Foreword

By Aaron Churchill and Michael J. Petrilli

America is wasting much of its human capital. That’s because many high-achieving, low-income students—we call them HALO students—fall off the education track and never reach the gates of our top colleges and universities. Regrettably, these capable young people won’t reap the benefits of a four-year degree, which remains a solid engine for upward mobility and career success. Their families, communities, and states won’t benefit from maximizing their capabilities. This represents a tragic loss of potential, while also leaving our best colleges and—later—our most promising occupations bereft of socioeconomic diversity and looking less like the nation writ large.

We at the Thomas B. Fordham Institute have long been troubled by this leaky pipeline from K–12 to higher education. Through various empirical studies, we have drawn attention to the disappointing outcomes of high achievers from less advantaged backgrounds. We have also worked to develop and widely share policies and practices that would help more high achievers reach their full potential.

The present study builds on those efforts to strengthen advanced education in U.S. schools, both generally and the Buckeye State particularly. We commissioned Ohio State University professor and Fordham research fellow Stéphane Lavertu to focus on the educational experiences and outcomes of HALO students in Ohio. His analysis addresses three main questions:

1. What does the K–12 education experience look like for Ohio’s HALO students?

2. How many of them attend four-year colleges, including more selective “top colleges”?[1]

3. What K–12 education factors best predict their college attendance?

To examine these questions, Dr. Lavertu used anonymous student-level records to study the trajectories of HALO students from elementary school into college (in and out of state).[2]

Key findings

This report presents irrefutable evidence that Ohio’s HALO students are underserved by most school systems, and that too many fail to reach college after high school. Compared to similarly scoring students from more affluent backgrounds, Ohio’s HALO students are:

• 26 percent less likely to take advanced math—primarily Algebra I—during middle school.

• 34 percent less likely to complete an Advanced Placement (AP), International Baccalaureate (IB), or dual-enrollment course during high school.

• 44 percent less likely to receive gifted services (e.g., accelerated coursework or enrichment) during middle or high school.

This systemic neglect has ramifications for their college-going. Figure A shows that only two in five HALO students enroll in any four-year college within two years of high school, and fewer than one in ten attends a top college. These college-going rates are far lower than those of high-achieving, high-income (HAHI) students. Specifically, Ohio’s HALO students are 42 percent less likely to attend a four-year college and 66 percent less likely to attend a top college than their similarly-scoring but more affluent peers.[3]

Figure A. HALO students attend college at much lower rates than their more affluent, similarly-scoring peers.

Underscoring how leaky this pipeline is for HALO students, consider this: While they are outnumbered 3-to-1 by their more affluent counterparts in the top achievement quintile in third grade, a decade later they are outnumbered 6-to-1 in four-year colleges and a staggering 10-to-1 in top colleges.

What would keep more HALO students on-track for college? Figure B shows that AP or IB and dual-enrollment course-taking during high school makes a difference in their college-going prospects, as do advanced math courses during middle school and gifted services in English language arts or math. For instance, HALO students who completed an AP or IB course had four-year college-going rates that are 29 percentage points higher than HALO students who did not complete such courses. These results provide compelling evidence that advanced learning opportunities are critical to keeping students on a college pathway.[4]

Figure B. HALO students who participate in advanced learning opportunities have higher four-year college attendance rates than HALO students who do not.

Policy implications

What should state and local policymakers do? Many solid recommendations were recently made by a national working group on advanced education, which can be found in a recent Fordham report. Here, let us highlight five ideas.

1. Ensure early identification of high-achieving students and provision of advanced education during elementary school. It’s hard for schools and teachers to provide advanced learning opportunities to HALO students if they don’t know who they are. Commendably, Ohio (and several other states) require universal screening in the elementary grades for gifted and talented programs (using non-state tests). Policymakers should go a step further and also require identification when students score at the highest levels on state exams—an important marker of academic ability. But identification alone won’t do students much good unless schools provide them opportunities to accelerate and advance. Here, states also have a role to play. They should require schools to provide advanced opportunities to identified students—and enforce that requirement through an audit-type process.

2. Promote automatic enrollment of high achievers in advanced middle school and high school math courses. A few states, including North Carolina and Texas, have automatic enrollment policies that guarantee placement in advanced math courses based on students’ prior-year test scores. For instance, under this type of policy, high-performing seventh graders are automatically placed in eighth grade Algebra I, which is key to more advanced math in high school. Tying course placement to objective measures of achievement ensures that HALO students are in advanced courses—far more reliable than systems that rely primarily on subjective judgments or recommendations for such placements.

3. Expand access to AP and IB coursework into more high-poverty high schools. This goal has long been on the education-reform agenda, but there’s still much to be done to bring such challenging course options into schools that serve the most HALO students (e.g., high-poverty rural or urban campuses). This can be done in many ways, including, for example, “itinerant” programs, whereby an AP teacher covers classes in multiple schools or school districts, or virtual options that bring advanced coursework into classrooms anywhere in the state. Another possibility is to offer schools or teachers financial incentives when students pass AP or IB exams, as North Carolina does. A “weighted” bonus that provides additional funds when low-income students pass these exams could further strengthen the incentive for educators to start programs in schools that have not traditionally offered advanced coursework.

4. Help more HALO students access suburban public schools. In analyses pinpointing which Ohio districts most effectively boost the college-going odds of HALO students, we find several wealthy suburban districts—Solon, Mason, and Dublin among them—topping the charts (see here). That is good news for the handful of low-income students who live in those districts. However, many suburban districts, including two of the three mentioned above, prohibit non-resident pupils from attending their schools. States should require all districts, including suburban, to participate in statewide interdistrict open enrollment, a policy that enables students to cross district boundaries to attend school (or in lieu of a mandate, provide sufficient per-pupil funding to encourage district participation). This would put suburban public schools—and the advanced opportunities they typically offer—within greater reach of HALO students residing in nearby cities.

5. Launch specialized magnet schools for high achievers. When it comes to Ohio’s best high schools for HALO students (see here for overall rankings and here for rankings focused on top colleges), this varied mix includes a handful of urban district high schools, such as Walnut Hills in Cincinnati and Cleveland School of Science and Medicine, both of which have selective admissions. Also near the top are a few non-district options, such as Dayton Early College Academy, a public charter school that Fordham authorizes, and Metro High School, an independent STEM school in Columbus. Unfortunately, such schools are rare. School districts should work to create specialized, “early college” schools for high achievers—open to students regionally—as could charter or STEM school developers. Following in the footsteps of Illinois, North Carolina, and Oklahoma, states themselves could also create, and directly fund, a residential, competitive public high school that enrolls high-achieving students from all corners of the state.

HALO students have demonstrated their potential to reach college and use it as a springboard to the American dream. Yet due to the widespread neglect of their talents, our education system allows too many of them to fall through the cracks. There is a clear pathway that can better nurture HALO students. It includes early identification and acceleration in elementary school, access to advanced math coursework starting in middle school, and enrollment in high schools with strong college-prep curricula and college-going cultures. The playbook is there. Now the question is whether states and school systems can execute and give HALO students the challenge and rigor that prepares them to compete for spots at our top universities.

 

Executive Summary

Ohio has a large store of untapped human potential. Every year, thousands of students with exceptional academic ability—many from low-income households—fall off the path to college, missing out on opportunities to improve their economic well-being and, consequently, the prosperity of their families and fellow Ohioans. This report examines Ohio’s public-school pipeline to college, particularly for high-achieving low-income (HALO) students, and identifies the characteristics of districts and high schools that are most effective at getting these students into four-year colleges and universities.

The analysis defines HALO students as those who qualify for free or reduced-price lunches and whose state test scores put them in the top 20 percent of Ohio students (as of third grade or eighth grade, depending on whether the focus is on districts or high schools). It first provides a descriptive look at these students, comparing their demographic attributes and college attendance rates with those of high-achieving peers who do not qualify for free or reduced-price lunches. The analysis then employs statistical techniques to estimate “college-going value-added” for the Ohio districts in which HALO students reside and the specific high schools (including charter and STEM schools) they attend. Those estimates capture the extent to which students in a district or school exceed (or fall short of) average rates of college attendance among students with similar characteristics. Using these value-added[5] estimates, the analysis then examines factors that might explain the relative success of some districts and high schools in getting HALO students into college. Finally, the study focuses on how taking advanced coursework in middle and high school relates to HALO students’ college-going. Below are some of the most noteworthy findings.

