of Colorado–Colorado Springs.

This article was updated to reflect news developments.

Class Clowns: How the Smartest Investors Lost Billions in Education
by Jonathan A. Knee
Columbia University Press, 2017, $29.95; 288 pages.

As reviewed by Julie Landry Petersen

In his new book, Class Clowns, Jonathan Knee wraps up his case studies with a list of lessons he frames as “cautionary reminders, empirical observations, and hopeful exhortations.” It’s a humble but accurate summary of the book overall. Although Knee is a business-school professor and former investment banker, his book reads less like a coherent analysis of the prospects of education businesses and more like a few standalone cautionary tales.

Knee profiles four large education companies—Edison Schools, Amplify, Houghton Mifflin, and Knowledge Universe—that either received investments from or were acquired by players in the media and entertainment industry in which the author formerly worked. His main argument is that despite a genuine desire to improve education, these companies and their investors dreamed too big, had flawed assumptions about the nature of the market, bungled many operational details, and failed to hedge against the realities of the market as their dreams dwindled. “These mistakes have often flowed from a basic misunderstanding of the education ecosystem specifically and the importance of industry structure more generally,” Knee writes, adding that the lessons he draws can “help not just investors and business people but also policy makers and administrators avoid some of the more treacherous and persistent pitfalls.”

If only Knee had taken this observation to heart and provided the reader with an upfront analysis of that education ecosystem. This broader context and the resulting strategic lessons are often obscured behind a laundry list of criticisms that range from the tactical to the personal. By contrast, the author explains little about the way public-education spending itself works, let alone about the customers’ decisionmaking processes or the skills required to manage and scale these businesses.

Knee does capture some unique journalistic details, taking the reader behind the curtains of businesses whose founders’ ambitions were often hazy and grander than would prove attainable. These details are sometimes useful to understanding the business itself, while in other cases they obscure the real picture of what went wrong. The first chapter centers on Edison Schools, a company that sought to take over the management of public schools. Here, the author expends more energy on personal attacks than on market analysis. Knee, like many other writers, portrays founder Chris Whittle as a lavish spender heavy on vision and light on operational expertise (a combination that Whittle himself dubs “robust naiveté”). Despite the reputation of its leader, Edison attracted significant capital from top-tier investors and went public in 1999, only to be sold four years later for pennies on the dollar. In the interim, Edison found more resistance than revenue as it tried to persuade states and districts to turn over management of their schools to a private company using a new, unproven school model.

The following chapter, on Amplify, is a more successful analysis of a market and the company’s challenges in serving it. It starts with an extended discourse on Rupert Murdoch’s rocky investment history and tendency to pursue things he liked without respect to their likely value. Knee then hits his stride, showing how these factors played out when Murdoch’s News Corporation acquired the learning analytics company Wireless Generation in 2012 to create an ambitious new “Amplify” division focused on educational tablets, games, curriculum, and data. Here, in contrast to his approach in other chapters, Knee illustrates well the politics that affected Wireless Generation and Amplify, and shows how Amplify’s management team underestimated the challenges of curriculum development and adoption while spending aggressively to expand its narrow product focus. Ultimately, Knee admires the Amplify investors and executives who in 2015 took the company private for their focus on “the specific incremental needs of the multiple constituencies that drive buying decisions, rather than the general transformational goals of the founders.”

The chapters on publisher Houghton Mifflin and the ambitiously diversified Knowledge Universe—whose services have ranged from childcare and toys to private schools and technology training—show how both businesses expanded in too many directions without ensuring there was real synergy to be gained. However, it is hard to extract meaningful lessons from this material. Houghton Mifflin, for example, started out with a competitive advantage in education, but over time, it struggled, thanks to the “fundamentally capricious nature of the K–12 end market,” as well as ineffective attempts to pursue globalization (though education content is fundamentally local), technology (despite schools’ resistance), and “edutainment” (rarely successful). These points are unfortunately buried among debt terminology and the mechanics of private-equity investment deals.

It’s easy to pick on a failed company in hindsight. What’s unclear is what Knee would recommend for future education entrepreneurs and investors, aside from a narrower vision and some increased caution. I tried to attack this question myself from the opposite direction in an article for Education Next by examining the few successful education “exits” (see “For Education Entrepreneurs, Innovation Yields High Returns,” features, Spring 2014). Ironically, because I measured “success” in terms of the likely returns to the original investors at the moment these companies exited—that is, went public or were sold—my list of companies overlapped greatly with Knee’s, including both Amplify and K12, a Knowledge Universe company. At the time, these companies seemed to be successful investments for their early-stage investors (though storm clouds were beginning to gather for both), but Knee deems them failures for their acquirers and later-stage investors. Clearly, the definition of success is critical, as are timing and perspective.

Ultimately, Knee is at his best when he focuses on the mechanics of building a strong education business, but he fails to help the reader understand how money and power flow in that context. He glosses over national policy and rarely examines the functions of state boards of education, district officials, and school principals, who all play a significant role in what products and services are adopted and used in schools. Knee also misses an opportunity to address the elephant in the room that any education business must face: the impact of such companies on their customers or end users, including effects on student learning, teacher quality, school productivity, or district cost savings.

In Knee’s view, “the greatest successes come from a series of targeted incremental steps forward.” But without a deeper look at any real success stories of this sort, it’s difficult to imagine this mythical success and why it eluded the companies Knee profiled. And without a comprehensive examination of the education market in which these companies operated, the reader is left unsure whether these firms were bad ideas or just badly timed investments—or whether students and schools have gained or lost anything as a result.

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

In his 2008 blockbuster, Outliers, Malcolm Gladwell makes the case that a person’s age relative to his or her cohort is a key predictor of success. That is, the older you are in relation to your peers, the more likely you are to perform at an elite level in sports, to excel in school, and even to attend college. We see this principle applied in college athletics when coaches “redshirt” freshman athletes, allowing them to practice with the team but not play in official games. Redshirting gives younger athletes an additional year to develop skills and extends their playing eligibility, since colleges allow these freshmen five years to attend and compete.

On the other end of the student age spectrum, many parents of preschoolers have bought into this concept, choosing to delay their child’s entry into kindergarten for a year—a practice known as academic redshirting. Their justifications parallel those of college coaches: these parents believe that their children need that extra year to develop the necessary skills and maturity to succeed in kindergarten. A redshirted child is a year older at kindergarten entry and thus becomes one of the oldest in his class and remains so throughout his school years, enjoying the presumed advantages of age.

