ETS Accused of Squelching New Approach on Racial Bias

Plan was an alternative to using race in admissions; testing service denies charge

Article tools

Printer friendly	E-mail article	Subscribe	Order reprints
	Discuss any Chronicle article in our forums
Latest Headlines Struggling Towns Turn to Small Private Colleges In economically troubled areas, small private colleges help generate development projects in large part as a matter of survival. Opinion: Even in Hard Times, Colleges Should Help Their Communities Endowment-Spending Talk Draws a Crowd of Tax Lawyers Learned Societies' Gathering Delves Into Political and Publishing Challenges Researchers Look Beyond Genetics to Help Doctors Predict Disease Militia Battles in Lebanon Rewrite Universities' Lesson Plans

By PETER SCHMIDT

A prominent higher-education researcher says scholars at the Educational Testing Service may have discovered a substitute for race-conscious college-admissions policies back in 1999, but their research project was suppressed — and eventually killed off — before they could put their findings through peer review and make them public.

A spokesman for the testing service says the study was dropped because it was "bad research."

Anthony P. Carnevale, a former vice president for assessment, equity, and careers at ETS, says he and other ETS researchers concluded in the summer of 1999 that it was theoretically possible for selective colleges to maintain or increase their black and Hispanic enrollments without giving extra consideration to applicants based on their ethnicity or race.

The researchers had developed a formula for using students' background data to identify "strivers" — those who had overcome adversity to an impressive extent — and had fine-tuned the formula to a point where it showed the promise of producing larger black and Hispanic enrollments at selective colleges than were being obtained through race-conscious admissions, says Mr. Carnevale, now a senior fellow with the Education Sector, a think tank in Washington, D.C.

Mr. Carnevale alleges that College Board officials put pressure on ETS to squelch the entire "striver" line of research, mainly because it added a new layer to the interpretation of SAT scores and they feared it would give federal courts reason to question colleges' need for race-conscious admissions policies.

Because the research was squelched midway, the researchers never got a chance to determine conclusively — and then demonstrate to ETS and the College Board — that they had found what they were looking for: a way to achieve racial and ethnic diversity at selective colleges without using affirmative action.

"The work never saw the light of day," Mr. Carnevale says.

In an e-mail message last week, Thomas Ewing, a spokesman for ETS, denied Mr. Carnevale's account of what transpired, saying "there was no pressure from the College Board to discontinue" the striver study. He said the study had been discontinued because "it was widely viewed at ETS as simply bad research," and the president of ETS, Kurt M. Landgraf, and the ETS research staff thought it "attempted to alter an objective measure (the SAT) inappropriately."

But some education researchers who were not involved in the strivers study said last week that they viewed the research as sound, and would like to see it continued.

"While there may be some valid criticisms, none of them suggest that the work as a whole is not meaningful and valid," said Roger E. Studley, who has done similar studies as assistant director for admissions research and evaluation in the office of the president of the University of California system. "I think it is perfectly good stuff."

Questions of Fairness

Although ETS has long administered the SAT, the College Board owns the rights to it. When newspapers first reported on the existence of such a study, in August 1999, Gaston Caperton, who had just been installed as president of the College Board, responded by assuring colleges that there was no new system for interpreting SAT scores, "only research," and that he would ensure that no such system came into use.

Last week Wayne J. Camara, the College Board's vice president for research, said that "we thought the research was shoddy" and that using such a method for interpreting SAT scores "would not only be scientifically indefensible but would raise many fairness issues."

Mr. Camara argued that the researchers involved in the striver study were proposing to adjust SAT scores based on such factors as the levels of poverty in a student's school or neighborhood — an approach he called both "unfair" and "imprecise" because, for example, students from low-income families could be enrolled in wealthy high schools. Mr. Camara also expressed concern about the negative impact the application of a striver formula would have had on students from wealthy families, who enjoyed many advantages while growing up.

"There was no way for them to go but down," Mr. Camara said. "No matter what they did, it would never be good enough."

Mr. Carnevale says he and others involved in the research were not talking about adjusting SAT scores, but, instead, were trying to find a way to determine when a student had encountered such disadvantages in life that his or her test score was much more impressive than it appeared on its face. Among the factors they took into account were the incomes and education levels of the student's parents, the number of books in the household while the student was growing up, whether the student's high school was inner city or rural, and whether the student attended a school that offered rigorous academic courses.