First, there is evidence of a “leaky” pipeline to college for HALO students:

• High-achieving low-income students are far less likely to attend college than their wealthier but similarly high-achieving peers. In third grade, HALO students are outnumbered 3-to-1 by similarly high-achieving students who are from wealthier households, but they are later outnumbered 6-to-1 at four-year colleges and 10-to-1 at top colleges (those where the average student scores in the top 20 percent on college admissions exams). Put differently, HALO students are 42 percent less likely to attend a four-year college than their wealthier high-achieving peers, and they are 66 percent less likely to attend a top college. If HALO students’ college-going rates matched those of their peers from wealthier households, 1,742 more of Ohio’s HALO students would enroll in a four-year college every year—and 1,153 of those students would enroll in a top college or university. 

• HALO students miss out on advanced coursework as they progress through school. The rate at which HALO students (identified in third grade) enroll in high school–level math courses (primarily algebra) while in middle school is 26 percent lower than it is for their wealthier peers. And the rate at which HALO students (identified in third grade) enroll in Advanced Placement (AP), International Baccalaureate (IB), or dual-credit options (as part of Ohio’s College Credit Plus program) while in high school is 34 percent lower than it is for their wealthier high-achieving peers. They are also 44 percent less likely than their wealthier peers to receive gifted services in English language arts or math in middle or high school (such as accelerated coursework or other specialized programs consistent with their gifted identification). If HALO students took advanced coursework at the same rate as their wealthier high-achieving peers, then 1,103 more HALO students in each third-grade cohort would ultimately take high school–level math courses during middle school and 1,644 more would take at least one AP course in high school. 

Second, the analysis indicates that some districts and high schools provide HALO students with a bigger boost—they provide greater “college-going value-added”—and that there are commonalities between these high value-added districts and schools:

• The academic achievement of HALO students’ peers is the strongest predictor of districts’ and high schools’ college-going value-added—far more predictive than common measures of district- and school-wide poverty rates or per-pupil spending. Going from a district where the average third-grade student tests at the fiftieth percentile to one where the average student tests at the sixtieth percentile (an increase of approximately one standard deviation) corresponds to a 3.5-percentage-point increase in the probability that a third-grade HALO student in the district will go on to attend a four-year college or university. Similarly, going from a high school where the average student tests at the fiftieth percentile to one where the average student tests at approximately the sixtieth percentile corresponds to an 8.6-percentage-point increase in the probability that a HALO student at the high school (identified in eighth grade) will go on to attend a four-year college. Consequently, the Ohio districts and high schools that provide the best pipelines to college for HALO students are predominantly (but not exclusively) located in high-achieving suburbs. (See here and here for lists of districts and high schools with the highest college-going value-added.)

• Rates of student participation in advanced coursework are the strongest predictors of whether districts and high schools get their HALO students into top colleges and universities—more predictive than peer achievement and far more predictive than common measures of district or school poverty and spending per pupil. High schools that feature greater participation in advanced coursework and gifted services (one standard deviation greater than average) are associated with an increase of 4.3 percentage points in the probability that a HALO student (identified in eighth grade) will attend a top college or university (one where the average student scored in the top 20 percent on college admissions exams). Participation in AP coursework drives this result, as opposed to rates of participation in College Credit Plus or gifted services in English language arts or math while in high school. Notably, 40 percent of the gap in AP course–taking between HALO students and wealthier high-achieving students is attributable to these students residing in different districts, as opposed to HALO students and wealthier high-achieving peers in the same district taking different coursework.

Finally, comparing HALO students who took advanced coursework or received gifted services with similar HALO students who did not, the analysis reveals major differences in college-going:

• Taking AP courses in high school is most predictive of college-going, but other dual-credit options also appear to benefit HALO students. HALO students (identified in eighth grade) who took at least one AP course in high school have a college-going rate twenty-nine percentage points greater than similar HALO students who did not take at least one AP course. Their matriculation rate for top colleges and universities is sixteen percentage points higher. The corresponding estimates for College Credit Plus (CCP) are, respectively, twenty-four and eight percentage points. Thus, HALO students who participate in CCP are substantially more likely to attend a four-year college but only half as likely as those who take AP coursework when it comes to getting into top colleges and universities.

• Taking advanced math courses in middle school is more predictive of college-going than receiving gifted services, but the results suggest that both benefit HALO students. HALO students (identified in third grade) who took at least one high school–level math course in grades six through eight (mostly algebra) have rates of college-going eighteen percentage points higher (six percentage points for top colleges) than similar students who did not take advanced math. Those who received gifted services in English language arts or math in those grades have college-going rates six percentage points higher (four percentage points for top colleges).

The results suggest that Ohio could better support its high-achieving low-income students by expanding their access to advanced coursework and high-achieving peers throughout the K–12 pipeline. Early interventions, so that students are ready to enroll in algebra by eighth grade and to enroll in AP coursework during high school, appear to be effective pathways to get HALO students into top colleges and universities. Expanding students’ schooling options, such as establishing selective-enrollment magnet schools and increasing districts’ participation in Ohio’s open enrollment program, could help families and schools overcome resource constraints that prevent them from providing HALO students with advanced coursework and access to high-achieving peers. Indeed, selective high schools located in city districts are some of Ohio’s most effective at getting HALO students into college. Offering families more help in navigating these and other school options, such as Ohio’s Educational Choice Scholarship Program (EdChoice), would likely further enhance their benefits.

 

HALO Students and College-Going: What Research Tells Us

Research indicates that the rapid growth in U.S. postsecondary education during the twentieth century increased incomes while reducing socioeconomic inequality. The college-going wage premium—the difference in income between those who attend at least one year of college and those who do not—remains substantial, but it no longer benefits low- and high-income students equally.[6] By the end of the twentieth century, the premium had increased for high-income students while shrinking for low-income students—what Zachary Bleemer and Sarah Quincy (2025) call “collegiate regressivity.”[7] Their analysis indicates that 80 percent of collegiate regressivity is due to low-income students disproportionately selecting into community and for-profit colleges, the divergence in quality between research-based universities and the teaching-focused colleges where low-income students tend to enroll, and higher-income students shifting their studies toward computer science and engineering.

That high-achieving students—whether rich or poor—might fail to attend and graduate from a high-quality postsecondary institution harms us all. The best available research indicates that human capital—individuals’ productive capacity—explains at least half of income differences across countries, that schooling is effective at developing human capital, and that today’s labor market increasingly favors higher-order thinking skills, such as problem-solving.[8] Research suggests that the college wage premium doubles throughout graduates’ careers because college is a gateway for individuals with strong cognitive skills to access nonroutine occupations associated with high wage growth.[9] Evidence also indicates that universities play an important role in delivering general economic benefits by developing students’ productive capacities and generating innovations that benefit their communities.[10] Failing to capitalize on the talents and potential of high-achieving students has an especially negative impact, as their forgone lifetime income is greater and society misses out on more economic growth.

Some barriers to HALO students’ college-going are easier to address than others. As James Coleman concluded decades ago, families’ socioeconomic circumstances have a large impact on their children’s performance in school, as does the performance of their peers.[11] For example, disruptive peers have been shown to have substantial negative impacts on students’ college-going and future earnings.[12] In a recent study using exceptionally rich data, economist Raj Chetty and his colleagues determined that kids’ economic connectedness—the extent to which low-income kids interact with higher-income kids—is a strong predictor of whether low-income kids improve their economic standing down the road.[13] They also find that schooling is a determinant of economic mobility for low-income students in part because schools facilitate connectedness between low- and high-income students.[14] Thus, a serious analysis of the postsecondary pipeline for HALO students must consider the potential impact of their residential and school peers.