Preschools and elementary schools often recommend redshirting, asserting that it bestows the “gift of extra time,” but parents should take such advice with a grain of salt. After all, a preschool stands to gain financially from the practice, since the school will likely capture another year’s tuition. And elementary schools may also have mixed motives: older children are easier to teach and they perform at higher levels, just by virtue of being older. In other words, older children make the school’s job a little bit easier.

How should parents decide whether they should enroll their child in kindergarten when he is first eligible or hold him back for a year? In this article, we draw upon our combined experience—Schanzenbach as an education researcher and Larson as a preschool director—to provide some practical, evidence-based advice. Notably, we find that Larson’s take on the issue, formed by 14 years of experience with preschoolers and their parents, accords perfectly with Schanzenbach’s conclusions based on academic studies: redshirting is generally not worth it.

Despite the weightiness of the decision, rest assured that a child is likely to be successful whichever path his parents choose. We know dozens of families who have redshirted their children and have been perfectly happy with the outcome. On the other hand, in most instances there is a good case to be made for resisting the pressure—not only from schools but sometimes from other parents as well—and sending a child to school when he is first age-eligible.

Why Delay? 

We recognize that deciding whether or not to redshirt a preschooler is difficult. Parents want what’s best for their children now and in the future, and they have to make the kindergarten-enrollment decision with limited and uncertain information. The concerns that lead parents to contemplate redshirting are most often related to the child’s physical, social, and emotional maturity as the parents perceive it. In particular, parents seem to wrestle most with the redshirting decision when they have a son whose fifth birthday falls just before the cutoff date for kindergarten eligibility, which is most commonly on or around September 1. If the child starts kindergarten “on time,” he will be among the youngest in his grade; if he is redshirted, he will be one of the oldest.

In terms of physical maturity, it is true that redshirting changes the child’s relative height in the kindergarten class. For example, using national data, we calculated that a summer-born boy who is in the bottom third of the national height distribution will be, on average, the fourth-shortest child in a class of 24. If he is redshirted, the height he gains during the additional year of preschool will move him closer to the middle of the pack. On the other hand, if the boy enters on time, he will also tend to gain relative height over the years, and by 3rd grade the odds are only fifty-fifty that he will remain in the shortest third of his class.

Emotional development presents a thornier issue. Parents correctly want their child to be able to walk into his kindergarten classroom with confidence—standing tall, asking questions, developing relationships with the teacher and with other students. If the child’s work habits or ability to sustain attention or fine motor skills are less advanced than those of his peers because he is younger, parents fear that the child will be left behind. One challenge here is that the child’s skill level in April, when the redshirting decision is often made, is a poor predictor of what his skills will be in September or October, when the school year is underway. Children’s development is highly uneven, with bursts of improvement in language, fine motor skills, and other capacities coming somewhat unpredictably.

Who Is Redshirted? 

Most parents wrestling with these issues ultimately decide to enroll their child on schedule, as the vast majority of children are not redshirted. Among parents of the kindergarten class that entered in fall 2010, 6.2 percent reported that they delayed their child’s school entry by a year, and the share was slightly higher for boys (7.2 percent) than for girls (5.2 percent, see Figure 1a). The rate was higher among highly educated parents (see Figure 1b), with college graduates approximately twice as likely to redshirt their sons as high-school graduates are. The variation by parental education was especially stark for boys born during the summer months: among those with college-educated parents, approximately one in five was redshirted, a rate that is about four times as high as that for summer-born boys with high-school-educated parents. Note, however, that even among summer-birthday boys with college-educated parents, the great majority of them enter kindergarten on time.

The Research

In his analysis, Gladwell overstates the benefits of redshirting to some degree. In fact, a balanced look at the research suggests that while children derive a short-term gain from being redshirted, that advantage dissipates quickly over time.

It is difficult to study the impacts of redshirting because students who are redshirted differ across a host of dimensions from those who start on time. As noted, children of more-educated parents are more likely to be redshirted; separating out the effects of the delayed school entry from those of other characteristics, such as family background, presents a challenge.

No one has conducted a true randomized trial related to redshirting. Instead, researchers have sought out opportunities to isolate the effects on student outcomes of the two variables that by definition change when a student is redshirted: age itself and the student’s age relative to those of classroom peers. Once these effects are known, one can simulate the impact of being redshirted by statistically aging a kindergarten entrant by one year, and predicting the impacts of absolute age and of relative age on his outcomes.

The research literature includes many serious attempts to estimate the impact of being one of the oldest students in a class or grade, using variation in students’ age or relative age that is driven by external factors. For example, a study by Todd Elder and Darren Lubotsky leverages cross-state differences in birthday cutoff dates for kindergarten entry. In some states, a child must turn five by December 1 to be eligible for kindergarten in a given year; in others, the cutoff date is September 1. In states with earlier cutoff dates, eligible children who enter on time (and not a year late or early) are, on average, older than their counterparts in states with later cutoffs. These differences in state policy allow researchers to estimate the impact of the child’s age at kindergarten entry.

Another study, co-authored by Elizabeth Cascio and Diane Schanzenbach, uses data from the well-known Project STAR experiment in which students were randomly assigned to classrooms prior to kindergarten entry. Project STAR was initially designed to study the effects of reductions in class size. The random assignment of students to classrooms, however, meant that pairs of children with the same birthday fell into different positions in their classroom age distribution by the luck of the draw.

Both studies find that the benefit of being older at the start of kindergarten declines sharply as children move through the school grades. In the early grades, an older child will tend to perform better on standardized tests than his younger peers simply by virtue of being older. This makes perfect sense—a redshirted kindergartner has been alive up to 20 percent longer than his on-time counterpart, which means his brain has had more time to develop and he has had that many more bedtime stories, puzzles, and family outings from which to build his general knowledge. This initial advantage in academic achievement dissipates sharply over time, however, and appears to vanish by high school when, as a 9th grader, the redshirted student is at most 7 percent older than his peers.

