Front Page
  Current Issue
  Back Issues
  About National CrossTalk

National CrossTalk Fall 1999
News Editorial Other Voices Interview

1 of 2 Stories

Making the Grade
The SAT versus the GPA

By Rebecca Zwick

In a 1995 study, researchers from the National Center for Education Statistics calculated the proportion of college-bound high school seniors who met academic standards typically used by selective colleges in making admissions decisions.

The analysis produced some striking findings about the percentage of students with grade point averages of at least 3.5: Twenty-nine percent of Asian Americans, 21 percent of whites, ten percent of Latinos, and four percent of African Americans had GPAs this high, a hefty difference among ethnic groups. Researchers also examined students of high, medium and low socioeconomic status, or SES (measured by parental education, occupation and income). Twenty-four percent of the high-SES group, compared to only ten percent of the low-SES group, was found to have GPAs of at least 3.5.

So, has the high school GPA been criticized as a racially biased "rubber ruler," or condemned as an index of family wealth? Have prominent business leaders, the Urban League or the president of the University of California recommended that grades be de-emphasized or even eliminated as a college admissions criterion? The answer, of course, is no. The large and persistent differences in GPA among ethnic and socioeconomic groups have produced no such demands.

Now consider the Scholastic Assessment Test (SAT). In February 2001, the president of the University of California, Richard Atkinson, delivered a speech advocating the elimination of the SAT as a criterion for admission to the university. According to him, reliance on test scores is "not compatible with the American view on how merit should be defined and opportunities distributed." Moving away from quantitative admissions formulas, he said, "will help all students, especially low-income and minority students, determine their own educational destinies."

Although score disparities among socioeconomic and ethnic groups received only a whisper of a mention in Atkinson's speech, these well-known score gaps featured prominently in the reverberating reactions to his proposal. In The Chronicle of Higher Education, for example, test critic Alfie Kohn suggested that the verbal section of the SAT, which includes some difficult vocabulary, merely measures "the size of students' houses." "Call it the Volvo effect," said journalist Peter Sacks, "a boost that peaks out at the highest levels of family income. Being white, on average, confers an extra 200-point advantage over a black test-taker." These remarks echo a recurring theme in discussions of standardized admissions tests, the idea that test scores vary across socioeconomic and ethnic groups because of some intrinsic property of the tests, either their content or their susceptibility to pricey test coaching.

It is indisputable that SAT scores vary among socioeconomic and ethnic groups. College Board data from 1997 show that the average combined SAT score (math score plus verbal score) for college-bound seniors whose families earned less than $20,000 was 908. The SAT average increased steadily with family income, reaching 1132 for students whose families earned over $100,000. And in 2000, the spread between the highest- and lowest-scoring ethnic groups was 95 points on the verbal section of the SAT and 140 points on the math section.

The 1995 National Center for Education Statistics study also found that SAT scores were strongly related to both socioeconomic status (SES) and race. Thirty-two percent of the high-SES students versus nine percent of the low-SES group were found to have combined SAT scores of at least 1100. And the percentage of students with SAT scores this high was largest for Asian Americans (28 percent), followed by whites (25 percent), Latinos (eight percent) and African Americans (three percent).

It is just these sorts of findings that have led test critics to claim, as did the watchdog group FairTest, that "the SAT is very effective at eliminating academically promising minority (and low-income) students…" The National Center for Education Statistics study provides an unusually good opportunity to examine this charge. A close look at the results shows that for each ethnic group, the percentage meeting the GPA standard is strikingly similar to the percentage satisfying the SAT criterion. Twenty–nine percent of Asian Americans met the GPA standard, 28 percent met the SAT standard; four percent of African Americans met the GPA standard, three percent met the SAT standard, and so on.

What the GPA and the SAT have in common, of course, is that they are indexes of previous achievement and, therefore, reflect past inequalities in educational opportunity. In a recent analysis of the achievement gap which appeared in The Nation, Harvard professor Pedro Noguera and co-author Antwi Akom made the point that "explaining why poor children of color perform comparatively less well in school is relatively easy: Consistently, such children are educated in schools that are woefully inadequate on most measures of quality and funding. This is particularly true in economically depressed urban areas, where bad schools are just one of many obstacles with which poor people must contend."

