Student pool manipulated for tests, report finds

• Article
• Comments ()
• Click-2-Listen
• Videos

By

The high stakes involved in results of standardized accountability tests are prompting some public schools to take steps designed to keep their worst students from taking them, a study charges.

The report by University of Florida economist David Figlio concludes that schools are "disproportionately punishing low-performing students during testing periods" to help ensure that fewer weak students are there to take the test and lower a school's overall score.

Mr. Figlio said research he and other investigators previously conducted found that schools use tactics, such as beefing up nutrition for students at test time, "classifying more marginal students as disabled" and temporarily moving them into special-education classes, where he says tests are less likely to be administered, to keep test scores high.

But John Nori, director of instructional leadership resources for the National Association of Secondary School Principals, voiced surprise as well as skepticism that schools are engaging in such tricks.

"Under the [federal] law, students are tested whether they are in special education or not. And it would not be ethical to suspend anyone [for poor performance on an earlier test]. Ninety-five percent of kids must be tested, and there is no hiding that I know of," he said in an interview.

Mr. Figlio countered that public school administrators have long argued "that they are under a lot of pressure at test time," and that "low achievers tend to misbehave more" at that time.

In his report, published in the May issue of the Journal of Public Economics, Mr. Figlio said accountability tests "play a critical role" in how much federal and state funding a school receives and can determine whether "teachers receive bonuses" and whether "principals get promoted or demoted."

Because of this, he said, schools are likely to focus on actions such as improving test-preparation skills of pupils and tailoring instruction to subjects likely to be included on the examination.

But "the more these scores can be manipulated, the less meaningful these ratings are," Mr. Figlio said.

Testing of students to evaluate a public school has been the biggest educational policy change in the United States in the past few decades. Mr. Figlio said it became popular in some states in the 1990s and became national with the enactment of the federal No Child Left Behind Act of 2001. That legislation requires annual testing of public school students in the third through eighth grades, and it rewards and sanctions schools based on how students perform.

Mr. Figlio concedes he has concern about the language in the No Child Left Behind Act, because it relies on "proficiency standards" and "precludes the use of student [academic] gains to measure a student's learning progress."

"Some states have low proficiency standards. Some have high standards ... but whenever you have a school using proficiency standards" rather than student gains to measure performance, "there is a chance to game the system." He said pressure is mounting for Congress to amend the law to assess learning gains.

For his study, Mr. Figlio analyzed suspension rates of students involved in more than 41,800 incidents at 504 Florida elementary, middle and high schools from 1996 through 2000. The period examined constituted the first four years after the state's introduction of its accountability exam, the Florida Comprehensive Assessment Test.

Mr. Figlio compared disciplinary actions taken against students for the same kinds of incidents to see whether the suspensions were influenced by a student's scores on the FCAT the year before. Students with the lowest math and reading scores on the FCAT the previous year had a higher risk for being suspended longer, including being kept out a week or more, than those who fared well on the test, Mr. Figlio said.

"While schools always tend to assign harsher punishments to low-performing students than to high-performing students throughout the year, this gap grows substantially during the testing window," says the article's abstract.