By third grade—the starting point of our analysis below—socioeconomic inequalities are largely baked into student achievement.[15] And rigorous research indicates that increased spending on schools has, on average, a modest impact on student achievement and attainment.[16] Yet schooling, including how money is spent on it, matters a great deal after third grade. For example, there is strong evidence that providing AP coursework causes increased college attendance and success.[17] This is consistent with a larger body of research documenting that AP coursework and, to a lesser extent, dual-enrollment options are highly predictive of college-going.[18] Research also indicates that the entire pipeline of coursework, back to elementary school, can affect college-going.[19] For example, taking algebra in eighth grade can substantially increase the probability that students take and succeed in advanced math coursework in high school, provided that they are sufficiently high-achieving students at the end of seventh grade.[20]

Preparing HALO students to participate in this advanced coursework is a significant challenge, as one must contend with some of the broader social forces noted above. But research also demonstrates that much can be gained by simply providing options and advising students and parents. For example, recent randomized controlled trials—the “gold standard” for estimating causal effects—have demonstrated that advising low-income high-school students to enroll in higher-quality colleges and universities can significantly increase their bachelor’s degree attainment.[21] Such advising decisions work further down the pipeline, as well, such as when schools directly enroll high-achieving seventh graders into eighth-grade algebra.[22]

When schools lack the capacity to provide the requisite coursework, instructors, and advising, HALO students could benefit from options such as selective magnet schools, private schools, or schools in another district. For example, research has demonstrated that charter schools and private-school vouchers can yield substantial educational benefits for low-income students, in terms of both achievement and attainment.[23] Importantly, these studies indicate that such benefits can accrue to both the students using those options and, via competitive effects, the students who remain in district schools. Rigorous studies of selective exam schools have often found null impacts on student test scores, but recent research suggests that may be because such schools are delivering educational benefits on dimensions not captured by tests.[24] That also appears to be the case with other school options, as evidenced by a recent study documenting higher rates of college-going among students who participated in Ohio’s EdChoice private-school scholarship program than among similar students who did not participate in the program.[25]

This literature review shows several paths through which communities and schools might affect the college-going of HALO students after third grade. The analysis below considers most of the above factors to some extent when examining the district- and school-level predictors of HALO students’ pursuit of higher education at four-year colleges and universities. It explicitly accounts for broader social forces that affect both school and home life (the affluence and achievement of students’ residential district and school peers), public investments in schools (spending per pupil), access to alternative schooling options that might better match students’ needs and generate competitive effects (student enrollment in charter schools and eligibility for private-school vouchers), and school inputs such as advanced coursework and services for students identified as gifted. Importantly, the focus is on factors that lead students to attend four-year colleges and universities—especially top colleges that serve similarly talented students—as research indicates that these institutions are most likely to help HALO students reach their academic potential.

 

A Descriptive Look at Ohio’s HALO Students

The analysis begins with a look at the attributes and education trajectories of low-income students in Ohio who were high-achieving third graders between the 2008–09 and 2013–14 school years. It compares their attributes and education trajectories with those of the average Ohio student, as well as those of similarly high-achieving but wealthier students. Consistent with past Ohio research—and to ensure the sample of students is large enough to conduct the analysis that follows—“high-achieving student” is defined as one whose average score on Ohio’s third-grade tests in English language arts (ELA) and mathematics was in the top 20 percent statewide.[26] Whether students were eligible for free or reduced-price lunches during that third-grade year serves as a proxy for income to identify whether students are high-achieving low-income (HALO) or high-achieving high-income (HAHI).[27] The analysis uses third grade as the baseline because that is the earliest point of the public-school pipeline for which student achievement data are available.

Baseline Characteristics of Ohio’s HALO Students

Table 1 reveals that only 25 percent of Ohio’s high-achieving third graders are low-income (5 of the top 20 percent) and that HALO students are approximately five times more likely to be Black or Hispanic than their wealthier high-achieving peers. Put differently, by third grade, academic achievement already diverges significantly by students’ household income and race. These results are not surprising, as research has long demonstrated that education gaps by income and race have largely set in by third grade. Thus, the analysis in this report is about what benefits a HALO student’s residential district or high school might provide after family, community, and schooling environments have significantly shaped their early childhood development. It does not address how earlier interventions might boost the proportion of low-income kids who are high-achieving third graders.

Table 1. Characteristics of HALO and Other Students in Third Grade between 2008–09 and 2013–14

Table 1 also indicates that HALO students’ test scores in grade 3 are 10–20 percent of a standard deviation lower than that of their higher-income peers (1.1 and 1.2 standard deviations above the average math and ELA score, respectively, as opposed to 1.3 standard deviations above the average score). This difference in achievement might appear to explain their being identified as “gifted” (in any area) at significantly lower rates (31 percent compared with 48 percent).[28] However, as the analysis below demonstrates, the differences in baseline test scores between HALO and HAHI students are not sufficiently great to account for significant differences in these students’ current and future schooling experiences.

A Leaky Pipeline: College-Going Among Ohio’s HALO Students

Table 2 provides comparisons like those in Table 1 but focuses on measures of students’ future educational attainment. It indicates that the HALO students identified above are only somewhat less likely than their wealthier peers to receive a high-school diploma (82 percent compared with 87 percent) but are far less likely to attend college (two- or four-year) within two years of graduating high school (52 percent compared with 76 percent). HALO students’ rates of college attendance are only marginally greater than those of the average Ohio student. Note that these percentages understate college-going rates because of data limitations (for example, students exiting public schools after third grade would be identified as not having graduated or attended college), but they nonetheless illustrate the disparities between HALO students and their wealthier peers.[29]

Table 2. Educational Attainment of HALO Students and Other Students (Grade 3 Cohorts from 2008–09 to 2013–14)

Comparing overall rates of college-going understates the severity of the situation, however. HALO students have the academic potential to excel at top four-year colleges, which this study defines as those where incoming classes have average ACT and SAT scores in the top 20 percent of the Ohio score distribution.[30] Although HALO students are twenty-four percentage points less likely to attend any college than their wealthier high-achieving peers (a rate of college-going that is 31 percent lower than HAHI students’ rate), they are twenty-eight percentage points less likely to attend a four-year college (a rate that is 42 percent lower) and nineteen percentage points less likely to attend a top college (a rate that is 66 percent lower) than their wealthier peers.

In other words, HALO students’ college-going rates fall increasingly short as a percentage of the rate of their wealthier peers as the degree level (two-year versus four-year) and selectivity increase. The statistics in Table 2 imply that if HALO students’ college-going rates matched those of their peers from wealthier households, 1,742 more HALO students would enroll in a four-year college every year—and 1,153 of those students would enroll in a top college or university.

Another way to characterize this “leak” in the public-school pipeline to college is to consider differences in student counts, as one can then consider how overrepresented one group is over another at each attainment level. The third column of Table 3 (below) reveals that wealthier high-achieving students outnumber HALO students 3-to-1 in third grade, but HALO students are later outnumbered nearly 6-to-1 at four-year colleges and nearly 10-to-1 at top colleges. In other words, the greater the postsecondary attainment level—two-year, four-year, or top college, in that order—the greater the extent to which HALO students’ wealthier peers outnumber them.

Table 3. Educational Attainment of HALO Students and Other Students (Grade 3 Cohorts from 2008–09 to 2013–14)

The tables above also indicate that HALO students are more likely to attend college within Ohio than their wealthier high-achieving peers—which is to be expected, given the costs associated with higher education. Table 4 (below) provides more information on the top Ohio colleges and universities that HALO students attend. It indicates that 58,900 of the 780,516 public-school students in third grade between 2008–09 and 2013–14 ultimately attended top Ohio colleges and universities. The table illustrates that HALO students are heavily outnumbered at all of Ohio’s top colleges and universities and that 75 percent of HALO students (2,299 of 3,036) attend the main campuses of just two schools: The Ohio State University and the University of Cincinnati.

