For the past two decades, the College Board has moved aggressively to expand the number of high school students taking Advanced Placement courses and tests — in part by pitching the program to low-income students and the schools that serve them.
It is a matter of equity, they argue.
“What if the best stuff in education were not just for the best to distinguish themselves — but could engage a much broader set of kids?” asked David Coleman, the College Board’s chief executive, in a January podcast interview. “Why are we holding it for some?”
Left out of that narrative is one of the most sobering statistics in education: Some 60 percent of A.P. exams taken by low-income students this year scored too low for college credit — 1 or 2 out of 5 — a statistic that has not budged in 20 years.
Nevertheless, the College Board, citing its own research, says its A.P. program helps all students, regardless of scores, do better in college — a claim that has helped persuade states and local districts to help pay for the tests.
A growing body of research, however, conflicts with the College Board’s claims. One expert went so far as to call the group’s research briefs “junk science.” And some research shows that other advanced programs may make it easier for high school students to earn college credit and lower tuition costs.
But expanding A.P.’s reach is crucial to the College Board’s future. From 2019 to 2022, revenue from its other signature product, the SAT, plummeted to $289.2 million from $403.6 million, as more colleges dropped testing requirements for admissions.
The College Board has tried to bolster Advanced Placement. It is developing new courses, such as business and African American studies. It pushed all A.P. students not just to enroll in the classes but to take the final exams. And in 2021, the nonprofit began circulating a research brief arguing that even students who perform poorly on the exams experience benefits.
That pitch has helped turn A.P. into the College Board’s most lucrative program, generating almost $500 million in revenue in 2022, a chunk of that from the taxpayers.
In its tax filings, the College Board declared that it receives only about $5 million to $6 million annually in direct government funding.
Yet, by examining College Board documents and state and local education budgets, The New York Times estimated that for A.P. alone, the College Board brings in about $100 million annually in public money.
In response to the Times reporting, the College Board said that this year, it received at least $90 million for test fees from the government.
A.P. curriculums, which are given to schools for free, can be enriching and valuable. But the grueling, multi-hour tests put many low-income students at a disadvantage. Their families have fewer resources to spend on test prep; they may not speak English as a first language; and they may have attended elementary and middle schools that provided less effective preparation.
Justin Cohen, an education consultant, served from 2009 to 2015 as the president of Mass Insight, a nonprofit group in Massachusetts that helped expand A.P. access to disadvantaged students.
Results — especially on test passing rates — were mixed, according to a series of studies by researchers at the University of Massachusetts.
“We should not assume the College Board is acting in the country’s best interest by lobbying for the expansion of this,” Mr. Cohen said. “What are the other things you could be doing if you weren’t investing money in expanding access to A.P.?”
A.P.’s High-Stress Tests
Given the American school system’s bleak history of diverting low-income and nonwhite students away from college-prep classes, many educators believe in the power of a rigorous liberal arts curriculum to lift all students.
And Advanced Placement is the go-to program for schools across the country, making it something of a de facto national curriculum. Eighty percent of public high school students attend a school offering five or more courses. This year, students took 5.2 million A.P. exams, up from 1.6 million in 2002.
Students from low-income families have been an important part of that surge, taking 1.1 million tests this year — the highest number ever — compared with 153,000 tests two decades ago.
Roxbury Prep, a charter school in Boston, has built its entire curriculum around A.P., modifying the courses to fit the interests of its students, who are largely from low-income Black and Hispanic families. Massachusetts and 33 other states cover many of the test fees for students like those at Roxbury.
“We keep all doors open,” said Chelsea McWilliams, Roxbury’s principal. “It’s not about lowering expectations. It’s about raising supports.”
For A.P. Language, an English composition course, Roxbury Prep built a rich unit on race in America, featuring Black writers like bell hooks and Ta-Nehisi Coates. Many students said the reading list was inspiring and shaped their college aspirations.
By March — nearly two months before the exams — students in another class, A.P. Literature, had moved from studying August Wilson’s play “Fences” to preparing for the test, studying a handout with 28 tips for the exam.
Intense preparation is necessary for the school’s students, who are competing against students from wealthier families who can afford tutoring, said Brett Peiser, co-chief executive of Uncommon Schools, the charter school network for Roxbury Prep.
Mr. Peiser said the tests are integral to the A.P. experience, an idea the College Board promotes.
“It’s a high bar and rigor is good,” he said.
Roxbury Prep points to alumni like Glahens Paul, who took eight A.P. exams and scored a 3 or above on seven of them. He earned a full scholarship to Centre College in Kentucky, from which he recently graduated. Only calculus resulted in college credit. But Mr. Paul, 22, said his English assignments on race in America gave him a chance to study issues he cared about.
“I came alive,” said Mr. Paul, who immigrated to the United States from Haiti at the age of 9, speaking little English.
Roxbury Prep officials said that 54 percent of their 2023 graduates earned at least one A.P. score of 3 or above.
But that statistic obscures a grimmer picture. Most students take many A.P. classes, increasing their chances of passing one exam, even if they fail many others. Out of 872 A.P. exams taken by recent Roxbury Prep graduates, 80 percent of the tests scored a 1 or 2, according to the school.