"This is pretty established social science," Mr. Carnevale says. "We did not do anything nutty."

In fact, in a February 2003 paper, "Inequality, Student Achievement, and College Admissions: A Remedy for Underrepresentation," Mr. Studley, who was then coordinator of research and evaluation in the office of University of California system's president, reviewed the ETS striver research that had been made public. "It was not altering scores," he said, "and was, in fact, advocating a practice in which colleges have long engaged: evaluating achievement in light of individual circumstance."

Mr. Studley described how he had developed his own theoretical admissions model, similar to the one that the ETS researchers had been working on, and concluded that it could "significantly increase the representation of socioeconomically disadvantaged and minority students" in his university system, which was precluded from considering race in admissions as a result of a constitutional amendment passed by Californians in 1996.

Mr. Carnevale says he and the other striver researchers did not claim to have developed a workable new system for evaluating applications, ready for colleges to use. They were seeking only to show that the development of such a system was possible.

"Would I use our study to pick a kid? Hell, no," Mr. Carnevale says. But, he says, "there was a broader point here."

"What we were saying here was there were a lot of kids out there who achieve above expectations who tend to be lower income," he says. "What we were saying, in the end, is that you ought to look at these kids."

The basics of Mr. Carnevale's account of what the striver research had produced were confirmed last week by Jeff Strohl, a research economist who worked at ETS from 1996 through 2001 and helped Mr. Carnevale with the study. Mr. Strohl said those involved in the project were "pretty confident" they had found a workable replacement for race-based affirmative action, and he has been able to refine their formula by doing additional work on his own about the identification of "strivers."

Mr. Carnevale says ETS officials first authorized him to undertake the striver research in late 1996. At the time, he says, he was pushing for such an undertaking at the urging of Clinton-administration officials, who feared that colleges' race-conscious admissions policies would not survive the challenges being brought against them in the courts and at the polls.

At a 1997 meeting of the American Association of Collegiate Registrars and Admissions Officers, Mr. Carnevale and another researcher in the strivers project, Elhum Haghighat, summarized their findings to that point. So far, they said then, they had not yet developed a formula that was as effective as race- or ethnicity-conscious admissions in producing campus diversity.

In the summer of 1999, however, the researchers began testing the hypothesis that colleges could produce then-current levels of diversity without considering ethnicity or race if they used a striver formula that took into account accumulated family wealth. That variable showed promise because, at the median-income level, white families have at least three times as much accumulated wealth as black families.

Mr. Carnevale ran student profiles through a version of their formula that had been revised to take into account how much the students' families had saved for college, and found that, with the addition of the family-wealth measure, black and Hispanic enrollments would be actually higher than they had been under affirmative action with racial preferences.

The striver researchers at ETS regarded family college savings as an imperfect variable. The chief problem was that the amount of savings was self-reported. But they were optimistic that they could develop a workable system if they crunched their student data through versions of the formula that considered other family-wealth measures, such as homeownership.

Race vs. Class

Given the preliminary nature of such findings, Mr. Carnevale did not mention them when he was interviewed for a Wall Street Journal article on the striver research published in August 1999. The article nonetheless caused a stir among colleges by suggesting that ETS would soon be promoting striver methodology, and officials of both the College Board and ETS scrambled to offer assurances that the research was preliminary.

At a March 2000 meeting involving representatives of ETS and various higher-education associations and civil-rights groups, it became clear that many on hand were concerned about what impact the striver research might have on the battles over affirmative action in higher education that were being fought in the federal courts. The flagship universities of Georgia, Michigan, Texas, and Washington had all become targets of lawsuits challenging their race-conscious admissions policies, and the fear was that the U.S. Supreme Court, which seemed destined to review one of the cases, would strike down such policies if it was convinced that viable race-neutral alternatives were available.

"There were concerns about a backlash use of the study," Nancy S. Cole, who was then president of ETS, recalled this week.

Theodore M. Shaw, who attended the meeting as associate director of the NAACP Legal Defense and Educational Fund (and who is now the organization's president), said, "I know what my concerns were then because they would be my concerns now." He said that "class inequality and race inequality are not completely identical issues." Because even many black students from upper-income families lag behind their white peers, he fears that the replacement of race-based affirmative action with class-based affirmative action would result in many such black students' being shut out of selective colleges.

"In absolute numbers," Mr. Shaw noted last week, "there are more poor, white students in this country than there are poor, black students."

