Until the president of the University of California catapulted it to fame, most
people had not even heard of the SAT II. President Richard Atkinson announced last
February that he wanted to eliminate the use of the SAT I: Reasoning Test-the test we
used to know as the Scholastic Aptitude Test-in UC admissions. He recommended that
tests be developed which are directly tied to the college prep courses required of UC
applicants. In the short term, he proposed an increased role for the SAT II in admissions
decisions.
While the SAT I focuses on general verbal and mathematical skills, the SAT II: Subject
Tests (formerly the College Board Achievement Tests) are intended to assess high school
students' knowledge in particular areas. More than 20 SAT II tests are available, including
writing, American History, math, physics, Spanish and Chinese. Like the SAT I, the SAT II
exams are primarily multiple-choice, although the writing test does include an essay.
UC applicants must take the SAT II math and writing tests, as well as a third test of
their own choosing. Currently, these three SAT II scores combined count twice as much as
the SAT I in determining eligibility for admission to the UC system. (Each UC campus
imposes its own individual admissions criteria as well.)
From Atkinson's point of view, the SAT II "begins to approximate...an appropriate test
for UC and other American universities since it
tests students on specific subject areas that are
well defined and readily described." A preliminary
report from the UC Office of the President
labels the SAT II "a fairer test for use in
UC admissions than the SAT I."
Any mechanism for sorting individuals inevitably
attracts criticism, however, and now that
an expansion of its role is being considered, the
SAT II is drawing fire. SAT II scores, like SAT I
results, show large gaps among ethnic groups,
and critics point out that performance on subject-area tests may be especially vulnerable to
disparities in instructional quality.
But it is the SAT II language tests that are drawing the most attention. Native speakers
of the languages included in the SAT II program have an opportunity to boost their
chances of admission by choosing a language test as their third SAT II. This could benefit
Latinos, a severely underrepresented group in academia, and possibly Asian Americans.
But what about African Americans, who often attend poor schools and typically can not
reap the second-language advantage? UC Regent Ward Connerly, best known as father of
California's Proposition 209, which banned public-sector affirmative action, told the San
Francisco Chronicle recently that putting a heavy emphasis on the SAT II disadvantages
black students. "It seems to me that we really have built in a cultural bias," he said.
Before UC decides whether to eliminate the SAT I and grant a correspondingly larger
role to the SAT II, it may be useful to take a closer look at two key issues. First, what would
be the effect of such a move on the composition of the freshman class? Second, what are
the implications of using a (non-English) language test in admissions decisions?
How would a greater role for the SAT II affect freshman class composition?
Fortunately, we can do more than conjecture about the relation between the SAT I and
the SAT II. Two recent analyses-one by the UC Office of the President and one by the
College Board-found a very strong association (a correlation of .84) between the SAT I
(verbal and math combined) and the SAT II (a composite of all SAT IIs taken by the
applicant).
The College Board study, by Brent Bridgeman, Nancy Burton and Frederick Cline, used
data from 14,000 students at ten colleges, including four UC campuses, to compare the
effects of implementing various admissions
criteria. High school grades were used in combination
with either SAT I or SAT II scores to
select the top two-thirds of candidates, who
constituted the hypothetical freshman classes.
(Although the admissions criteria were applied
to students who were, in fact, already enrolled in
college, the analyses are nevertheless informative
about the effects of alternative screening
processes.)
An admissions model that used SAT I scores
was compared to a model that used scores on all
SAT IIs taken by each candidate. (Both models
also included high school grades. All test scores
received equal weight, and grades received as
much weight as all test scores combined.) For 86
percent of the students, admissions decisions
were found to be the same, regardless of which model was used. Not surprisingly, the
overlapping "freshman classes" produced by each model were very similar in terms of
average freshman grades and ethnic composition. The SAT II model yielded a class with a
slightly greater percentage of Latino students (eight percent versus 6.3 percent), and a
slightly smaller percentage of Asian American and white students, than the SAT I model.
The percentage of African American students was the same for both models (1.6 percent).
The finding that the SAT I and SAT II produce similar freshman classes is not without
precedent. In their 1988 book, "The Case Against the SAT," James Crouse and Dale
Trusheim compared admissions models involving the SAT I and the SAT II (under its
earlier name), and concluded that "in most colleges, scores on ETS's achievement tests are
almost interchangeable with SAT scores."
