Teacher Credentials Don't Matter for Student Achievement

From NBER Digest;

In How and Why Do Teacher Credentials Matter for Student Achievement? (NBER Working Paper No. 12828), co-authors Charles Clotfelter, Helen Ladd, and Jacob Vigdor study the effect of teacher credentialing on student achievement using data on 75 percent of all children in North Carolina in grades 3, 4, and 5 from 1994 to 2003. Their results show that having a graduate degree has little effect on student achievement. Teachers who entered teaching with a master's degree, or who earned it within five years of beginning to teach, were as effective as teachers without a master's degree. Teachers who earned a master's degree more than five years after they started teaching were less effective than those without master's degrees.

As in previous studies, the authors find here that teachers with more experience are better teachers. This is the case even after accounting for the fact that the teachers who remain teachers may, on average, be less effective than those who leave. The benefit of experience peaks at 21-27 years of teaching and adds 0.092 to 0.119 standard deviations to student achievement scores. More than half of that gain occurs during the first years of teaching. Teachers who come from competitive undergraduate institutions are somewhat more effective than those who come from uncompetitive colleges or universities, the researchers find.

By comparison, increasing class size by five students reduces math achievement by 0.015 to 0.025 standard deviations, and reading achievement by 0.010 to 0.020 standard deviations. Having a parent with only a high school degree decreases math scores by about 0.11 standard deviations relative to having a parent who has a college degree. Having a parent who is a high school dropout reduces achievement by another 0.11 standard deviations. The effects for reading are slightly larger than for math.

Overall, the authors find that having a math teacher with low scores on the licensing exam, little experience, an undergraduate degree from a noncompetitive college, and an emergency license, is roughly equivalent to having poorly educated parents. This suggests that schools that put teachers with weak credentials into classrooms with educationally disadvantaged children tend to widen the already large achievement gaps associated with various socioeconomic differences.

While the effect of teacher credentials on mathematics achievement for third, fourth, and fifth graders is quite large compared to class size or parental education, the effects on reading achievement are noticeably smaller. Even highly credentialed teachers will likely not offset the effects of educationally impoverished family backgrounds on reading. In view of this, the authors conclude that a "real challenge for policymakers is to find ways to direct the teachers with strong credentials to the students who most need them."

The data used in this study are administrative records; they allow the researchers to link specific teachers with the reading and math performance of specific children, provide substantial information on each child's socioeconomic status, and contain each child's standardized test scores. The records also identify each teacher's license type, licensing exam score, years of experience, undergraduate college or university, and advanced degrees or National Board certification.

Previous work suggests that teachers with stronger credentials tend to end up teaching students who perform better academically. This rich dataset allows the authors to correct for the possibility that, while teachers with better credentials can command better students and thus are associated with high student achievement, the better results may not stem from better teaching.

The authors focus on credentials that can be affected by policy. These include the number of years a teacher has taught, whether a teacher has a regular or an emergency license, whether the teacher has an advanced degree or National Board Certification, his score on the state's licensing exam, and the competitiveness of his undergraduate institution. A student's standardized end-of-grade test scores in reading and math are assumed to be dependent on achievement in the previous year and on student, teacher credentials, and classroom characteristics in the current year.