Richard D. Kahlenberg, a senior fellow at the Century Foundation who has extensively studied and advocated class-based affirmative action, said he thinks "a fair reading of the social-science evidence is that you get a lot of racial diversity" through the types of affirmative action that the striver study was examining. But, he said, "as a political matter, you end up helping low-income and working-class African-Americans and Latinos, whereas race-based affirmative action tends to help upper-middle-class African-Americans and Latinos."

"Traditionally," Mr. Kahlenberg said, "the strongest constituencies within the civil-rights community have been middle-class and upper-middle-class people of color."

The striver study did enjoy the support of some prominent black scholars who were not on hand for the meeting, however. They included Lani Guinier, a professor of law at Harvard University, and Christopher F. Edley Jr., now dean of the law school at the University of California at Berkeley, who at the time was one of President Clinton's top advisers on matters of race.

"Anything we can do that makes access to elite higher education more than a device whereby the privileged get to perpetuate themselves, that is a real service," Mr. Edley said last week.

Ms. Cole and Mr. Carnevale assured those present at the March 2000 meeting that any additional findings from the striver research would be discussed with them before publication, and Mr. Carnevale and his colleagues continued to work behind the scenes. Until she retired from ETS, in August 2000, Ms. Cole remained supportive of the striver research, despite repeated objections to it from officials at the College Board.

In an interview last week, she said she thought Mr. Carnevale "had some contribution to make" and seemed close to finding a new route to achieving campus diversity. At the same time, she said, she found herself having to play the role of intermediary between the striver researchers and those who objected to the study, and she thought Mr. Carnevale "believed it could be a solution beyond what even I thought it could be."

"It was very controversial," Ms. Cole said. "This country does not discuss class very well."

Ms. Cole was replaced as president of ETS by Mr. Landgraf, a former chairman and chief executive officer of DuPont Pharmaceuticals, who was much less supportive of the striver study. Mr. Strohl, the research economist at ETS, said those involved in the research began to hit resistance from within ETS "in the form of obscure technical criticism." Meanwhile, he said, ETS leaders were not offering any technical support from the nonprofit organization's staff of hundreds of researchers to help overcome the flaws being identified. The striver research fizzled out over the course of the following year.

At a March 2003 news conference discussing a separate, Century Foundation-financed study of class-based affirmative action, Mr. Carnevale stressed that the hypothetical admissions model being described should be thought of as a complement to, rather than a replacement for, race-based affirmative action because it could not produce as much racial and ethnic diversity as the types of policies then being challenged before the U.S. Supreme Court.

In hindsight, Mr. Carnevale says, he shaded the truth: The model may have produced enough diversity if it was tweaked to include family wealth, but he could not make such an adjustment without invoking the striver research against the wishes of his bosses at ETS.

Mr. Carnevale left ETS in 2004. He says he is talking about what happened with the striver research now in hopes of attracting financial support to resurrect the study and develop a striver formula that college admissions offices could use effectively.

Same Outcome?

Writing for the majority in the U.S. Supreme Court's 2003 decision to uphold race-conscious admissions at the University of Michigan's law school, Associate Justice Sandra Day O'Connor accepted the university's argument that it had considered race-neutral alternatives and none was adequate.

Race-conscious admissions remain a divisive issue, with Michiganders voting this week on a ballot initiative that would ban the use of racial, ethnic, and gender preferences by state agencies, including public colleges. Nonetheless, several prominent critics of affirmative action said last week that they did not see the use of striver formulas as a viable or acceptable alternative to race-conscious admissions policies.

"I doubt that the strivers research would have changed the outcome of the University of Michigan cases, nor should it have," said Terence J. Pell, president of the Center for Individual Rights, which helped represent the plaintiffs suing the University of Michigan. He said those involved in the study seemed "interested only in finding a way to get minority students into college by any means necessary, regardless whether the means promoted or hindered later minority academic achievement."

Ward Connerly, founder of the American Civil Rights Institute and a leader of the campaign to ban preferences in Michigan, said any class-based policy that seeks to give black and Hispanic applicants an edge in admissions "is almost the same thing as affirmative action" and offends him as discriminatory.

Roger B. Clegg, president of the Center for Equal Opportunity, mainly objected that ETS had been consulting advocates of affirmative action in deciding what research to publish, and when. "The idea that ETS's outcomes were being manipulated by people with a political agenda is, I think, extremely disturbing," he said.