Table 4. Attendance at Top Ohio Colleges (Grade 3 Cohorts from 2008–09 to 2013–14)

Potential Source of the Leak: Advanced Coursework and Gifted Services

HALO students’ rates of participation in advanced coursework and gifted services provide some suggestive evidence of how schools may be contributing to the leaky public-school pipeline described above. Table 5 indicates that HALO students’ enrollment in high school–level math courses (primarily algebra) while in middle school is eighteen percentage points lower than the enrollment rate of their wealthier peers. The divergence in advanced coursework is even greater in later grades. The rate at which HALO students enroll in AP, IB, or other dual-credit college courses (as part of Ohio’s CCP program) while in high school is twenty-five percentage points (34 percent) lower than the rate of their wealthier high-achieving peers. And HALO students are approximately twelve percentage points (44 percent) less likely to receive gifted services in English language arts or math in middle school or high school.[31] These statistics imply that if HALO students took advanced coursework at the same rate as their wealthier high-achieving peers, then 1,103 more HALO students in a given third-grade cohort would ultimately take high school–level math courses during middle school and 1,644 more HALO students would take at least one AP course in high school.

Table 5. Courses Taken by HALO Students and Other Students as Percentages of the Respective Samples (Grade 3 Cohorts from 2008–09 to 2013–14)

 

HALO Students’ Districts of Residence and College-Going

This portion of the analysis examines how much the districts where HALO students live affect their probability of attending college. Focusing on where students live, as opposed to which schools they attend, has several benefits. First, one can consider the broader ecosystem that influences third graders’ future college-going, beyond school-specific inputs. Factors include the affluence and academic performance of their residential peers and the availability of publicly funded school options beyond district schools. For example, charter schools affect the educational outcomes of students who attend them, of course, but they also affect students attending traditional district schools that must compete with these alternatives.

A second benefit of focusing on HALO third graders’ district of residence is that doing so makes it straightforward to analyze the factors that predict college-going. The elementary schools students attend vary significantly in terms of grades served and governance structures, making it difficult to identify a broad set of comparable metrics and to attribute college-going to particular schools. For example, some students attend K–4 schools that could influence their future trajectories for just one more year, whereas others attend K–12 schools that could shape their futures straight through to graduation. But all third-grade HALO students reside in districts that serve students through twelfth grade.

The analysis below uses statistical estimates of “college-going value-added” to characterize districts’ impacts on the probability that HALO students attend college. These estimates capture the extent to which students in a district exceed (or fall short of) average rates of college attendance for students with similar characteristics. Generating these estimates first requires calculating for each student their expected probability of attending college based on their third-grade achievement levels and attributes such as parents’ income level, sex, race, and whether they were identified as gifted, homeless, a migrant, residing in a non-English-speaking household, having repeated third grade, having an individualized education plan (IEP), or having limited English proficiency. Then, for each district, one calculates the difference between the actual rate of college-going for those students and the expected (average) rate of college-going for students with similar characteristics.[32] The analysis focuses on attendance at any four-year college and, in particular, more selective top colleges, as these are institutions that HALO students have the academic potential to attend and that would likely best enable them to realize their potential.[33]

Note that the analysis below is primarily concerned with identifying district-level factors—including those related to peer characteristics, school spending, and the availability of advanced coursework and gifted services—that predict the extent to which a district’s HALO students attend college (relative to expectations based on averages for students with similar characteristics). This portion of the analysis is thus about assessing how the broader ecosystem in a HALO student’s district of residence predicts their college-going, as opposed to estimating the causal impact of the specific schools they attend or the coursework they take. Analyses in subsequent sections of this report consider the impact of schools and coursework more directly.

Districts’ College-Going Value-Added for HALO students

Before examining patterns in the college-going value-added of residential school districts across Ohio, there are some important facts to consider about the value-added estimates. First, the value-added of districts in which students reside explains to a significant degree the gap in college-going between HALO students and their wealthier high-achieving peers. Most notably, 37 percent of the gap in attendance at top colleges between HALO students and their wealthier high-achieving peers is attributable to these students residing in different districts, as opposed to HALO students and their wealthier high-achieving peers in the same district attending college at different rates.[34] In short, the district in which students reside appears to matter considerably—particularly when it comes to attending top colleges and universities.

Second, college-going value-added estimates reveal that districts that serve as effective pipelines for Ohio’s average student also serve as effective pipelines for HALO students. Figure 1 plots districts’ college-going value-added for HALO students against their value-added for all students, focusing on the 355 (of 615) Ohio districts in which at least ten HALO students resided while in grade 3 from 2008–09 through 2013–14.

Figure 1. HALO College-Going Value-Added (VA) and Overall College-Going VA (Grade 3 Cohorts from 2008–09 to 2013–14)

The figure reveals that districts where the average student’s probability of attending a four-year college is ten percentage points above expectations (0.10) are those where, on average, HALO students’ probability of attending college will also be approximately ten percentage points above expectations based on their grade-3 characteristics (the slope, beta, is just about 1.0). The correlation coefficient is 0.69, just below the cutoff of 0.7 often used to characterize a statistical relationship as “strong.” Put differently, the R-squared coefficient indicates that a district’s college-going value-added for the average student “explains” 47 percent of the variation in its college-going value-added for HALO students. That districts that are effective postsecondary pipelines for the average student also tend to be effective for HALO students is important, as the value-added measures based on all students are more precise and available across all Ohio districts (in other words, even if they educate fewer than ten HALO students). On the other hand, as the analysis below indicates, the districts and schools that provide the biggest boost to HALO students have distinguishing characteristics—particularly when it comes to getting them into top colleges and universities.

District Attributes That Predict College-Going Value-Added for HALO Students

The research reviewed earlier in this report summarized the range of factors that might bear on HALO students’ education and life outcomes, including the socioeconomic factors affecting them in and out of school and the quality of their school options. The analysis in this section examines the predictive value of such factors—including peer affluence, peer achievement, district spending, the availability of advanced coursework and services, and the availability of alternative schooling options, including charters and private-school vouchers—using reasonable (though imperfect) proxies captured at the district level at the time students were in third grade (2008–09 to 2013–14). In other words, these variables should capture the district conditions at the third-grade baseline and do not measure students’ actual experiences as they progressed through school. That is why these factors are characterized as district-level “predictors” of students’ future college-going (net of expectations based on students’ characteristics).

The district-level predictors include the standardized fraction of students residing in a district who are not eligible for free or reduced-price lunches (“peer affluence”)[39]; the standardized average of third-grade test scores among students residing in the district (“peer achievement”); the standardized spending per pupil in district schools; the average of the standardized fraction of students enrolled in charter schools and the standardized fraction of students eligible for private-school vouchers (“school options”); and the average of the standardized fraction of students taking high-school math courses in grades 6–8 and the standardized fraction of students receiving gifted services in ELA or math in any grade (“advanced courses and services”). These variables are standardized so that the magnitudes of their relationships to district value-added are directly comparable in each figure below (but not between figures). The relationships between these district-level attributes and residential districts’ value-added is estimated using basic regression models that include all attributes at once, such that the estimate for any given attribute is based on holding the others constant.[40]

Figure 2 (below) presents the estimates with respect to value-added at four-year colleges and universities. It reveals that peer achievement is by far the strongest predictor of college-going value-added for HALO students. Going from an average district to one with test scores one standard deviation above average (equivalent to a district’s average student going from approximately the fiftieth percentile to approximately the sixtieth percentile on the third-grade test-score distribution) corresponds to a 3.5-percentage-point increase in the probability that district students attended a four-year college within two years of graduating high school.[41] Notably, spending per pupil, peer affluence, the proportion of district students taking advanced math courses or receiving gifted services in ELA or math, and the availability of school options have no statistically significant predictive value.[42] In other words, having high-achieving peers—and the family, community, and school factors that contributed to their high achievement leading up to third grade—appear to have outsize effects on students’ future college-going.

