Standardized testing plays an important role in reform, but it is not a silver bullet. A statewide testing program informs parents and legislators and exposes schools that are performing poorly. Unfortunately, most of the gain comes when the information is used not merely to inform debate but to give schools incentives to improve. Parents can create such incentives by exercising school choices informed by test results. Alternatively, a state can create a system of penalties or rewards for schools. School choice, however, is rare, and reward systems are hard to construct.
For years, I was a test skeptic because I thought that the benefits from testing (without accompanying incentives) were too small to justify the cost. I was wrong. Not only had I underestimated how much schools would improve to avoid being exposed, but I had not realized how inexpensive testing is. Testing is undoubtedly the school reform with the highest ratio of benefits to costs.
It costs about $2.50 per student to buy, grade, and publish a full battery of standardized tests, such as the Stanford 9 exams. California, for instance, is currently reimbursing districts at the rate of $2.52 per student for the cost of its testing program. Even if states were to pay proctors, the total cost would be under $4.50 per student. Just so that no one quibbles about a few cents here or there, let's round up to $5 per student for annual standardized testing. What else could we get for that amount?
We could reduce class size by two-thousandths of a student. You could raise teacher salaries by 0.24 percent—about one-quarter of 1 percent. While incumbent teachers would not mind a raise, such a tiny increase will not draw talented people into teaching or retain a teacher with good outside job opportunities. You could keep students off the streets and involved in after-school activities for almost two hours per year, assuming that the activities are cheap. Although children need after-school supervision and stimulation, part of one afternoon will not help much.
You could lengthen the school year by one-tenth of one day, but will one-tenth of one day really make a difference? You could give every child a college scholarship that would pay for about 1.5 ten-thousandths of the cost of a typical American four-year college education ($32,000). To be fair, the child will be in twelve grades that use standardized tests before he reaches college, so you could add up the cost over those twelve years. The total would be a scholarship worth 1.8 one-thousandths of what a typical education costs.
Because the federal government accounts for about 7 percent of spending on elementary and secondary schools, it is often hard for federal policymakers to affect schools. As a rule, the federal government simply does not have enough money on the table to induce state and district officials to enact policies that they don't already want to enact. Testing, because it is so inexpensive, may be the exception to the rule. The current federal push for testing just may work.