The need to raise entry standards into the teaching profession is generally met with nods of approval--that is, until forced to confront the ugly, sharp distinctions between and Black and white test performance and the challenge reconciling what often appear to be two competing needs: high standards and a diverse teaching force.

A recent paper published in the AERA Journal by economists Dan Goldhaber and Michael Hansen (a finalist in NCTQ's 2009 TR3 research competition) complicates matters even further. Three main take aways:

Looking at the test performance of 4,000 teachers in Grades 4 through 6 on North Carolina's two required Praxis II exams (both tests are mostly focused on pedagogy, with only a smattering of content), they find that licensure tests do a pretty poor job of predicting which teachers, regardless of their race, will produce gains in students' reading performance. Conversely, they do a good job of predicting which teachers will produce gains in students' math performance.

Second, the two licensing tests used in North Carolina--and presumably all 122 (!) licensing tests offered by the Educational Testing Service and other companies-- differ in their ability to predict teacher effectiveness. In this case, white teachers did better on the test that was in a multiple choice format whereas Black teachers did better on the test comprising essay questions.

Third, and this is where their findings offer critically important implications for policy: Black teachers, even though their scores on the licensing tests were lower, turned out to be just as good, if not better, than a lot of white teachers with higher scores. This finding is particularly pronounced when Black teachers are assigned to teach Black children.

Goldhaber and Hansen hypothesize that student performance would be better in a majority Black classroom taught by a Black female teacher who has failed the licensing tests than by a white female teacher who (barely) passed the tests. It turns out that the Black teacher would get the greatest student gains, and significantly so. The relative effectiveness of the Black teacher would decrease, but only marginally, if the number of white children in the hypothetical classroom increased. The Black teacher's effectiveness would not be equaled in the classroom by substitution with a white teacher until the white replacement was one who had scored well on the licensing tests.

Pedagogy tests predict teacher performance better on math than reading...white teachers do better on such tests in multiple choice format tests, while Black teachers do better in essay format...Black teachers under-perform on such tests relative to their actual effectiveness...Black children perform better for Black teachers. Without a doubt, the way to craft policies designed to increase the potential for student achievement in view of these findings is a riddle wrapped in a mystery inside an enigma.