Figure 2. Predictors of Districts’ Value-Added for Four-Year Colleges

The predictive value of achievement is similar whether or not one controls for the other factors. It’s important to note, however, that the results in Figure 2 are partly attributable to the analysis’s putting more weight on districts that enroll more HALO students. Consistent with research reviewed earlier in this report, the primary predictor of college-going value-added for the average Ohio student is the affluence of residential peers—that is, the percentage of students who do not qualify for free or reduced-price lunches—with peer achievement as the second-strongest predictor (see Table D2 in Appendix D). The divergence in results could be because peer affluence does not alter HALO students’ probability of college attendance after third grade (which is what Figure 2 implies, although it may still play a significant role in driving peers’ third-grade achievement). But it could also be because the measure of peer affluence does not vary sufficiently among the districts that HALO students attend, meaning there is relatively little opportunity to observe its effects. Relatedly, peer affluence is not measured as precisely as peer achievement. There is significant variation in household income even among students who qualify for free meals, and this unmeasured variation could in fact matter for HALO students.

What one can say is that based on available data and an analysis focused on altering HALO students’ college trajectory after third grade, peer achievement is the strongest predictor of college-going value-added across the districts where HALO students are concentrated. One can reasonably conclude, therefore, that whether district peers are high-achieving is the best predictor of whether a HALO student’s district will serve as an effective pipeline to a four-year college. But there is also good reason to suspect—based on past research and the analysis of college-going for the average Ohio student—that HALO students would also benefit from residing in districts with wealthier peers for reasons other than their achievement level.

Figure 3 (below) presents the results of a similar analysis but focuses on attendance at top colleges—the outcome for which a student’s district of residence (or attendance) most greatly influences the gap in college-going between HALO students and their wealthier high-achieving counterparts (see prior subsection of this report).[43] Notably, peer achievement is no longer a significant predictor of college-going value-added. Figure 3 reveals that when one focuses on HALO students attending top colleges, the significant predictors are the availability of and participation in alternative schooling options and coursework or services for advanced students.

Figure 3. Predictors of Districts’ Value-Added for Top Colleges

Specifically, the results in Figure 3 indicate that a one-standard-deviation increase in the proportion of district students attending charter schools or eligible for private-school vouchers corresponds to a 1.3-percentage-point increase in the probability that a district’s students attend a top college within two years of graduating high school. The results also indicate that students in districts with higher rates of enrollment in gifted services in ELA and math or in advanced math coursework in grades 6–8 have a 0.8-percentage-point higher probability of attending a top college. Note that disaggregating the impacts of advanced math coursework and gifted services (which often consists of advanced coursework), as well as charter school enrollments and rates of voucher eligibility, reveals that these results are driven primarily by rates of charter school attendance and gifted service provision (see Table D3 in Appendix D). The average attendance rate at a top college is 9 percent for HALO students, so the combined “impact” of a one-standard-deviation increase in school options, advanced math coursework, and gifted services (1.3 plus 0.8, a 2.1-percentage-point increase) is a 24 percent increase in the probability that HALO students residing in a district attend a top college.

One cannot directly compare the estimates in Figure 2 and Figure 3, as they are based on different effective samples of students.[44] For example, the analysis of value-added for top colleges puts more emphasis on data for HALO students who reside among higher-achieving peers, as those are the students for whom we can most reliability estimate top-college attendance. What one can take away from Figure 3 is that, among districts for which we have relatively reliable measures of top-college-going value-added, those that have more school options and provide more advanced coursework and gifted services are more successful at getting their HALO students to attend top colleges.

Once again, it is worth noting that for the average Ohio student, the affluence of district students is the strongest predictor of whether they attend a top college, just as it is the strongest predictor of whether a student attends a four-year college. In other words, the results suggest that even after third grade, the affluence of a student’s residential peers can significantly alter that student’s schooling trajectory. That is also the case for HALO students if one does not put greater weight on value-added estimates that are more reliable (see Table D5 in Appendix D). What the analyses in figures 2 and 3 tell us, though, is that when one focuses on the best available data on HALO students’ college-going—based on where HALO students tend to reside—the average achievement of their peers is the primary predictor of whether they attend a four-year college, and the availability of alternative schooling options, advanced coursework, and gifted services are the best predictors of whether they get into top colleges. The analysis of high schools (below) provides an informative test of the robustness of these results.

 

A Closer Look at Districts’ College-Going Value-Added for HALO Students

The predictors identified above provide a general portrait of the characteristics of districts where HALO students are most likely to realize their potential. The analysis here adds to this portrait by illustrating how districts with varying levels of college-going value-added are distributed across the state, and it identifies (and lists the value-added estimates for) the twenty-five Ohio districts that provide the most opportunity for HALO students. Doing so requires adjusting the value-added estimates for statistical uncertainty. Appendix C describes the procedure for generating these empirical Bayes estimates. Note that, like the previous analysis above, this analysis includes only those districts that educated at least ten HALO students between the 2008–09 and 2013–14 school years.

Figure 4 (below) presents a pair of heat maps that illustrate the geographic distribution of district value-added for HALO students—providing a visual look at the estimated quality of public-school pipelines for recent HALO students across Ohio. The maps yield a couple of patterns. First, district value-added for HALO students’ attendance at four-year colleges and universities (panel a) seems relatively stronger in suburbs surrounding (but not including) large urban districts—particularly near Cleveland and Akron and, to a lesser extent, Columbus, Cincinnati, Toledo, and Youngstown. But the pattern is fuzzy, and there are significant clusters of high-value-added districts in some remote areas.

Figure 4. Districts’ College-Going Value-Added for HALO Students

Second, district value-added for HALO students’ attendance at top colleges is highest in the Cincinnati and Dayton areas, followed by the Columbus and Cleveland areas, respectively. The clustering may be because of Cincinnati and Dayton’s proximity to top public universities in Ohio, including the University of Cincinnati, the University of Dayton, and Miami University. Research indicates that geographic proximity and lower tuition rates are big factors driving the enrollment of low-income students, and the public universities near Cincinnati and Dayton likely offer both to students in nearby districts.[45]

Table 6 (below) confirms the patterns in Figure 4. It presents the twenty-five districts with the highest average rank across the four-year and top-college-going value-added estimates (weighted equally). Twenty of the top twenty-five districts are suburban. Most have high-achieving and relatively affluent students, and none is in a city. There are a few exceptions, such as Winton Woods City School District, Waverly City School District, Bath Local Schools, and Union-Scioto Local School District, which provide a relatively strong pipeline for HALO students despite relatively high poverty rates. But these districts do not educate many of Ohio’s HALO students. Indeed, across six third-grade cohorts (from 2008–09 to 2013–14), only Dublin City School District, Fairfield City School District, and Pickerington Local School District educated over 100 HALO students—just twenty-six to twenty-eight HALO third graders per year in each of these three districts. 

Table 7 (further below) presents the districts with the highest numbers of HALO third graders. It demonstrates that Ohio’s HALO students are most concentrated in large city and, to a lesser extent, suburban districts and that their rates of college-going are typically well below what one would expect based on their third-grade test scores and demographics. For example, the 1,346 HALO third graders in Columbus City Schools (approximately 225 per year from 2008–09 to 2013–14) attend four-year colleges at rates that are eighteen percentage points below the statewide average for other similar students. Residing in Cleveland, Cincinnati, or Toledo similarly impedes four-year college–going among HALO students—though, consistent with the map above, residing in Cincinnati or Dayton seems to have benefits when it comes to attending top colleges and universities.

Table 6. Districts with the Highest College-Going Value-Added for HALO Students

Note: The table lists the districts that most exceeded expectations in terms of the probability that their grade-3 residents would attend a four-year college or a top college (where students have an average score of approximately 25 or above on the ACT and 1200 or above on the SAT). Specifically, focusing on districts that had at least ten HALO residents attending public schools (including charters) in grade 3 between 2008–09 and 2013–14, it identifies districts with HALO college-going value-added estimates that rank among the top twenty-five statewide (i.e., those whose average ranking across both estimates puts them in the top twenty-five). The value-added estimates listed in the two columns next to district names indicate the increased probability of college attendance among HALO students. (For example, Bay Village City School District’s value-added of 0.26 indicates that the fraction of HALO students going to college was twenty-six percentage points greater than expected based on those HALO students’ grade-3 characteristics.) The remaining columns summarize district characteristics in those baseline years (2008–09 to 2013–14) when HALO residents were in third grade, summarizing how those districts stack up compared with other Ohio districts regarding the baseline achievement level of students in grade 3 (average test scores in ELA and math), the proportion of grade-3 students who qualify for free or reduced-price lunches (“poverty”), spending per pupil, the fraction of students in grades 6–8 who are enrolled in advanced math courses, and the fraction of students receiving gifted services in ELA or math. The possible categories are “high” (top 20 percent), “high/moderate” (sixtieth to eightieth percentiles), “moderate” (fortieth to sixtieth percentiles), “moderate/low” (twentieth to fortieth percentiles), and “low” (bottom 20 percent). “NCES” is an abbreviation for the National Center for Education Statistics. “TPS” is an abbreviation for traditional public schools, indicating that public school spending captures only that spent on district schools (not charter or state STEM schools).

