UCOP methods in SAT study flawed, says UCLA professor

By Noah Grand
Daily Bruin Reporter

A UC report ““ used to argue that the SAT I should be
omitted in admissions ““ is flawed, according to UCLA
statistics professor Richard Berk.

In November, a UC Office of the President study said that the
SAT II is a better predictor of how a student will perform in UC
than the SAT I. University officials are currently debating a
proposal to omit the SAT I from UC admissions.

Berk presented his findings at a meeting of UCLA faculty and
administrators Wednesday to help them formulate a recommendation to
the systemwide Academic Senate, which will help determine the fate
of the SAT I.

Berk tried to verify the UCOP study ““ conducted by Saul
Geiser and Roger Studley ““ by repeating it. He said he found
multiple problems with the way data was gathered and processed.

“You can’t justify dropping the SAT I based on the
UCOP study,” Berk said at a seminar Wednesday.

The Geiser and Studley study looked for a correlation in how
76,000 freshmen entering the UC between 1996 and 1999 performed
their freshman year of college, compared to their high school GPA,
SAT I and II scores, and socioeconomic status.

According to the study, the SAT II is a preferred test because
it better reflects high school curriculum and is based less on
socioeconomic factors, such as accessibility to test preparation
courses.

Geiser, the lead researcher in the UCOP study, said their method
was the “best way” to conduct such a study and that it
was “the standard operating procedure in the field”
when the study was released.

Neither Geiser, nor any UC officials, were available for
comment.

Berk said that Geiser and Studley used a biased sample because
they only studied students admitted to UC.

He suggested that better data could be gathered from a random
sample of people who apply to the UC and tracking their performance
in college, even if that college is not a UC.

He also said freshman GPA may not be the most accurate measure
of college success.

The other main problem is that a person’s scores on the
SAT I are highly correlated with their scores on the SAT II, making
it very difficult to distinguish if one is better than the other,
Berk said.

“The correlation between the SAT I and SAT II means you
can’t make any conclusions about either one with this
evidence,” Berk said.

His analysis did confirm some of the UCOP report’s
findings about ethnicity, however. Both studies found that African
American and Latino students had lower high school GPAs and test
scores than the average UC student.