One benefit that redshirting might indeed confer has to do with grade retention and special education placement. Statistically, older children are less likely to be retained in a grade or to be diagnosed with learning disabilities such as ADHD. This may be because schools make judgments about retention and referrals based on a student’s relative achievement within a grade, and by virtue of their age, older students are less likely to have very low achievement. Most parents who are considering redshirting, however, have children who are not likely to perform at levels that would put them at risk for grade retention; thus, we would argue that the slightly decreased probability of retention afforded by redshirting should in most cases be given relatively little weight.

The Influence of Peers

Both of us have stories of children who were redshirted and would likely have had a better school experience if they had enrolled on time. Larson tells the story of Joshua, a preschooler with a spring birthday who was on the low end of the normal developmental range in terms of work habits: he had trouble sitting still during circle time, for instance, and finishing multi-step projects. His parents decided to hold him back and give him an extra year of preschool. By fall, though, he had matured tremendously and clearly would have been flourishing in the kindergarten classroom. The following year, when he entered kindergarten at age six, Joshua was well ahead of his classmates and was often bored in class. Over subsequent years, he became demotivated and even developed behavioral problems. He was physically and emotionally more mature than his younger classmates and had trouble making friends.

Schanzenbach teaches Sunday school to preschoolers. It turns out that her experience in teaching college students does not transfer to good classroom-management practices for preschoolers, and the room can be somewhat chaotic at times. While some children sit on the story blanket and are engaged in listening to the short Bible lesson, others—typically the high-energy boys—tend to misbehave, goof off, and need constant reminders to pay attention. One little girl who was especially small for her age, Julia, would always sit front and center, actively listening and asking questions about the lesson. After story time one Sunday, as the kids were transitioning to the next activity, Julia walked up to her teacher, got within an inch of her face (as kids will do), and demanded to know, exasperated, “What is wrong with those boys? Why can’t they just sit still and listen?”

Not 30 minutes later, over coffee, the girl’s mother mentioned to Schanzenbach that they were planning to redshirt Julia because of her small stature. Schanzenbach tried to convince the woman that her tiny, bright, self-possessed girl would do just fine despite her small size, and in fact would be worse-off being delayed, because attending school with less mature peers would frustrate her. Nevertheless, Julia’s parents decided to redshirt her. Now in 3rd grade, she is bored, and frustrated by her less-mature classmates—and is still among the shortest in her class. It is likely that Julia, with her sharp intellectual curiosity and mature self-possession even at age five, would have been better-off starting school on time.

What do Joshua and Julia have in common? Because they were redshirted, they were both matched with younger peers. As the research literature confirms, the peer composition of a classroom is very important: not surprisingly, children benefit from being in a class with well-behaved, high-achieving children, and are harmed by the presence of poorly behaved classmates. One recent study by Scott Carrell, Mark Hoekstra, and Elira Kuka was able to measure the lasting detrimental impact of being in class with a disruptive peer in elementary school, suggesting that the presence of just one disruptive student out of 25 reduces the earnings of other students in young adulthood by a modest but measurable 3 to 4 percent.

Consistent with this evidence, the research on relative age indicates that being among the youngest in the class has benefits, in both the short and long term. Why? Because older classmates tend to be higher achieving and better behaved. They model positive behavior, and the younger students achieve greater academic gains from learning and competing with older ones. And two studies cited above—the one by Elder and Lubotsky and the one by Cascio and Schanzenbach—find that, with age held constant, learning with older classmates boosts students’ test scores.

For the older students, on the other hand, the positive impacts of being more mature are offset by the negative effects of attending class with younger students.

This is not to suggest that students would benefit from being in classrooms with far older children. The studies finding positive effects from older peers are based on classrooms where the age variation is typical—usually a few months. These results cannot be extrapolated to situations where a child is learning alongside much older students.

But one implication is that if a handful of families in a school decide to redshirt their children, they may be doing the other families a favor by improving the children’s peer group. Rather than feeling pressure to follow in their footsteps, parents may want to thank them.

In the Balance 

Looking at the evidence, we advise parents to redshirt their child only in unique circumstances. Just what are those situations? Here we have less research to draw upon, but experience suggests a few scenarios. One is extreme developmental delay, outside of the normal range, to such an extent that another year’s development will potentially put the child in range of his classmates. Another is when a child is experiencing trauma, such as having a terminally ill parent or sibling.

Many instances of redshirting, however, involve parents who are trying to gain an advantage for their child down the line. Parents sometimes ask us whether redshirting their child will make him more likely to be accepted into gifted and talented programs. Although we could find no direct evidence in the research literature that this is the case, we can extrapolate from the literature that older children may indeed be more likely to be placed in a gifted and talented program. This depends on how schools assign students to such programs, in particular, whether a child’s eligibility is based on a comparison to other children in his grade or to other children his age. With variations according to individual talent, children’s test scores increase both as they get older and as they experience more years of schooling. So, within a given kindergarten classroom, a six-year-old will, on average, score higher than a five-year-old by virtue of being older. And a six-year-old in 1st grade will, on average, outscore a six-year-old in kindergarten, because the 1st grader has had an extra year of schooling.

If qualification for the gifted program is limited to the top 10 percent of the 1st graders, then being older may give some children an edge. (We are quick to add, though, that age tends to convey only small advantages that are likely to be trumped by differences in innate academic talent, and that there are many young-for-grade students in gifted programs.) On the other hand, if qualification is based on a test that is normed by age—and therefore six-year-olds are compared to other six-year-olds no matter what grade they are in—then the advantage may tip to the relatively younger child who has been in school longer.

And finally, parents should bear in mind that redshirting can even have an effect on their child’s economic future. Starting school at age 6 instead of age 5 means heading off to college at age 19 instead of age 18. (Perhaps spending an extra year with a teenager in the house is another cost to consider!) Then the student graduates from college and enters the workforce at age 23 instead of age 22. The redshirted individual will ultimately spend one less year in the labor force and forgo the returns of an extra year of experience throughout his working life. Assuming that, as research seems to indicate, being redshirted has no net long-term impacts on skill level, we can estimate the cost of losing that year in the labor force for a college-educated male who retires at age 67. Over the course of the worker’s career, working full time and year-round, he can expect to earn $80,000 less.

