Test Scores Are Deceptive


About those test scores

By Ernest Dumas
September 27, 2002

 As governor of Texas, George W. Bush showed the country how to reform public education. If your children score poorly on national tests, write an exam that they can do well on, teach it until the kids are whizzes on it, then declare the battle won.

Bush ran for president on the strength of a healthy turnaround in student achievement based on the Texas tests. No one noticed that on nationally standardized tests little had changed. On the most widely used nationally normed test of high school seniors, Texas in 2002 still trails even Arkansas.

Arkansas will be able to claim total victory, too, very soon, maybe even before Nov. 5, the general election.

Weary of turning up near the cellar of every national test ranking, Arkansas decided five years ago to write its own benchmark tests to measure how much kids know of what we think an Arkansas youngster ought to know at certain grade levels. Devising your own tests carries the bonus of not permitting invidious comparisons with other states. Even then, Arkansas students generally were scoring in lousy ranges. But the state Education Department eventually worked out the kinks and students began to score really well.

This week, the Education Department proudly announced the results of the spring 2002 exams. Fourth and sixth graders blew the top out of the tests, relatively speaking. Particularly astounding were a 22-percent gain from last year in the number of fourth graders achieving at grade level in reading and a 19-percent gain in sixth graders who were satisfactory in math. Eighth graders unaccountably were stuck close to last year's scores.
States that devise their own benchmark tests usually can show sizable leaps, but increases of 19 and 22 percent in a single year are extraordinary. Even when great things are happening in the classroom, improvement in knowledge testing is normally glacial.

Legislators from both parties, who had scheduled a hearing on the tests for the next day, were cynical. The company that the Huckabee administration is paying $3.2 million to administer and score the tests missed the deadline for returning the results by three months. The state let the deadline pass without assessing penalties because preliminary results showed students performing much worse, not better, than last year and the administration
said it wanted to get the right figures. Between the preliminary and final calculations, the Democrat-Gazette reported, the number of fourth graders judged to be proficient climbed from 4 to 65 percent. Education Department director Ray Simon said they checked out the better results and concluded that everything was done right and that the state could go to the bank with the new scores.

No one mentioned that the timing couldn't be better. Huckabee re-election commercials boasting of soaring test scores on his watch were already on the air. We can make too much of the governor's bragging. Every governor seizes upon random statistics to show that he or she has turned the schools around. Bill Clinton was able to find national statistics in every election that showed his school programs were having results. (Huckabee, in fact, is taking credit for Clinton's most spectacular achievement. He has been pointing to a national study that shows Arkansas in the vanguard in the percentage increase in teacher salaries during the decade of the '90s. The big increase was in 1991-92, following the tax increase that Clinton pushed through the legislature in 1991, not in the raises of Huckabee's four years in that decade.)

We also may be too skeptical of the spectacular gains of fourth and sixth graders on the Arkansas benchmark exams. It may not be good education, but high-stakes testing is supposed to produce those kinds of results.

Under the 2001 federal education law, schools that do not show good progress on grade exams are supposed to be punished, including alerting each parent in the lagging schools this year of options for transferring their child to another, higher-performing school if there is one. Arkansas school administrators hoped to escape that remedy with high scores on the Arkansas exam but they learned last month that federal law required them this year to
still use results of the national Stanford Achievement Test, on which most
Arkansas schools tested very badly.

With finances and embarrassment riding on the tests, anecdotal evidence suggests, administrators have had teachers drilling students on the state tests with single-minded passion. The idea is not to leave a problem that will be tested until all the kids are proficient in it. In the old days, teaching to the test was thought to be unprofessional.

The results are good for educationists and politicians who need robust figures, but it is doubtful whether constricting knowledge to the parameters of a few tests raises a wiser, healthier citizenry. 

Posted on this website by permission of  the author, Ernest Dumas


Educational Issues









Hit Counter