It should be noted that the linkage between expenditure levels and student performance, which is usually measured by some standardized test, is tenuous at best. In fact, if you look at the performance of the average pupil in a state, versus that state's spending per pupil, the relationship is slightly negative. In the 1997-98 academic year, for example, New Jersey spent over $10,000 per pupil, New York about $8,800, and quite honestly, on average, their students do poorly compared to places like North Dakota, which spent $5,100 per pupil and whose students do very well. But spending varies all over the map—Wisconsin spent $7,100 per student and Iowa $6,000, and their students are also top performers. California spent $5,500 per pupil; the national average was $6,000.

When you look at California's test scores correctly, in my view, the data are not that discouraging. Since the early '60s, the proportion of high-school seniors taking the SATs and the ACTs has gone way up. These used to be elite tests that only the college-prep kids, a small minority in each class, would take. Now nearly everyone takes them. So for as many stu-dents to be doing as well as they are is actually, by some definitions, a remarkable accomplishment. When you break down SAT scores by ethnicity, you see that over the last 20 years, there has been„not so much among white students, but among black and Latino students„a very steady and impressive improvement.

Expenditure data can also be misleading. For example, the L.A. Unified School District employs one of the largest police forces in the state of California. That money shows up on the books as educational spending. And a few years ago there was a proposal to require that 90 percent of the education budget had to be spent in the classroom and not on administration. I can't conceive of that affecting what a school district does. You just call stuff "classroom" and not "administration." So you really have to be careful with these data.