Depending upon which new story you read, either economist Dan Goldhaber has just come out with a study that shows teacher licensing tests don't matter a twit or that they do in fact predict who will be an effective teacher. In a recent paper presented at the 2006 AERA conference in which he looks at North Carolina's teachers, Goldhaber examines the usual teacher characteristics, but with access to how well teachers did on their licensing tests, enabling comparison of those scores with their student achievement gains over time. Here's our takeaway:

*Similar to recent findings by Eric Hanushek, the great majority of variation in teacher test scores is due to variation within schools and not between schools. This is an emerging consensus finding from research and one which should be considered in the context of paying great teachers more to work in high poverty schools. They may already be there.

*Teachers with the highest test scores are more likely to be teaching middle class white children than poor or minorities.

*Some licensing tests--though all were Praxis or NTE tests--correlate with teacher effectiveness; others do not.

*Licensing tests clearly correlate with teacher effectiveness; it's a question of degree, though. The higher a teacher scores, the better the teachers' students perform. However, Goldhaber tries to control for the fact that high scoring teachers tend to teach in high performing schools, and the strong correlation that he had observed becomes considerably weaker. In the final analysis Goldhaber terms teacher licensure a "weak signal of teacher quality."

Other findings:

Like many others before him, Goldhaber finds no correlation between having a master's degree and student achievement. He found some evidence that having a PhD characterizes a less effective teacher than one who just has a bachelor's.

National Board teachers are more effective teachers.

There's no evidence that experience matters after five years in the classroom. Like a lot of other research, he finds that teachers are pretty effective after only one or two years in the classroom.

Consistent with Thomas Dee's research, matching student and teacher race for black students might help, though the numbers aren't especially telling in that direction. High-performing black teachers, however, did produce better gains in black students than similarly performing white teachers with black students.