Nationally, it is common for students to take multiple exams and emerge with a single passing score — a practice the College Board celebrates as evidence that students are maturing and able to succeed in the program.
But this year, 38 percent of all test scores were 1 or 2. And failure rates were higher for low-income, Black, Hispanic and Native American students.
Trevor Packer, the head of Advanced Placement, said that giving more students access to rigorous material, and identifying those who are able to excel, justifies the failure rates.
He acknowledged that in terms of academic preparation, the program is now serving a much broader group of students than it had in the past.
“American education has an ugly track record when it comes to deciding in advance which students are deserving of the best opportunities,” he wrote in a statement.
The College Board has also argued that even low scores help students. In a 2022 presentation to school district leaders, the College Board pointed to its research brief claiming that students with average A.P. scores of 1 or 2 were significantly more likely to enroll in four-year colleges than peers who had not taken the tests.
The brief, which is all of two pages, also stated that students who received 2s had slightly higher grade-point averages in introductory college classes on the same subjects.
The brief said that the study considered the records of 1.5 million students in the class of 2017, and controlled for factors such as parental education level, race and gender.
But the College Board has released only the brief — not its study or its data.
The brief did not break out the performance of students who received subsidies to take A.P. exams, arguably the most pertinent group for policymakers. And the study did not control for factors within schools that play a strong role in academic success, such as poverty levels and funding.
The College Board has long resisted outside scrutiny of its A.P. data, said Kristin Klopfenstein, director of the Colorado Evaluation and Action Lab at the University of Denver, which uses data to analyze social problems.
Though the College Board employs highly skilled researchers, Dr. Klopfenstein said that this brief, like some others from the nonprofit, amounts to “junk science,” reading more like marketing material than a rigorous summary of a study’s findings.
Independent research points to problems with Advanced Placement.
A 2018 paper analyzing dozens of studies on the program found, on the whole, “minimal to no impacts” of the courses on college outcomes. Students who earned a score of 3 or above on the tests had better college outcomes, the paper found, but the impact was small.
In another study, a team led by Dylan Conger, a professor at George Washington University, conducted a randomized controlled trial, the gold standard of social science research. It looked at 1,800 students, largely low-income, who were randomly selected to enroll in A.P. science classes at schools new to the program.
Those students became more interested in science than peers, but experienced “lower confidence in their ability to succeed in college science, higher levels of stress and worse grades,” the study reported. Another paper, published this year, found the A.P. students were no more likely to enroll in college, and 8 percentage points less likely to attend a competitive college.
Jessica Howell, vice president for research at the College Board, emphasized in an interview what she called the small scale of the Conger study, and noted that it measured different outcomes than the Board’s internal research, which focused on college grades.
She also defended the Board’s restrictions on releasing papers and data, noting that her team occasionally submits its work for peer review and invites select outside scholars to collaborate.
“These are data we take great care with,” she said.
Some studies have shown that competitors may get better results. A federal analysis comparing Advanced Placement with the International Baccalaureate program found that in 2014, 61 percent of I.B. students from low-income families earned high enough scores to earn college credit, compared with 39 percent of A.P. students from low-income families.
The final scores for I.B., unlike the vast majority of A.P. courses, are partially determined by projects and presentations.
There are also dual enrollment programs, in which high school students take courses developed by local college faculty. College credit is generally based on class grades and performance, not a standardized test.
The College Board holds assets that are akin to a wealthy university — approximately $2 billion at the end of 2022, including more than $200 million in offshore accounts. Mr. Coleman, the chief executive, made $2.1 million last year.
Because of the way A.P. tests are funded, government agencies do not have an easy way to track the total amount of money that flows into the program. The Department of Education said it did not track total federal spending on the program.
Of the $90 million the College Board said it received in government funding for A.P. this year, approximately $37 million goes toward low-income students’ exam fees. The other $53 million came from states that pay for tests in certain subjects, like science or math, for all students.
Government funding for A.P. tests accounts for roughly 18 percent of the program’s total revenue. But in a written statement, the College Board called this funding stream “minimal” relative to its overall operating revenue.
“We believe these are worthwhile investments in preparing all students for their futures,” the organization said. “We recognize that all students are not yet receiving equitable preparation for A.P. coursework, and that such work requires addressing inequities that occur years prior.”
Legally, the College Board does not have to state its total government revenue on its tax forms. That is because test fees are considered payments for services, part of the nonprofit’s stated mission to help students succeed in college, said Erica Harris, an accounting professor at Florida International University.
Still, the lack of transparency raises important accountability questions, said Norman Silber, a professor of law at Hofstra University and expert on nonprofits.
Lobbying government agencies to pay for tests, while not revealing the extent of those payments, allows the College Board to more effectively resist any public pressure to lower fees, he said.
College Board executives are well aware that growing skepticism toward standardized testing could imperil the A.P. revenue stream.
In an acknowledgment of the intractable class and race disparities in A.P. test scores, some of the board’s newest courses culminate not in tests, but in assessments that include projects or presentations.
The College Board expects students to pay fees for these project-based assessments, too.
Stephanie Saul and David A. Fahrenthold contributed reporting. Sheelagh McNeill contributed research.