Educators often rely on standardized tests to make any number of important decisions about their students, including whether to let students graduate, enter a gifted- or special-education program or move to the next grade. However, the playing field for such high-stakes tests may not be level for all students, said psychologists at APA's 2001 Annual Convention.
No matter how well a test is constructed, there is concern that many schools use such test results in inappropriate ways and that the tests themselves fail to adjust for America's increasingly diverse student population, especially minority students, those from low-income families and those who do not speak English as their first language, the speakers explained.
Forty-seven states have established standards for what students should know and be able to do, said educational researcher Linda Darling Hammond, PhD, and about half use tests to make promotion, graduation and other crucial decisions. While psychologists and educators have developed a document--the "Standards for Educational and Psychological Testing"--stating that schools shouldn't rely on only one test to make such decisions, "we have almost half of the states in this country violating the standards the profession has set for the uses of tests," said Hammond.
"These [policy-based] tests have substantial unintended consequences," explained Hammond. For example, even though the percentage of students passing high-stakes achievement tests has increased, Hammond warned psychologists such numbers may be misleading: In many schools, the percentage of students passing tests has increased only because poorly performing students are not included in the pool, she said. To make test averages look better, some schools might, for instance, hold students back or exclude or separately report the scores of special education students.
Studies have shown, said Hammond, that such methods leave poorly performing students even worse off. For example, researchers at the Consortium on Chicago School Research have found that students who were held back had significantly higher dropout rates and lower achievement than those of similar ability who were promoted.
"Although the reforms were popular, the policy-makers and educators simply ignored a large body of research showing they would not produce academic gains and would increase dropout rates," said Hammond. "This was a policy with no probable educational benefits and large costs. The benefits were political and the costs were borne by at-risk students."
Inappropriate use of test results isn't the only problem. The tests themselves often fail to take into account the increasingly diverse students who take them, said the presenters. Just one example, given by George Mason University psychologist Jack A. Naglieri, PhD, are tests that use English vocabulary and cultural-specific knowledge to measure intelligence. Such tests measure achievement and learning, Naglieri argued, not intelligence. For instance, a talented student who speaks Spanish as her first language may do poorly on vocabulary portions of an intelligence test, potentially compromising her gifted-education placement.
"It's really inappropriate to have content that puts certain groups at disadvantages," Naglieri said. His research has shown that when children's intelligence is measured without achievement-based items, minority students score significantly higher--"which has very important implications for dealing with the problem, for example, of over-representation of African-American children in [special education] classes."
Even though standardized testing isn't perfect, tests should still be used, said some speakers, since properly developed and used tests provide critical measures of students' progress. In fact, the testing standards that APA helped develop (available at APA Testing Standards) explain that appropriate testing can be key in identifying lower-performing students and schools so that they can get the extra resources they need.
In addition, there's much that educators and psychologists can do now to level the playing field. Instead of reconstructing tests, educators and psychologists should focus on learning how to be culturally sensitive when selecting and administering tests, adapt tests to students' needs and assess their own biases when interpreting the results, said University of Texas at Galveston psychologist Freddy A. Paniagua, PhD.
To test general knowledge, for example, testers could ask students to tell them the colors of the flag in their country, instead of just asking about the American flag. And to prevent what Paniagua calls student "thought blocking"--difficulty translating what they're thinking in their native language into spoken English--the instructor should use the language of the student.
Making adaptations such as these is becoming increasingly important, explained University of St. Thomas, Houston, psychologist Kurt F. Geisinger, PhD, because of the nation's changing demographics. From 1999 to 1995, 43 percent of the 2.8 million immigrants to the United States were Hispanic, 25 percent white (mostly Eastern Europeans), 24.5 percent Asian and 7 percent black--and about 90 percent of these individuals don't speak English as their first language, if at all, said Geisinger.
Assessing those students with tests that do not take into account their acculturation puts them at a serious disadvantage, the psychologists said.
But in Hammond's eye, the best solution to biased testing would be to follow the model of states like Nebraska, Minnesota and Connecticut--three of the highest achieving states in the country--that conduct performance-based assessments, instead of relying solely on high-stakes tests. In general, these assessments, scored by multiple teachers, look at samples of students' work, take into account behavioral observations and measure whether students have learned the material covered in course syllabi.