What, then, do these findings say about the likely effect on campus diversity of abandoning the SAT as an admissions criterion? The best evidence comes from the University of California. This very topic was addressed in a December 1997 report issued by the university president's office, based on supplementary analyses of data from a study conducted by the California Postsecondary Education Commission. Transcripts, test scores and background information from a random sample of more than 15,000 students who graduated from California public high schools in 1996 were analyzed to determine the effect of applying various admissions criteria. In particular, the study considered the impact of eliminating standardized admissions test requirements on the rates of UC eligibility, which at that time was based on the completion of certain college preparatory courses, GPA for those courses, and, if the GPA were below a certain level, scores on the SAT or ACT. (Students who are judged UC-eligible are then subject to the admissions criteria of the individual UC campuses.)

The study's conclusion was surprising to some: Elimination of the admissions test requirement, when combined with other mandated features of UC admissions policy, would produce very small changes in the eligibility rates for Latinos (from 3.8 percent to four percent), African Americans (from 2.8 percent to 2.3 percent) and Asian Americans (from 30 percent to 29 percent). The largest change would be an increase in the eligibility rate for whites (from 12.7 percent to 14.8 percent).

The minimal change in the predicted eligibility rates for African American and Latino students in the California study is less remarkable in light of the finding that "low test scores rarely are the only reason for a student's ineligibility." In fact, fewer than three percent of California public high school graduates were ineligible solely on the basis of inadequate admissions test scores. (It is perhaps ironic that this key piece of information comes from the office of the same university president who has become an instant hero of the anti-SAT movement.)

Most students, 63 percent of graduates overall, were ineligible because they had major course omissions or grade deficiencies, or because they attended "schools that did not have a college preparatory curriculum approved by the University." The percentage of students in this ineligibility category was much higher for African Americans (77 percent) and Latinos (74 percent) than for whites (59 percent) and Asian Americans (39 percent).

So, if we were to condemn any admissions criterion that varied among ethnic and socioeconomic groups, we'd have to cross high school grades and course background off our list along with test scores. It's a sad reality that educational disparities in our country make it unlikely that we could find any reasonable measure of educational achievement that was unrelated to ethnic or socioeconomic status. Discrepancies in high school grades, admissions test scores and course background are reflections of the same educational system with all its flaws and inequities.

What are the implications of this situation for admissions policy? First, any policy that relies heavily on traditional evidence of past academic achievement, whether it be test scores, grades or course completion, will tend to perpetuate societal inequities. So, in the short term, attaining campus diversity requires an admissions system that places significant weight on other factors. Of course, the most effective means of achieving ethnic diversity is through explicit affirmative action programs, but this is no longer legally possible in many parts of our country.

We need flexible admissions polices that allow for the inclusion of a wide range of student characteristics. Among the student attributes that warrant further investigation are motivation, perseverance and "spike talents" in particular areas. Such factors are already considered at many schools, but we need more research to determine how best to measure these characteristics and to assess their predictive value.

Second, the focus in the longer term must be on improving K–12 education for all students. The evidence of inequalities in academic preparation among ethnic and income groups is overwhelming. These achievement gaps are not surprising in light of the enormous variation in the quality of K–12 education across districts and states, as reflected in teacher-pupil ratios, teachers' educational level and years of experience, and school resources. According to an Education Week survey, for example, state per-pupil expenditures for 1999 ranged from about $4,000 to about $8,700, after adjustment for regional cost differences. Even more striking is the range in spending among districts within a state, which exceeded $3,000 in several states, according to 1997 calculations.

Finally, although public scrutiny of high-stakes tests is certainly appropriate, we don't need to point to the difficulty of vocabulary questions or to the cost of test coaching to explain the differences in SAT results among ethnic and income groups. While ignoring test score disparities would be unforgivably irresponsible, attributing the score gap to an inherent property of tests has its own dangers, the diversion of time, energy, editorial space, and resources that could otherwise be used toward the improvement of educational opportunity for all children.

E-Mail this link to a friend.
Enter your friend's e-mail address:



National Center logo
© 2000 The National Center for Public Policy and Higher Education

HOME | about us | center news | reports & papers | national crosstalk | search | links | contact

site managed by NETView Communications