Who Played With Test Score Facts?

* Ron Unz accuses Times columnist Agustin Gurza of “misrepresenting the facts” on bilingual education (Letters, July 30).

But it is Unz who misrepresents the facts when he concludes that Proposition 227 is responsible for the increase in test scores at San Juan Elementary School.

There are several plausible reasons why Stanford 9 scores went up at San Juan, other than Proposition 227.

One reason is possible selective testing. In cases where test scores increased for the same group of children at San Juan, the number of students tested declined. This was not so in cases in which test scores did not increase.

For example, at San Juan Elementary, fourth-graders scored 18 on the test in 1999 and increased to 26 in 2000. But 143 children took the test in 1999 and only 111 did so in 2000. This raises the suspicion that the excluded children would not have done as well on the tests.

San Francisco Chronicle writer Nanette Asimov has presented data showing that similar test score increases and declines in the number of children tested have taken place in several Bay Area districts.

Another possible reason is there were fewer underserved children. Before bilingual education was reduced at San Juan, the school reported that 28% of the limited-English-proficient population received no services at all. In 1999 it reported that only 3% received no services.

A sheltered immersion plan is far better than doing nothing. Thus, nearly 25% of the limited-English-proficient children at San Juan were better off in 1999 than they were in 1998.

Other possible reasons include intensive test preparation and test score inflation. Test scores generally rise each year with commercial tests. This is why tests need to be recalibrated every few years.

It also needs to be pointed out that bilingual education was not dropped at San Juan. It was reduced. Before Proposition 227, 38% of San Juan children were in bilingual education. After the ballot measure, 14% were in bilingual education. Whatever the effect of reducing bilingual education, it affected only about 25% of the student population.

Most important, Unz’s observations are what researchers call post hoc. They were made after the fact. Of the thousands of schools, this one was picked because its results appeared to conform to a certain pattern.

This is improper scientific procedure. Proper procedure is to do a real experiment, with groups selected before the treatment, not after.

STEPHEN KRASHEN

Professor of education

USC