In sum, we find that redshirting at the kindergarten level bestows few benefits and exacts some substantial costs. Both research and experience suggest that the gains that accrue from being an older student are likely to be short-lived. Because of the important role of classroom peer effects, redshirted children can be educationally and socially harmed by being with others who are performing and behaving at lower developmental levels. Furthermore, while it is hard to predict a child’s likely growth trajectory in the months prior to his expected school entry, the perceived developmental delays and immaturity that prompt parents to choose redshirting in the spring have often resolved themselves by fall.

As any good educator will tell you, parents know their children best. There are multiple factors to weigh when deciding whether a preschooler is ready to thrive in kindergarten. What research tells us about the “average” child or “most” children may not apply to a particular son’s or daughter’s unique abilities and delays. Families might want to speak with parents of older children about their own kindergarten-enrollment decisions and how they view them in retrospect. And then, parents should follow their own best judgment.

Diane Whitmore Schanzenbach is professor of education and social policy at Northwestern University and a research associate at the National Bureau of Economic Research. Stephanie Howard Larson is the director of Rose Hall Montessori School in Wilmette, Illinois. 

The Case for Contention
by Jonathan Zimmerman and Emily Robertson
The University of Chicago Press, 2017, $22.50 (paper), $68 (cloth); 144 pages.

As reviewed by David Steiner

A people who mean to be their own Governors must arm themselves with the power which knowledge gives.

—James Madison (1822)

In Immanuel Kant’s famous essay, “What Is Enlightenment?” the 18th-century philosopher challenged Western societies to display some courage and “dare to know.” Kant and his contemporaries took it as fundamental that a dedication to understanding, to unfettered empirical inquiry, and to moral reasoning would become the engine of human education and advancement. The so-called Enlightenment Project persisted throughout much of the 20th century; in the 1980s and ’90s, social theorists still maintained that respectful open dialogue on foundational matters would advance truth and strengthen democracy.

In regard to the American public school classroom, Jonathan Zimmerman and Emily Robertson, authors of The Case for Contention, place themselves squarely in this tradition:

Discussion of controversial issues helps students develop an array of skills and dispositions embodying the ability to formulate and evaluate arguments, thus fostering their capacities for rational thought and action. . . . Leading students in discussion of controversial issues can be regarded as essential to effective and efficient education.

The book is divided into two parts. The first is a rapid romp through the history of the teaching of controversies in American education. Here, the basic lesson is that such teaching has always been difficult—constrained by political forces, legal challenges, parental objections, limited class time, and understandably hesitant teachers who are ill-prepared for the complex task of teaching divisive issues.

Among these pressures, it is perhaps the courts that have done the most to limit teachers’ willingness to take risks. Despite the U.S. Supreme Court’s finding in Tinker v. Des Moines (1969) that “it can hardly be argued that either students or teachers shed their constitutional rights to freedom of speech or expression at the schoolhouse gate,” the court would later hold in Garcetti v. Ceballos (2006) that teachers’ speech is part of their work for their employers—the public school board—and thus belongs to the board. Post-Garcetti, the U.S. Court of Appeals for the Seventh Circuit found that “the school system does not ‘regulate’ teachers’ speech as much as it hires that speech.” Zimmerman and Robertson approvingly cite the “stirring dissent” by Colorado judge Gregory Hobbs: “When we strip teachers of their professional judgment, we forfeit the educational vitality we prize. . . . When we quell controversy for the sake of congeniality, we deprive democracy of its mentors.”

The book’s historical survey has its arresting moments. I didn’t know that Horace Mann, as he worked to magnify the reach of public education, was intolerant of introducing controversial subjects in the classroom: “If the day ever arrives when the school room shall become a cauldron for the fermentation of all the hot and virulent opinions, in politics and religion, that now agitate our community, that day the fate of our glorious public school system will be sealed, and speedy ruin will overwhelm it.” We surely owe to such sentiments the often-tame quality of so many textbooks and curricula in the United States. But the history summary is too brief to be systematic: it may be important, for instance, that in a 1953 survey, Ohio teachers reported “teaching the controversies over the federal take-over of the Steel Mills, the Firing of General Douglas MacArthur, and the use of the Atomic Bomb,” but the reader cannot judge if this instruction was unusual, or geographically limited, or widespread across the nation. At best, this part of the book offers general support for the authors’ claim that “we simply do not trust our teachers to engage students on controversial issues in a knowledgeable and sensitive manner.”

The second part of the book attempts to define what that trust could mean if we took it more seriously. What constitutes a controversy worth teaching, and how should educators approach teaching it? Here, we find the book’s principal contribution: to argue for a distinction between topics on which strong disagreements divide the public but “expert” opinion is largely settled, and those on which both public and experts’ judgments diverge. Regarding the former kind of issue (evolution and global warming are offered as examples), the authors argue that teachers should not remain neutral but rather teach specific respect for expert judgments and general respect for the expert “epistemologies” that support those judgments:

When the controversy is one between experts and portions of the public, the teacher . . . has an obligation to let the students know the settled judgment of those qualified to investigate the issue.

In regard to topics where neither experts nor the public have reached consensus, the authors “agree that it is important for teachers to ‘teach the controversy’ rather than directing students toward a particular conclusion.” They go further, suggesting that “parents may legitimately ask that the schools represent their side of the issue.”

So for issues such as abortion, the assumption is that teachers would need to take on the role of neutral discussion facilitator. It is striking, however, that abortion is mentioned only twice in the whole book and is not the subject of direct discussion. The death penalty is referred to once, with no analysis. In short, the book is light on discussion of some of the most obviously controversial ethical issues of our time. Moreover, the authors punt even on issues they do address: on white privilege, we are told that the issue of whether “the concept of white privilege [should] be taught in schools . . . may itself be a maximally controversial question.” Note their use of the word “may” and the fact that the question is left unanswered. In general, if the issue is judged by teachers to be too “hot” in their particular communities, the authors conclude that “avoidance could be a reasonable strategy.”

The endless raising of unanswered questions, the vague explanations of what makes an issue “maximally controversial” (who exactly gets to decide how much intra-expert disagreement gives an issue that standing?), and the generally anodyne quality of many of the conclusions (“We acknowledge . . . that deciding how to classify a given issue can sometimes be difficult”) combine to leave the reader feeling undernourished by both the theoretical and the practical content that is served up.