Table 7. Districts with the Greatest Number of HALO Third Graders from 2008–09 to 2013–14

Note: The table lists the districts with the greatest number of grade-3 HALO students (from 2008–09 to 2013–14) included in the analysis. Once again, the list is restricted to districts that had at least ten HALO residents attending public schools (including charters) in grade 3 between 2008–09 and 2013–14, and the value-added estimates listed in the two columns next to district names indicate the increased probability of college attendance among HALO students. The remaining columns summarize district characteristics in those baseline years (2008–09 to 2013–14) when HALO residents were in third grade, summarizing how those districts stack up compared with other Ohio districts regarding the baseline achievement level of students in grade 3 (average test scores in ELA and math), the proportion of grade-3 students who qualify for free or reduced-price lunches (“poverty”), spending per pupil, the fraction of students in grades 6–8 who are enrolled in advanced math courses, and the fraction of students receiving gifted services in ELA or math. The possible categories are “high” (top 20 percent), “high/moderate” (sixtieth to eightieth percentiles), “moderate” (fortieth to sixtieth percentiles), “moderate/low” (twentieth to fortieth percentiles), and “low” (bottom 20 percent). “NCES” is an abbreviation for the National Center for Education Statistics. “TPS” is an abbreviation for traditional public schools, indicating that public school spending captures only that spent on district schools (not charter or state STEM schools).

 

HALO Students’ High Schools and College-Going

Focusing on high schools provides important advantages. First, the attrition issues with the district-level analysis of HALO students (because of students exiting public schools after third grade and not being matched to postsecondary data) are minimal because the analysis is limited to students linked to specific high schools in ninth grade. Ninth-grade cohorts are those that the Ohio Department of Education and Workforce matches to National Student Clearinghouse data, so there is less chance that a student is missed and incorrectly coded as not having attended college. A second advantage is that one can observe students’ entry into high schools, as a large majority of high schools have the same entry grade. Thus, unlike the district analysis—which awards districts no value-added credit for educational gains to which they contribute prior to grade 4 (grade 3 is the baseline for the value-added calculations)—one can generally examine students’ education trajectories before and after entering high school.

A third advantage of focusing on high schools is that one can estimate the impact of more immediate preparation for college, particularly via college-level coursework. Approximately 40 percent of the gap between HALO students’ AP course–taking and credit for Ohio’s public colleges (score of 3 or above) and those of HAHI students is attributable to these students residing in different districts.[46] However, data on AP and other high school coursework are unavailable for the 2008–09 to 2013–14 baseline years used in the district-level analysis. By setting a new grade-9 baseline for school years 2014–15 to 2018–19, one can examine the link between the availability of AP courses in high school and the probability of college attendance.

A significant share of the ninth-grade students used in this analysis are the third graders used in the district analysis, even though this analysis identifies HALO students using seventh- and eighth-grade data instead of third-grade data.[47] As with districts of residence, the high schools students attend in grade 9 explain a significant share of the gap in course-taking and college-going. Indeed, nearly 50 percent of the gap between HALO students’ AP course–taking and credit for public colleges (score of 3 or above) and those of HAHI students is attributable to students attending different high schools in grade 9. Similarly, although where students go to high school accounts for 13 percent of the four-year college–going gap between HALO students and their wealthier peers, it accounts for 41 percent of the gap in attendance at top colleges.[48]

The analysis of high schools proceeds like the analysis of districts. It first involves estimating how much the schools students attended in ninth grade affected their probability of attending college—isolating, to the extent possible, high schools’ influence from students’ cognitive, demographic, and socioeconomic attributes as of eighth grade (see Appendix C for details). The analysis then examines the extent to which school-level attributes observed at the time students were in eighth grade are predictive of college attendance (relative to expectations based on averages for students with similar characteristics). Given the methodological advantages enumerated above, this analysis provides a valuable check and extension of the district-level analysis.

 

High School Attributes That Predict College-Going Value-Added

The predictors are like those used in the district-level analysis but are measured at the school level. They include the fraction of students who are not eligible for free or reduced-price lunches (“peer affluence”),[49] peers’ average grade-8 test scores (“peer achievement”), spending per pupil, and the average share of students receiving gifted services in ELA or math, taking AP/IB courses, and getting dual credit through CCP (“advanced courses and services”). Unlike the district-level analysis, however, there is not a measure of school options, as there is no analogous measure when shifting the focus from the district of residence to the school of attendance.[50] Descriptive statistics, results in tabular form, and sensitivity analyses are available in Appendix E. 

Figure 5 (below) presents the estimates with respect to value-added for four-year colleges and universities. It reveals that when comparing high schools’ attributes at baseline, peer achievement is by far the strongest predictor of college-going. An increase of one standard deviation in the average test score of one’s high-school peers (equivalent to a school’s students going from approximately the fiftieth percentile to the sixtieth percentile on the grade-8 test-score distribution) corresponds to an 8.6-percentage-point increase in the probability of attending a four-year college.[51] School spending comes in second, with an increase of one standard deviation—$2,350 per student—corresponding to an increase of 1.9 percentage points in the probability of attending college. Notably, as in the district analysis, there is no positive relationship between the proxy for peer affluence or advanced course–taking and a school’s value-added with respect to attending a four-year college. Indeed, HALO students’ college-going is higher when they attend school with other students who are eligible for free or reduced-price lunches. The negative result for peer affluence may be because of the limitations of using free or reduced-price meal eligibility as a proxy for student poverty.[52]

Figure 5. Predictors of High Schools’ Value-Added for Four-Year Colleges

Figure 6 (below) presents the estimates for attendance at top colleges and universities. Once again, like the district-level analysis, it reveals that advanced coursework and gifted services overtake peer achievement as the strongest predictor of college-going. Specifically, a one–standard deviation increase in participation in advanced coursework and gifted services corresponds to a 4.3-percentage-point increase in the probability of attending a top college. This result is driven entirely by AP coursework, as opposed to other dual-credit options or gifted services.[53] Peer achievement comes in a close second (an estimated effect of 3.2 percentage points), and HALO students’ college-going is again somewhat higher when they attend school with other students who are eligible for free or reduced-price lunches.[54]

Figure 6. Predictors of High Schools’ Value-Added for Top Colleges

Note that null or negative estimates (e.g., for advanced coursework in Figure 5 and peer affluence in both figures) do not imply that those factors do not affect college-going. This analysis is merely identifying school-level predictors of changes in students’ schooling trajectories after grade 8.[55] Additionally, as discussed in the district analysis, whether the analysis can identify such predictors depends on the effective sample used. The schools that educate a sufficient number of HALO students and that are reliably effective or ineffective at changing these students’ trajectories when it comes to four-year colleges may not vary all that much in the coursework they offer (e.g., few may offer AP courses), but those that are particularly effective or ineffective at changing HALO students’ trajectories when it comes to top colleges may indeed vary significantly in the availability of AP coursework. That does not mean that schools effective at getting their students into four-year colleges without the aid of AP coursework would not benefit from it. Indeed, an analysis that includes all Ohio schools and students (not just HALO students) reveals that the availability of AP coursework (but, once again, not CCP or gifted services) is predictive of getting the average Ohio student into a four-year college or top college.[56]

The above results largely corroborate those of the district-level analysis and, reassuringly, are less sensitive to the sample used and statistical weighting.[57] Once again, the analysis points to the predictive power of peer achievement and advanced coursework in getting HALO students into four-year and top colleges, respectively. This is an important addition to the set of results from the district-level analysis, as a student’s district of residence explains much of the variation in AP course–taking.

