EDITORIAL WHEN TESTS FAIL.

Defects in the API make it hard to gauge the status of state schools

IF the state's Academic Performance Index is any indication, California schools are showing some long-overdue signs of improvement.

That's encouraging, but we wouldn't rush to judgment. The API is just one measure of the state's academic performance - and not a very reliable one at that.

California education officials have reported that 16 schools across the state have been caught cheating on the Stanford 9 test. Sixteen schools out of 6,500 is hardly evidence of massive corruption. But for every school that gets caught, there are no doubt others that don't. And there are still others, to be sure, which is why state Superintendent Delaine Eastin is belatedly recommending tougher policing.

There also are other ways that scores are being boosted without technically violating the law.

In the Coachella Valley, for example, many Spanish-language students simply didn't take the Stanford 9 this year. That produces higher average test scores, but it's no indication of progress on the education front.

Moreover, because the Stanford 9 recycles the same questions year after year, many teachers - whether consciously or not - are teaching kids test answers in advance.

Large financial bonuses, worth as much as $25,000, provide teachers with a powerful incentive to teach ``for the test,'' or compromise the test in various ways.

The result is that test scores are going up, but we'll have to wait a few years before leaping to the conclusion that kids are actually reading, writing or solving math problems better than before.