There is a sense, however, of worrying about deck chairs on the Titanic. Recent measures of academic achievement in the United States suggest that the Enlightenment Project is in trouble: our high-school students’ performance is largely unchanged since the 1990s, and students in many other countries have leapfrogged over them. According to a 2013 analysis by Kei Kawashima-Ginsberg of Tufts University, teaching the controversies does indeed appear to increase students’ civic knowledge; perhaps such discussions allow teachers to raise the energy level in the classroom and thus better capture the attention of students. But ours is an age in which teachers are expected to teach critical thinking about nothing in particular. It is tough to “dare to know” when any canonical knowledge itself is disdained. Teaching the controversies may indeed be one important way in which to engage students with knowledge. Any contribution to that end is most welcome.

David Steiner is professor and executive director of the Institute for Education Policy at the Johns Hopkins School of Education.

It’s 10 o’clock on a Wednesday morning at New York City’s Richard Rodgers Theatre, and the buzz in the house is palpable. The matinee of Hamilton, the acclaimed musical about the “10-dollar founding father without a father,” won’t start for four hours, but the audience of 11th-grade students is ready. While most of the play’s actors aren’t even in the building yet, the show is about to begin.

Backstage a group of about 35 high schoolers take selfies, warm up their voices, and pace. They are getting ready to mount the same stage where the play that won 11 Tony Awards is presented eight times a week. In a flash, Okieriete Onaodowan, the actor who plays Hercules Mulligan and James Madison, appears in the wings.

“EduHam” is on.

Long before the play about the nation’s first treasury secretary opened in February 2015, its creator, writer, and star Lin-Manuel Miranda knew his work would be useful for teachers. That’s because before Hamilton won a Pulitzer, a Grammy, and all those Tonys, before its cast album was streamed a half billion times on Spotify alone, before the top ticket prices soared to a Broadway record $849, Miranda debuted the show’s first song, “Alexander Hamilton,” at the White House Poetry Jam in 2009.

When video of the four-and-a-half-minute performance hit YouTube, the number-one comment posted was, “My teacher showed us this in APUSH,” Miranda told Newsweek, using the acronym for the AP U.S. History class.

So six years later, when the play first opened at the Public Theater, Hamilton biographer Ron Chernow invited Lesley S. Herrmann, then executive director of the Gilder Lehrman Institute of American History, to be his guest at the show. The two had common interests. It was Chernow’s 818-page biography of Hamilton that had kindled Miranda’s interest in the material and that became the inspiration for the hit musical. Gilder Lehrman, based in New York City, is a nonprofit dedicated to supporting history education.

As soon as the stage went dark that night, Herrmann turned to Chernow and said, “We have to get this into the hands of kids.”

Students Onstage

Fast-forward to that Wednesday morning at the Richard Rodgers Theatre, in November 2016.

As the students look over their material and emcee Onaodowan goes onstage to welcome the audience, Lulu Rivera stands by himself. The Harvey Milk High School junior will perform a spoken-word piece about Hamilton’s best friend, John Laurens. Rivera based his piece on the long, intimate letters the two wrote to each other. Clad spectacularly in black and wearing a pair of platform boots, Rivera says, “I tried to dress like history, but mostly like Prince.”

During his performance, he’ll earn solid applause from the 1,300 students in the audience. But the most raucous reception will be saved for the rock-’em, sock-’em performance of Traci Ann Pauda and Madeline Ferraris from the Brooklyn School for Music and Theatre. The duo portray Hamilton’s wife, Elizabeth (Eliza) Schuyler, and his mistress, Maria Reynolds, locked in a fight over their man, swapping insults and even shoves as the crowd hoots its approval.

Later in the program, after another group of students trade invectives in a rap battle, Onaodowan comes onstage and jokes to the audience, “That hurt my feelings. I know they were acting, but …”

The student performances are the capstone of their participation in EduHam—officially known as the Hamilton Project—an in-class history curriculum based on Hamilton. After their presentations, the students will participate in a “talkback” session with members of Hamilton’s cast, break for lunch, and then return to watch the matinee performance of the play.

During 2016–17, its pilot year in the New York Public Schools, the education program has become almost as successful as the play itself. As it wraps up its first year, the program will bring 20,000 11th graders to the 46th Street theater to perform and to see the play; that’s one out of every four juniors in the nation’s largest school district. EduHam expanded to Chicago in February 2017, and it followed the touring company when the play opened in San Francisco in March.

The total number of students served at the end of the school year will be more than 35,000, says Sasha Rolon Pereira, the director of Gilder Lehrman’s Hamilton Project.

“In the next four years, we will reach 100,000 kids in 50 to 100 cities,” adds James G. Basker, Gilder Lehrman’s president.

If the results on the National Assessment of Educational Progress are any indication, the teaching of American history can use this leg up. Although the average student score in U.S. history climbed a bit between 2001 to 2014, it remains well below the level indicating proficiency. Only about 10 percent of students passed that benchmark.

The Creation of EduHam

So how did the hottest show on Broadway team up with a nonprofit organization and a major foundation in an attempt to reinvent how American history is taught—and motivate 16-year-olds to interact with primary documents from 240 years ago?

Unlike the writing of the play, which took Miranda and his co-creators the better part of seven years, the partnership between Hamilton, Gilder Lehrman, and the Rockefeller Foundation was a whirlwind project, going from idea to fully realized program in less than a year. But in a sense, the seeds of the project were sown much earlier.

Miranda knew how empowering it feels for a young person to create his own artistic project; indeed, that’s how he got his start in musical theater. He wrote three original songs when he was in 8th grade to help teach classmates the content of The Chosen, a novel by Chaim Potok set in 1940s Brooklyn. “My first musical I ever wrote was a class assignment,” Miranda revealed to Arrive magazine.

Hamilton producer Jeffrey Seller himself has a history of bringing Broadway to high school students. He created an educational program for the 1990s musical Rent, his first theatrical success.

And Gilder Lehrman has a long track record of developing history programs that benefit schools. Indeed, two-thirds of the students who take AP U.S. History visit the institute’s web site, and its total traffic jumped to 10 million visitors last year, up from fewer than 2 million two years ago. So when Seller and Miranda’s father, Luis Miranda Jr., visited Gilder Lehrman’s 45th Street office last summer, the institute’s director of education, Tim Bailey, showed them a recent curriculum program he had written called Vietnam in Verse. The lesson plan used poetry and music from the 1960s and ’70s to address the issues of that era. Seller was impressed: “You’re in,” he told Bailey. A partnership was born.