 

A Closer Look at High Schools’ College-Going Value-Added for HALO Students

The preceding analysis identifies peer achievement and AP course enrollment rates as top predictors of HALO students’ probability of attending four-year and top colleges, respectively. This section looks for similarities in high schools that are particularly effective at increasing HALO students’ probability of attending a four-year or top college. First, Table 8 (below) lists the top twenty-five high schools in terms of the average statewide rank of their college-going value-added estimates for four-year and top colleges and universities. (The appendix provides a high-school value-added ranking focused exclusively on top colleges and universities.[58])

Table 8 reveals that the high schools most effective at changing students’ attainment trajectories after grade 8 are primarily in the suburbs, but there are some notable exceptions. Portsmouth High School, for example, is in a district that federal data label as a town, and it has relatively low average test scores, high rates of school meal eligibility, and moderate-to-low rates of participation in AP coursework. This school is unusual, however, in that it also includes a junior high school (it serves grades 7–12). Zanesville High School is another high-poverty and low-achievement school in a town, but notably, it boasts a moderate-to-high rate of enrollment in AP courses. Perhaps the biggest standout is the Cleveland School of Science and Medicine—one of the Cleveland Metropolitan School District’s selective “criteria schools.” It is a high-poverty school in a city district, but its students are high-achieving and have access to advanced coursework.

Table 9 (further below) restricts the rankings to schools in the top 40 percent in terms of the proportion of their students who qualify for free or reduced-price lunches (those indicated as having “high” or “high/moderate” poverty rates in Table 8). It includes several schools in city districts that enroll large numbers of HALO students, with a solid mix of schools from towns, suburbs, and rural areas. Consistent with the analysis of value-added for top colleges and universities, several schools on the list have moderate-to-high rates of student participation in AP coursework and moderate-to-high average test scores. Among these schools are four of Cleveland Metro’s five selective “criteria schools.[59] It is beyond the scope of this study to thoroughly examine similarities across these schools, but the presence of selective and early-college high schools stands out.

Table 8. High Schools with the Highest College-Going Value-Added (VA) for HALO Students

Note: The table lists the high schools that most exceeded expectations in terms of the probability that their grade-9 students would attend a four-year college or a top college (where students have an average score of approximately 25 or above on the ACT and 1200 or above on the SAT). Specifically, focusing on districts that had at least ten HALO residents attending public schools (including charters) in grade 9 between 2014–15 and 2018–19, it identifies schools with HALO college-going value-added estimates that rank among the top twenty-five statewide (i.e., those whose average ranking across both estimates puts them in the top twenty-five). The value-added estimates listed in the two columns next to school names indicate the increased probability of college attendance among HALO students. (For example, Turpin High School’s value-added of 0.25 indicates that the fraction of HALO students going to college was twenty-five percentage points greater than expected based on those HALO students’ characteristics in grade 8 and earlier.) The remaining columns summarize school characteristics in those baseline years (2014–15 to 2018–19) when HALO residents were in ninth grade, summarizing how those schools stack up compared with other Ohio high schools regarding the baseline achievement level of students in grade 8 (average test scores in ELA and math), the proportion of grade-8 students who qualify for free or reduced-price lunches (“poverty”), spending per pupil, the fraction of students in grades 9–12 who are receiving gifted services in ELA or math, and the fraction of students taking AP and IB coursework. The possible categories are “high” (top 20 percent), “high/moderate” (sixtieth to eightieth percentiles), “moderate” (fortieth to sixtieth percentiles), “moderate/low” (twentieth to fortieth percentiles), and “low” (bottom 20 percent).

Table 9. High-Poverty High Schools with the Highest College-Going Value-Added for HALO Students

Note: The table lists the high schools that most exceeded expectations in terms of the probability that their grade-9 students would attend a four-year college or a top college (where students have an average score of approximately 25 or above on the ACT and 1200 or above on the SAT). Specifically, focusing on districts that had at least ten HALO residents attending public schools (including charters) in grade 9 between 2014–15 and 2018–19, it identifies schools with HALO college-going value-added estimates that rank among the top twenty-five among high- and high/moderate-poverty schools statewide (i.e., those whose average ranking across both estimates puts them in the top twenty-five among schools with poverty rates in the top 40 percent). The value-added estimates listed in the two columns next to school names indicate the increased probability of college attendance among HALO students. (For example, a value-added of 0.20 indicates that the fraction of HALO students going to college was twenty percentage points greater than expected based on those HALO students’ characteristics in grade 8 and earlier.) The remaining columns summarize school characteristics in those baseline years (2014–15 to 2018–19) when HALO residents were in ninth grade, summarizing how those schools stack up compared with other Ohio high schools regarding the baseline achievement level of students in grade 8 (average test scores in ELA and math), the proportion of grade-8 students who qualify for free or reduced-price lunches (“poverty”), spending per pupil, the fraction of students in grades 9–12 who are receiving gifted services in ELA or math, and the fraction of students taking AP and IB coursework. The possible categories are “high” (top 20 percent), “high/moderate” (sixtieth to eightieth percentiles), “moderate” (fortieth to sixtieth percentiles), “moderate/low” (twentieth to fortieth percentiles), and “low” (bottom 20 percent).

HALO Students Taking Advanced Coursework

The preceding analysis identifies commonalities among the districts and schools where HALO students are most likely to exceed expectations in their college-going. One clear takeaway is that, holding other inputs constant, districts and schools where students take advanced coursework at higher rates are also those where HALO students attend top colleges at higher rates. The analysis cannot tell us, however, how the postsecondary prospects of HALO students compare between those who participate in advanced coursework and those who do not participate in advanced coursework. The analysis in this section does so by focusing on observationally similar HALO students who attended the same baseline school. Specifically, it focuses on similar students who attended the same elementary school in grade 3, comparing the college-going of those who received gifted services in ELA or math with the college-going of those who did not receive gifted services in ELA or math (often advanced coursework) and comparing the college-going of those who took advanced math coursework in grades 6–8 with the college-going of those who did not, and it also focuses on similar students who attended the same middle school in grade 8, comparing the college-going of those who participated in AP/IB coursework or dual-credit options via CCP with the college-going of those who did not.[60]

Note that although this research design is relatively strong, it is unlikely to isolate the causal impact of taking advanced coursework or receiving gifted services on students’ college attendance. There are school-related factors and non-school-related factors that might lead talented HALO students to enroll in advanced coursework (or not), and these other factors may affect the probability of college-going beyond their impact on course-taking. For example, school guidance counselors and students’ friends may affect their aspirations, which may affect both course selection and college attendance. These environmental forces may be more important levers to pull to increase college-going among HALO students (hence the preceding school- and district-level analyses). That said, because the comparisons below are between similar students—who attended the same baseline school and who have similar prior test scores, demographic characteristics, gifted identification, and the like—this analysis should capture, to a reasonable extent, the impact of advanced course–taking and gifted services on college-going among HALO students.

Figure 7 (below) indicates that HALO students who took advanced coursework—from algebra in middle school to AP classes in high school—have substantially higher rates of college attendance than similar HALO students who did not. HALO students who took high-school math courses in grades 6–8 (mostly algebra) have college-going rates that are eighteen percentage points higher than similar students who did not take advanced math coursework. And Figure 8 (further below) indicates that HALO middle-school students’ probability of attending a top college is six percentage points higher if they take advanced math courses. One cannot directly compare the magnitude of the estimated impacts between middle schools and high schools, as they are based on different samples of students. The middle-school sample also less reliably links students to higher-education data, which most likely leads to underestimating true effects. One should instead focus on the fact that the estimates are large and positive across all types of coursework and services in both middle school and high school. 

