Educational Testing, Done Right

A year ago, an article about longitudinal testing of teaching exposed me to some new ideas. Recent web searching has turned up other articles. These generally refute the value of assesment tests like those mandated by No Child Left Behind, and point to a more effective use of testing to improve teaching.

Izard and Jeffrey. Testing for Teaching: A longitudinal formative assessment project. 2003.

Hanushek and Rivkin. How to Improve the Supply of High-Quality Teachers. 2003.

These papers make the case for some unpopular ideas. First, toss out punitive testing of entire schools. These scores seem to be designed to encourage the idea of "school choice" and are probably motivated by the private school subsidy voucher movement.

Implement testing, but test to assess, not to evaluate after the fact. Test to find out how teaching needs to be modified to maximize the educational gains for a particular class.

Start doing longitudinal testing of teacher effectiveness by measuring student achievement over several years, and, at the same time, reduce teacher testing. Then, reward effective teachers. This idea is unpopular with unions. Longitudinal testing would help identify the bad teacher and the good teachers, irrespective of the abilities or preparedness of the students. (A class that starts 3 years below grade level and ends the year 2 years below grade level is not a sign of a bad teacher. If they end the year 1.5 years below grade level, that's a great teacher!)

Expect to pay more for teachers who teach disadvantaged children. Expect this to be the case for eternity. Expect this to effectively improve the education of the working class.