Working through the finances to create the education program was a complicated task. The first hurdle was to get the production to discount all seats for the student matinees. About 1,100 of the Richard Rodgers Theatre’s 1,321 seats cost between $179 and $199 apiece for a performance of Hamilton. The 200 or so center-orchestra seats fetch $849, by far the highest ticket price on Broadway. The play, which nets close to $2 million per week, is sold out until November 2017. The play’s principals agreed to sell the tickets for student matinees for $70, essentially the breakeven price point.

In October 2015, the Rockefeller Foundation put up $1.5 million to pay for Gilder Lehrman to create the curriculum and to subsidize $60 of each student ticket. Students pay the remaining $10 (a “Hamilton”) for each ticket, so they’re invested (except in San Francisco, where students attended for free because of a strict state law that prohibits them from paying for any educational experiences).

“Works like this don’t come around very often, and when they do we must make every effort to maximize their reach,” says Judith Rodin, former president of the Rockefeller Foundation. “Here’s a story that talks about American history and the ideals of American democracy . . . in a vernacular that speaks to young people, written by a product of New York public education,” Rodin told the New York Times. “Could there possibly be a better combination in terms of speaking to students?”

In June 2016 the foundation upped its commitment to $6 million to fund year two of EduHam in New York City and extend it to Chicago and the touring company.

Teachers have applauded the project. “One thing we always try to do as social studies educators is to make content from the past relevant and meaningful to students today,” says Stephen LaMorte, president of the New York State Council for the Social Studies. “It’s exciting to be able to tie important historical events like the founding of our country to such a high-profile piece of popular culture.”

Popular culture has been used for decades to help teach history and other topics. Lawrence Paska, executive director of the National Council for the Social Studies, mentions films such as Glory and JFK. Teachwithmovies.org is a popular site that offers lesson plans for using more than 400 films in the classroom, from the Oscar-winning Spotlight to documentaries such as Super Size Me to the 1972 musical 1776.

The Curriculum

When Gilder Lehrman’s Tim Bailey started working on the Hamilton Project in late 2015, he knew he wanted to have students deal directly with primary sources. Gilder Lehrman owns 60,000 documents from American history, and Bailey recognized the value of reading and responding to these original materials. Secondary sources that merely summarize such documents and the events behind them tend to simplify the subject and rob students of the opportunity to analyze and interpret them for themselves. And the Common Core State Standards call for increased use of primary documents. But Bailey knew that asking students to read documents written more than 200 years ago could prompt lots of eye rolling.

Gilder Lehrman’s Basker puts it more succinctly: teaching the founding of the country can be the “castor oil of education,” he quips.

Bailey says that to capture kids’ interest in primary materials, “We have to teach the students the skills to unlock those sources. We provide enough structure so that students won’t freak out.”

Bailey’s 24-page study guide attempts to do just that. It starts by laying out the timeline of Hamilton’s life and his era, from his birth in the Caribbean to his death in a duel with Aaron Burr in New Jersey. Students do a close reading of two documents, loyalist Samuel Seabury’s Free Thoughts, on the Proceedings of the Continental Congress and Hamilton’s A Full Vindication of the Measures of the Congress. The guide instructs students to select key words from the excerpts and summarize the readings in language the authors might have used. Then students restate each excerpt in their own words.

In the next segment of the study guide, Bailey has students mine the excerpts from Seabury and Hamilton to discern exactly where each line in the song “Farmer Refuted” originated, teaching them how Miranda transformed facts into verse.

When teachers use films and plays in their instruction, Paska notes, it’s important for them to ask guiding questions so students don’t take fictional portrayals as gospel.

In the companion book to the musical, Hamilton: The Revolution, Miranda adds notes to each of the play’s 40-plus songs, often coming clean about historical inaccuracies. For instance, he writes that while the play depicts the three Schuyler sisters as having no brothers, in reality they had “loads” of them. Miranda leaned on artistic license here because he wanted to portray Angelica Schuyler as a world-class thinker in a society that didn’t accept women as intellectuals, and he thought it would be more effective to omit the four brothers.

Squeezing the details of someone’s life into two and a half hours almost always involves shortcuts. Miranda has Hamilton meet Laurens, Mulligan, and the Marquis de Lafayette in a bar on a single occasion. In fact, Hamilton met the three men separately over a number of years. But the early song “My Shot” propels the play’s action forward and establishes how the four characters will work together against England.

Multimedia Materials

While Bailey worked on the classroom materials, colleagues set up a private Web portal where students could see excerpts from five songs performed during the show. The play’s creators insisted on limiting how much of the piece they would expose, for fear of diluting the play’s potential earnings on tour. Still, says Bailey, “We have amazing access to the show. It’s unprecedented.”

On the web site, students can view nine video interviews created exclusively for them. The videos feature Miranda explaining how Hamilton was different from other Founding Fathers, Chernow discussing the artistic license used in nonfiction writing, and actors reading from original documents of the period.

In one video, Miranda holds an actual love letter from Hamilton to his future wife Eliza and reads: “You not only employ my mind all day; but you intrude upon my sleep. I meet you in every dream and when I wake, I cannot close my eyes again for ruminating on your sweetness.” He looks up and tells students, “This puts whatever R&B song you’re listening to right now to shame.”

The web site also features information on 30 different historical figures, ranging from Martha Washington to Hercules Mulligan, the tailor who used his access to British troops to spy for the patriots. The site highlights 14 key events from the era, as well as 20-plus documents, including The Federalist Papers and Thomas Paine’s Common Sense.

While EduHam’s materials are robust, the program requires only two to three class periods to complete, Bailey says. Most of the student work, such as the suggested three hours of rehearsal, takes place outside the classroom. The program includes an 11-page teacher guide that discusses objectives, procedures, and the four Common Core standards the lessons align with. There is also a rubric to guide teachers in assessing student work.