Figure 7. Comparing Attendance Rates at Four-Year Colleges Between Similar HALO Students Who Did/Didn’t Take Advanced Coursework or Receive Gifted Services

Figure 8. Comparing Attendance Rates at Top Colleges Between Similar HALO Students Who Did/Didn’t Take Advanced Coursework or Receive Gifted Services

Figure 7 also indicates that taking AP coursework likely provides the strongest path to college for HALO high-school students—whether they are eyeing a four-year college or a top college. Students who took at least one AP (or IB) course increased their college-going by twenty-nine percentage points relative to HALO students who did not take at least one AP (or IB) course.[61] Figure 8 indicates that the margin for top-college attendance is sixteen percentage points. The estimates for CCP are also positive and substantial, but they are half the size of the AP estimates when it comes to HALO students getting into top colleges.

Finally, the figures indicate that students who receive gifted services in ELA or math are indeed more likely to attend four-year colleges (by six percentage points) and top colleges (by four percentage points) relative to HALO students who do not receive such services—which are a variable mix of advanced coursework or other programming linked to students’ areas of gifted identification.

The above analysis does not allow one to completely isolate the causal impact of course-taking from other factors that might lead students to take advanced courses in the first place. And again, one should not compare the magnitudes of the estimates between the middle-school and high-school estimates, as they are based on different samples. What is clear, however, is that HALO students who took more advanced courses and received gifted services fared far better than those who did not. The results also suggest that the impacts of advanced math coursework in middle school and AP coursework in high school are larger than the impacts of gifted services and CCP coursework, respectively.

 

Summary and Policy Implications

The analysis confirms that HALO students in Ohio are significantly less likely to attend a college or university, especially a top college or university, than their wealthier peers who had similar academic performance and demographic characteristics back in third grade. The results suggest that students’ district of residence and the high school they attend play a significant role in this college-going disparity—particularly when it comes to top colleges and universities. A major reason appears to be that the academic achievement of HALO students’ peers and their access to advanced coursework (particularly AP classes) are strong predictors of college-going that vary significantly across districts and schools.

The districts and schools that provide the greatest boost to HALO students’ college-going tend to be in high-achieving, wealthy suburbs that enroll few HALO students. But the analysis also indicates that schools serving primarily low-income kids—often located in cities and towns—are sometimes effective at getting HALO students into college at high rates. This is particularly true if they surround HALO students with high-achieving peers and provide them with advanced coursework and gifted services. For example, four of Cleveland’s five selective “criteria schools”—which have high poverty rates but offer a variety of dual-credit options and have relatively high-achieving student bodies—are among the top twenty-five high-poverty high schools in Ohio in terms of boosting HALO students’ college-going. More generally, across all Ohio districts and schools, HALO students who took advanced coursework or received gifted services in middle and high school were far more likely to attend four-year colleges and universities than other HALO students who attended the same schools but did not pursue advanced coursework.

Thus, this study suggests Ohio could improve its K–12 pipeline to higher education by providing HALO students with more opportunities to take advanced coursework, receive gifted education services, and surround themselves with similarly high-achieving peers. Schools should encourage HALO students to take advantage of those opportunities when they are available. Rigorous research demonstrates how effective high-school guidance counselors can be in getting HALO students into top colleges and universities simply by making them aware of those opportunities.[62] The results of this analysis suggest that HALO students should receive encouragement to take advanced coursework and to get on a path to college well before it is time to choose a college, however. Policy proposals to automatically enroll high-achieving kids in algebra in eighth grade are a good start. Doing so could, in turn, increase AP coursework in high school—the strongest predictor of college-going among HALO students.[63] The National Working Group on Advanced Education’s (2023) report enumerates several opportunities for schools and policymakers to implement such an approach throughout the K–12 pipeline.[64]

Grouping high-ability students together and offering advanced coursework requires enough high-achieving students to make it financially viable, and it requires access to a supply of teachers with the requisite expertise. Many HALO students reside in districts that don’t meet these conditions.[65] But as this analysis shows, there are many districts with substantial concentrations of HALO students that do not provide much of a boost in HALO students’ college-going. For example, Columbus City Schools has the highest number of HALO students residing within its boundaries, and these students’ rate of college-going is eighteen percentage points lower than the state average for students with similar test scores and other attributes in third grade. That’s forty HALO third graders in Columbus alone, every year, who won’t go to college but would have if they resided in an Ohio district of average quality (where HALO students’ college-going is still significantly lower than that of their wealthier high-achieving peers). Indeed, among the twelve districts with the largest number of HALO students—including Ohio’s eight large urban districts—only in Akron does the rate of four-year college–going among HALO students not fall below what one would have expected based on these students’ third-grade test scores and attributes.

One way to address this problem is for Ohio policymakers to expand access to high-quality school options and to provide families with the information and decision-making assistance they need to make the best use of those options. For example, policymakers might establish selective schools that draw high-achieving students from several districts to get the critical mass of students needed to make the model viable. Each district in the region could have a right to some share of seats. Note, however, that to reach HALO students, there may need to be seats set aside for low-income students. For example, Metro Early College High School—a state STEM school that draws students from several districts in central Ohio—is effective at boosting HALO students’ chances of getting into top colleges and universities. But it does not make the top-twenty-five among “high-poverty” high schools for getting HALO students into college, because it educates relatively few low-income students.[66] Moreover, Ohio’s state STEM schools do not have test score requirements, which likely undermines their effectiveness when it comes to improving HALO students’ rates of college-going.

Another option is to expand Ohio’s interdistrict open enrollment program to all districts so that HALO students could attend a neighboring high-achieving district, free of charge, if that district has open seats.[67] Suburban districts filled with high-achieving students might be open to admitting HALO students from surrounding districts if there were financial incentives to doing so—for example, if they were to receive the substantially higher per-pupil aid allocated to low-income students attending high-poverty districts. The recent expansion of EdChoice, which provides considerable financial assistance to low-income households, is also a promising route, given its success in getting Ohio’s low-income students into college.[68]

The state of Ohio excels at providing information that is predictive of schools’ college-going value-added. For example, the report card grades capturing high-school achievement levels, test score growth, and college preparation during the baseline years (2014–2019) were all highly predictive of students’ future college-going.[69] Thus, as new school options become available, the information is largely available to assess the quality of those options. The missing link seems to be helping families incorporate this information into their decision-making. Research demonstrates that with such support, parents make decisions that better align with educational goals.[70] Such support becomes increasingly important as school options expand.

It is worth remembering that the decision to attend college often hinges on factors beyond the K–12 public schooling system. For example, one reason the concentration of high-achieving students and high rates of AP coursework in schools might lead to more college-going is that colleges are more likely to target these high schools in the recruitment process.[71] Similarly, as the analysis suggests, HALO students’ attendance at top colleges increases if their district of residence is geographically near top public colleges and universities, which is consistent with research demonstrating the impact of cost and proximity on low-income students’ college-going decisions.[72] Nevertheless, this report also clearly documents the educational benefits of HALO students attending public schools that provide access to advanced coursework and high-achieving peers, in spite of factors such as college cost and proximity.

The analysis in this report focuses on how districts and schools might alter students’ academic trajectories beyond third grade—after family-, community-, and school-related factors have already played a significant role in putting kids on or off track for college—and it relies primarily on descriptive statistical methods to do so. Yet the results are strikingly consistent throughout, line up with the results of other rigorous studies, and point to clear opportunities for substantially improving college-going among Ohio’s HALO students. Unlocking this talent not only would transform these students’ futures but would likely also deliver broad economic and social benefits for Ohio.

 

Acknowledgments

I am grateful for the extensive and helpful feedback I received from Fordham’s Chad Aldis, Aaron Churchill, Chester Finn, and Michael Petrilli (listed in alphabetical order by last name), as well as from external readers Richard Kahlenberg, Vladimir Kogan, and Jonathan Plucker. The report improved greatly because of their feedback. The weaknesses that remain are entirely my fault.

- Stéphane Lavertu

I offer my deepest gratitude to Stéphane Lavertu for his fine work carrying out this project. Special thanks also to the Ohio Department of Education and Workforce for providing the data that made possible this report. I wish to thank David Yontz, who copy edited the report, and Dave Williams, who designed the layout. Finally, I thank my Fordham Institute colleagues Jeff Murray, Victoria McDougald, Meredith Coffey, and Stephanie Distler for their support in report publication and dissemination.

- Aaron Churchill
Ohio Research Director

 

Endnotes