Should language tests be used in admissions decisions?
In the College Board study, using the SAT II Spanish test in admissions decisions was
found to increase the percentage of Latinos in the freshman class. (No parallel effect
occurred for Asian Americans, but only eight percent took Asian language tests, while
nearly half the Latino students took the Spanish test.) When language tests were excluded
from consideration, the SAT II admissions model yielded no more Latinos than the SAT I
model.
This study suggests that including the SAT II language tests in admissions decisions may
have potential as a strategy for increasing the Latino presence on campus (though it does
nothing to alleviate the under representation of other groups, such as African Americans).
But merely increasing Latino representation is not the intent of the University of California,
according to a letter to the San Francisco Chronicle from C. Judson King, provost
and senior vice president for academic affairs. "The change has nothing to do with
benefiting particular groups and everything to do with measuring academic potential," he
said.
What does the College Board study say about the academic success of Latino students
selected with and without using Spanish test scores? An analysis that focused on Latino
students who took the Spanish test revealed that those selected using the "SAT II with
Spanish test" model were less likely to receive freshman GPAs of at least 2.5 (the
researchers' definition of academic success) than those selected using the SAT I model (71
percent versus 82 percent). The "SAT II without Spanish test" model yielded an intermediate
success rate of 76 percent. (This figure was derived by combining the authors'
results for Mexican American and "Other Latino" students.)
Why would inclusion of the Spanish test in the screening criteria yield a lower success
rate for the admitted students? The likely reason is that most Latino students who took the
Spanish test learned Spanish at home, rather than through formal academic training. Skills
acquired at home, obviously, are no less valuable than skills acquired at school. But when a
language is learned in the natural course of family interaction, skill in that language is less
informative about students' academic skills and therefore less useful in predicting subsequent
academic success.
An interesting irony emerges in considering the role of language tests in the Atkinson
proposal. On one hand, the SAT II is described as more equitable than the SAT I because
it focuses on content that is learned in school. On the other hand, students who take tests in
their native languages are rewarded for knowing material that they did not learn in the
classroom, material that is instead a part of their culture. This situation is oddly reminiscent
of the familiar claim that white middle-class students have a built-in advantage on
standardized tests because their own culture is ingrained in the tests.
Of course, using language tests in admissions may nevertheless be a good thing. First of
all, many SAT II takers have learned second languages in the classroom. In any case,
maximization of the expected grades of the freshman class should certainly not be the sole
purpose of an admissions policy. According to UC Provost King, "UC faculty has deemed
mastery of a second language to be an important skill for college-bound students." Indeed,
bilingual students can enrich the college environment, and a policy that rewards bilingualism
can encourage second-language learning, which is useful in many careers and in
community service.
SAT II Pros and Cons
Let's consider what we know about the pros and cons of eliminating the SAT I in favor
of a greater role for the SAT II.
First, because the tests are closely related, a
heavier reliance on the SAT II is unlikely to
have much impact on the academic competence
and diversity of the freshman class,
or on the score disparities among ethnic
groups.
Second, exams that have demonstrable ties to
school learning are often considered more
equitable than "aptitude" tests, so even if
they don't help colleges to pick better entering
classes, they could have an advantage
from a public-relations standpoint. After all,
UC President Atkinson opposes the SAT I
partly because it is perceived as unfair and is
viewed as being "akin to an IQ test." On the
down side, the SAT II exams are not tailored
to the curriculum of any particular school
district and will inevitably include material
that is not taught-or is poorly taught-in some schools. This potential source of
inequity is sure to draw more attention if the prominence of the SAT II is increased.
Third, the inclusion of language tests in admissions criteria needs to be evaluated
carefully. Critics have suggested that this strategy is intended as a backdoor affirmative
action measure; its fairness and effectiveness for this purpose are doubtful. But if UC
considers the recruitment of a bilingual freshman class to be valuable in itself, then
including these tests is justifiable.
Finally, another rationale for relying on subject-based admissions tests is that it may
improve classroom teaching. Ideally, if admissions tests focused on course content, the
much-deplored practice of "teaching to the test" could become indistinguishable from
just plain teaching. This is perhaps the most compelling argument in favor of tests like
the SAT II.
