Students whose teachers score high on state licensing exams learn more mathematics over the course of a school year than peers taught by teachers with low scores, according to a new study that draws on 10 years of test-score data on North Carolina schoolchildren.
Coming at a time when experts and policymakers are divided over what makes a “highly qualified” teacher and how to produce one, the findings provide important evidence that teacher-licensing exams may be meeting their purpose in screening out ineffective teachers. But Dan D. Goldhaber, the University of Washington researcher who did the analysis, said his study also found a downside to some such tests: They keep some good teachers out of the classroom, too.
“There are clear trade-offs states are making in using these tests,” said Mr. Goldhaber, a research associate professor at the university, located in Seattle. “So it really becomes a value judgment.”
Mr. Goldhaber presented his findings here April 7 at the annual meeting of the American Educational Research Association, a Washington-based group that represents 25,000 scholars from around the world.
The report, “Everyone’s Doing It, But What Does Teacher Testing Tell Us About Teacher Effectiveness?” April 2006, is posted by the Center on Reinventing Public Education.
All but two states require teachers to pass some sort of licensing exam before entering the profession. Few studies, though, have explored whether the exams work: Do teachers who pass them produce bigger gains in student learning than those who fail to make the cutoff? Mr. Goldhaber looked to North Carolina to test that idea because it is one of a handful of states with a data system that can link student test scores to specific classrooms and teachers. He analyzed testing data collected from 1994 to 2004 on 701,000 students in grades 4-6 across the state and gathered licensing records for 24,000 teachers.
Differences in Math
Since the 1960s, North Carolina teachers have been required to pass a variety of tests before setting foot in the classroom. The state began with the National Teachers Exam and then, in 1996, replaced it with the Praxis II, a set of subject-focused tests developed by the Educational Testing Service of Princeton, N.J.
In 2000, North Carolina lawmakers made changes in the combination of cutoff scores required to pass the tests. As a result of the changes, Mr. Goldhaber was able to gather data for classes taught by teachers who would have failed the exams had they taken them a few years later or earlier as well as for those taught by teachers with passing scores.
What the researchers found was that, on average, students whose teachers earned passing scores on licensure exams made slightly larger gains in reading and mathematics than did the students of teachers who would have failed the tests under different conditions. But the improvements were significant in math only.
In that subject, the learning boost that students got translated roughly to an additional 3-point gain on state tests, according to Mr. Goldhaber. (That’s on a test on which all students score an average of 19 points more from year to year.)
Because high-scoring teachers tend to end up in schools and classrooms with higher-achieving students, Mr. Goldhaber cut the data a different way to take into account previous achievement differences among students and other factors. Analyzed that way, the gap between students of high- and low-scoring teachers shrunk slightly, but continued to point in the same direction: Higher-scoring teachers had students who scored higher from year to year on state tests.
However, the changes that North Carolina had made in its minimum-score requirements were slight. What would have happened, Mr. Goldhaber wondered, if the state had followed the lead of some other states and raised the bar higher—making it much harder to pass the licensing tests—in an effort to improve teacher quality?
To find out, he looked to Connecticut, which uses the same gateway tests but sets a higher cutoff score. When Mr. Goldhaber applied Connecticut’s standards to the North Carolina data, he found that the tests were less useful for sifting out productive teachers.
That’s because the new estimation yielded significant numbers of “false positives” and “false negatives.” In other words, both cutoffs would let in ineffective teachers, and the higher bar would keep out some whose students had test-score gains that were higher than the median for the state despite the fact that their teachers’ certification scores fell below the cutoffs.
Mr. Goldhaber was cautious about saying how his results would figure in current debates over how to improve teacher quality. Some experts argue for raising the bars that would-be teachers have to cross to enter the field, while others argue for creating alternative routes into the profession to make it easier for talented professionals from other fields to enter the classsroom.
Linda Darling-Hammond, a Stanford University education professor and expert on teaching who reviewed the study, said the findings highlight the need to develop better assessments of teachers’ ability to teach. “If we had better assessments, we could both certify teachers more effectively and also drive reforms in the field,” she said. “And maybe we could finally put the alternative-certification-versus-traditional-certification debate to rest.”