Students are given wide latitude as to what, and how, they perform in EduHam. They can present a rap, song, poem, monologue, or scene. And while their performance must represent the Revolutionary War era, they can choose from key people, events, or documents, even if they aren’t in the play. During the November 2016 performances, one group of students compared the struggle between America and Britain to the Crips–Bloods gang battles in California. Students’ reactions were so enthusiastic it was hard to hear the end of the performance. One girl recited poetry about the African American poet Phillis Wheatley, who isn’t in the play, and another reworked the rapper Drake’s piece “5AM in Toronto” to depict the Boston Massacre.

Extending the Impact

According to an internal survey at Gilder Lehrman, 70 percent of teachers said they would continue to use the curriculum even if their future students couldn’t see the play. With the soundtrack, various performances from the White House and television, and a 90-minute PBS documentary about the making of the musical, there is plenty of material that teachers can access to use Hamilton in the history classroom.

Indeed, before EduHam was available, teachers had already been using some of these materials. Jim Cullen at the Ethical Culture Fieldston School in the Bronx created an entire course about Alexander Hamilton and the play. He took students to a performance of Hamilton at the Public Theater before the show hit Broadway. One California teacher, Angelica Davis, used the music and clips of performances to teach her students about both the play and the Founding Father.

While the program still hasn’t completed its first year, “We’ve had nothing but praise from students, teachers, and administrators,” says Bailey.

Looking forward, Miranda has mentioned an even bigger potential impact. When the play is approved for school productions, likely in five years or more, students will have a chance to play these roles themselves. Theater licensing agencies estimate the play could be produced at more than 600 schools during a typical year.

Using Diversity

Hamilton’s multicultural cast is integral to the production, and it makes sense that this diversity would also inform how students learn about both the historical period and the play. Students attending the performance so far come from low-income high schools and most are students of color. (San Francisco is an exception. The city has so few Title I schools that the program will bring in students from Oakland to participate.)

Mystic Morrison, a student from Brooklyn, says the cast’s racial composition reflects “real life. It’s nice to see not just one race [represented] onstage.”

Adds Joe White, a teacher at the John Adams High School in Queens: “The play is in language they understand.”

Miranda himself, who is of Puerto Rican descent, said during one of the talkbacks after a student performance, “When it comes to diversity on Broadway, it’s a prerequisite for what I do.” But he also mentioned a specific benefit of employing a multicultural cast. “I wanted to eliminate any distance between the story that happened 200-some-odd years ago and this audience of today. By using music of today and a set of actors that looks like our country, we eliminate that distance and see the whole story with a fresh set of eyes.”

Including Controversy

Of course, not all the drama around Hamilton has occurred onstage. At a performance in late November 2016, the cast addressed Vice President Elect Mike Pence, who was in the audience.

Actor Brandon Victor Dixon, who played Aaron Burr in that performance, told Pence: “We, sir, are the diverse America who are alarmed and anxious that your new administration will not protect us, our planet, our children, our parents, or defend us and uphold our inalienable rights. We truly hope that this show has inspired you to uphold our American values and to work on behalf of all of us.”

Miranda, who is prolific on Twitter, is not shy about trumpeting his political views, which lean decisively to the left.

When asked if this controversy would make schools less likely to use the play as a learning tool, Lawrence Paska says: “That’s going to depend on local school curriculum choices and planning. Some teachers and schools may use recent events as a way to highlight the intersection of history, art, and current events. Others may choose not to use works like Hamilton because they want to focus on historical events and not on recent activism.”

Praise from On High

No one involved with the play or the education program thinks this work is the be all and end all of U.S. history education.

“Hamilton isn’t the endpoint for these kids’ education, it’s the jumping-off point,” Miranda has said. “I hope it encourages people to learn more about their history. I never really cared much about American history because I didn’t see any connection between [that history] and the world I lived in.”

“It’s a story for all of us, about all of us,” former president Barack Obama has said.

When the cast performed at the White House in spring 2016, Obama praised the educational component of the program, saying, “I’m thrilled they are working with New York public schools. There’s now a curriculum to give students context and a deeper meaning—or deeper understanding of our nation’s founding. I hope this helps every teacher who spent hours trying to make The Federalist Papers teenager-friendly. The remarkable life of Alexander Hamilton will show our young people the possibilities within themselves and how much they can achieve within the span of a lifetime.”

As Miranda tells students when they come to see the play, the big question is, “What kind of world do we want to create? It’s no less than that. What kind of world are you going to create when you grow up?”

Wayne D’Orio is an award-winning education editor and writer who has covered education topics for more than a decade. 

Since the 2001 passage of the No Child Left Behind Act, test-based accountability has been an organizing principle—perhaps the organizing principle—of efforts to improve American schools. But lately, accountability has been under fire from many critics, including Common Core opponents and those calling for more multifaceted measures of teacher and school performance. And yet the Every Student Succeeds Act, NCLB’s successor law, still mandates standardized testing of students and requires states to have accountability systems. So: is accountability on the wane, or is it here to stay? If accountability is indeed dying, would its loss be good or bad for students?

In this issue’s forum, we present three different viewpoints on those questions from Morgan S. Polikoff, associate professor of education at the University of Southern California’s Rossier School of Education; Jay P. Greene, professor of education at the University of Arkansas; and Kevin Huffman, former Tennessee commissioner of education.

 

 

Why Accountability Matters, and Why It Must Evolve
by Morgan S. Polikoff

 

 

 

 

Futile Accountability Systems Should Be Abandoned
by Jay P. Greene

 

 

 

 

If Parents Push for It, Accountability Can Work
by Kevin Huffman

 

Try to think of an education policy that 1) has been shown, in dozens of studies across multiple decades, to positively affect student outcomes; 2) has the overwhelming support of parents and voters; 3) reinforces many other policies and facilitates quality research; and 4) has been used widely at the district, state, and national levels for decades or more.

You might be thinking that such a policy doesn’t exist, and if it did, we’d surely want to keep it around. But the truth is precisely the opposite. Such a policy does exist—it’s called school accountability—yet the powers that be seem increasingly ready to throw it out and leave education to the whims of the all-but-unregulated free market.

School accountability, specifically test-based accountability, has been a staple of K–12 education policy since the 1990s (and even before that, in some states and districts). Over that time, we’ve learned quite a lot about it.

