Five Challenges to the S.A.T.
Next week, more than 1.7 million high-school students will spend two and a half hours clutching their No. 2 pencils and sweating over words like "perfidy" and "aviary." Like several generations of young people before them, they will be hoping to score high enough on the Scholastic Aptitude Test to win admission to one of the 1,500 colleges that require the test.
The College Board and the Educational Testing Service--the nation's two largest nonprofit testing organizations--have persuaded colleges and universities over the years that requiring sat scores is important to their admissions process. Yet virtually all of the information these organizations have provided colleges about the value of the test has been highly misleading.
From six years of research on the sat--drawing on three national surveys, hundreds of studies conducted by colleges, and analyses prepared by ets's own validity study services--we have found that the test does not help colleges improve their admissions decisions. Nor does it help applicants choose schools at which they will be successful or further equality of opportunity for minority and low-income students.
Colleges should rely instead on already available indexes of students' potential--such as their high-school coursework and grades. And new achievement tests that measure mastery of high-school courses should be developed to replace the sat
We call on ets and the College Board to show that the sat can meet five simple challenges:
Challenge 1: Do sat scores change enough admissions decisions to justify colleges' use of the test?
ets's own data in fact indicate that colleges would make identical decisions to admit or reject for at least 84 percent of their applicants with an admissions policy based only on the high-school record as compared with one based on the high-school record and sat scores. Indeed, the percentage is in the mid-90's at many colleges and for many programs within individual universities.
A rate of altered decisions that may fall below 10 percent would not seem to warrant requiring some million and a half students to take the test.
Challenge 2: Do the scores help admissions officers significantly enough even in a small percentage of cases to justify mandating the test?
There is no evidence that the sat appreciably improves decisions in the few cases where it changes them. In fact, according to our analysis, the typical gain to colleges from considering sat scores as well as high-school records is an increase in average freshman grades of only 0.02 on the typical four-point scale.
The improvement resulting from use of the sat is equally small when college graduation is the criterion. Such gains hardly suggest that the test is indispensable for admissions purposes.
Challenge 3: Does the test help high-school students identify colleges whose requirements do not exceed their ability?
For many years, ets and the College Board have told colleges that the sat helps students choose colleges where they can meet the academic demands--and have encouraged prospective applicants to use their scores in this way. Yet the most accurate predictions applicants can make about their academic performance in college by using their sat scores are essentially identical to the conclusions they would reach without considering them.
Our findings indicate that if applicants could make the best predictions possible, their two estimates would differ by an average of only 0.09 of a grade.
Challenge 4: Does the sat promote equal opportunity?
According to ets and the College Board, the sat promotes equal opportunity for blacks and other minorities by making all students "run the same race" even though they may come from widely divergent backgrounds.
If colleges use the same admissions standards for whites and blacks, however, consideration of the sat in addition to the high-school record reduces the number of blacks whom the schools accept--sometimes by more than half--but does not decrease the number of whites they admit.
Nor does use of the test improve admissions decisions for these black students. Adding the sat as a criterion leads to the indiscriminate rejection of potentially successful candidates as well as those unlikely to succeed--exactly the meaning of a test having an adverse impact on black applicants.
Even if a college admits a specified number of blacks or low-income applicants, the sat does not help the admissions office select the best of these students.
Challenge 5: Will ets and the College Board have the courage to organize an independent blue-ribbon panel to study alternatives to the sat?
As the creators of the test, they are in a better position than any other group to initiate such a project. And indeed, as nonprofit organizations holding public trust, they have an obligation to do so.
An independent group might consider, for example, whether colleges could with confidence drop the sat and rely instead on applicants' grades in academic courses, class ranks, caliber of the schools attended, and other information to make sound admissions decisions. Neither ets nor the College Board has ever shown that they couldn't.
In fact, several colleges have recently made the sat optional--most notably Bowdoin and Bates. The experience at these two colleges has proved positive. And reports surface periodically that other major universities are considering the same step.
It does not follow, of course, that if colleges could drop the sat and make equally good admissions decisions from applicants' high-school records, then they ought to abandon admissions testing altogether. Other kinds of tests, in fact, may have virtues the sat lacks.
Why not study the possibility of creating new kinds of achievement tests? In many countries, for example, selective universities require tests that measure mastery of secondary-school curricula.
But the sat covers only a very restricted portion of American secondary curricula--vocabulary, reading comprehension, and general mathematics. If selective colleges in this country were to advocate tests that certify mastery of high-school courses instead of the sat, college-bound students might be encouraged to take their academic work more seriously.
We do not think, however, that colleges should simply increase their use of ets's current batch of achievement tests, or support expansion of that program. The existing tests have been criticized for their superficiality and the ease with which they can be coached. Nor does ets, as far as we know, conduct any serious research to find out whether the material tested by its current achievement battery is covered by the courses students actually take.
Indeed, a better model for subject-matter achievement testing--one that tries to ensure a match between what is tested and what is taught--is the College Board's Advanced Placement program. Participating high schools offer courses based on detailed specifications worked out by representatives of both colleges and high schools.
The idea of the ap courses could be extended to college-preparatory courses generally--and to the admissions process. An achievement-testing program comprised of standardized assessments covering specified curricula in a large number of areas--but with tests and courses pitched at varying levels of difficulty--could both encourage greater diversity among high-school curricula and foster among students the notion that taking challenging courses is a good thing. Such a program would also afford admissions officers as much common ground as the sat offers--if not more--for judging applicants from different high schools.
Unfortunately, instead of rethinking what tests like the sat can and cannot accomplish, ets and the College Board typically ignore or try to discredit criticisms of the sat Too often their reactions are defensiveness and politically motivated attacks of their critics.
We should expect more than this from our nation's two most important nonprofit testing organizations. An independent blue-ribbon panel might help to correct the situation.
Vol. 08, Issue 08, Page 28