First, we’ve learned that it can work. We’ve seen this in studies of individual districts, individual states, and the nation as a whole: David Figlio and Susanna Loeb’s 2011 review of research summarizes this literature comprehensively. The effects observed in many studies are substantial, especially given that they typically occur schoolwide. The effect of No Child Left Behind (NCLB) on students’ mathematics achievement documented by Thomas Dee and Brian Jacob and confirmed by Manyee Wong and colleagues is equivalent to the gain from spending three or four years in an average urban charter school, according to the latest data from Stanford University’s Center for Research on Education Outcomes. Accountability doesn’t seem to do a great job at closing achievement gaps (though it certainly shines a light on underperformance), but there’s considerable evidence that it can raise student achievement.

Second, we’ve learned that parents and voters feel strongly that accountability is essential. Polls show overwhelming bipartisan support for the common-sense idea that schools receiving public dollars to educate children should be accountable for providing a good education. Education Next’s 2016 poll reported at least two-thirds support for annual testing among both Republicans and Democrats. In the 2016 PACE/USC Rossier poll of Californians that I led, we asked what schools should be held accountable for; voters rated standardized test results last among the options presented, but 69 percent of them still believed accountability for test results was important. We also know that parents prioritize student achievement when selecting a school for their children. In our increasingly resource-constrained and globally competitive world, this desire for outcomes will only intensify.

Third, we’ve seen that accountability mutually reinforces other policies and provides essential data to support education research and improvement. For instance, there is suggestive evidence that charter schools perform better in contexts where accountability is high (that is, where strong authorizing laws shut down poorly performing schools) than where it is weak or nonexistent. Accountability was intended to provide weight to state standards and encourage teachers to implement them, and evidence suggests it does focus teachers’ attention on the content that state policymakers want teachers to emphasize. Not to mention that the data emerging from the same tests used for school accountability have powered a revolution in education research that has allowed scholars to dramatically improve the relevance and rigor of their work.

Finally, we’ve learned a lot about how to design accountability policy to better target the schools that most need improvement. It is now generally understood that the simplest performance measures—those that defined test-based accountability under NCLB—mainly tell you who’s enrolling in a school, not how well the school is educating those students. We know that performance indexes and growth measures are much fairer and more accurate ways to classify school performance. There’s also a growing consensus that in the next generation of accountability policies, we must broaden the criteria beyond test scores, and the new federal education law, the Every Student Succeeds Act (ESSA), encourages this kind of creative rethinking.

We’ve also learned that the design of accountability policies can affect the way teachers respond to them. For instance, policies that focus attention on raising the achievement of low-performing students may be more effective than those that offer rewards for high student performance in general. And teachers do seem to respond rationally to accountability policies by focusing more on the grades and subjects that are tested. As for concerns about NCLB’s negative impact on teacher working conditions, Jason Grissom and his colleagues have shown that the law’s implementation did not diminish teachers’ job satisfaction or increase their levels of stress. While the unintended consequences of accountability can be pernicious, they can also be addressed, at least in part, through policy design.

Countering the Opposition

Despite this track record of modest success, many parties seem poised to throw the policy overboard and use the guise of “parental choice” or “local control” to return us to a time when we had little idea which schools were educating children well and which were not. The opposition to accountability in education is largely political; my 2016 analysis of California poll data, for instance, found that disapproval of President Obama was among the strongest predictors of Common Core opposition, and Education Next and others have routinely found that voters support “common standards” or “common assessments” when they are not tied to the Common Core name.

There are of course more principled concerns with accountability, and it is worth taking a moment to address them. One issue is that accountability in general, and test-based accountability in particular, can have negative effects on instruction, such as a dumbed-down, narrowed curriculum. This problem can be addressed in large part by improving content standards and the assessments used to gauge student performance. Recent work I co-led with Nancy Doorey indeed finds that the two state-assessment consortia have made considerable improvements over even the best NCLB-era tests. This issue can also be addressed by broadening the set of indicators against which schools are evaluated, which many states are poised to do under the new federal accountability law.

Another concern is that the tests used for accountability do not predict important life outcomes, and thus that we might be focusing on the wrong things. To be sure, studies do not show a perfect one-to-one relationship between impacts on test scores and impacts on later life outcomes—no one expects they would. But several studies do show longer-term effects of accountability policies; we have strong evidence from Raj Chetty and colleagues that impacts on test scores do predict impacts on other important life outcomes; and, again, many states appear poised to broaden accountability measures beyond just test scores.

What Comes Next?

There is no doubt that the coalition that once supported accountability policy has frayed. The Republican leaders in the executive and legislative branches, which once championed accountability, have turned to school choice as the primary strategy to produce reform (even as public opinion on choice, especially more extreme forms such as vouchers, has begun to sour). But choice without accountability is unlikely to work. Without test results, for instance, we would not know that online and virtual charters appear to be demonstrably harmful to students, as are many Louisiana private schools attended by students using vouchers. Nor would we know that Boston’s well-regulated charter high schools produce truly stunning positive effects on students’ test scores and early college decisions. Choice programs that do not contain accountability provisions offer us zero assurances that educational dollars are being well spent.

Where should we go from here? We must continue to recognize that the design of accountability policy matters, and we must refine our policies over time. ESSA allows states to do this. It allows states to include better test-based measures of school performance, and they should. It allows them to incorporate measures of school climate, student attendance and discipline, and progress toward college and career readiness, and states should adopt and experiment with these measures. It allows them to target consequences on a smaller subset of low-performing schools and move away from NCLB-era interventions that were largely ineffective, and states appear to be focusing their efforts on more promising interventions that target growth and effective practices. Will the next round of state accountability policies be perfect? They will not. Will they be better than what they replaced? They almost certainly will.

Over the last several decades, we have made real, if incremental, progress in education. Test scores on the National Assessment of Educational Progress are up for every student subgroup (even accounting for the downward blip in 2015), and graduation rates are, too. Accountability systems have worked well with other reforms—such as effective choice policies, the expansion of early-childhood-education and other school-readiness programs, and efforts to improve the teaching force through evaluation and tenure reform—to improve education for children around the country. There is simply no reason to think that abandoning accountability at this point would be an effective strategy. The coming years will see new and creative uses of accountability in states and districts. We must encourage and study this innovation if we are to continue improving America’s public schools.

This is part of a forum on test-based accountability. For alternate takes, see “Futile Accountability Systems Should Be Abandoned,” by Jay P. Greene, or “If Parents Push for It, Accountability Can Work,” by Kevin Huffman.