To the Editor:
In response to Randi Weingarten’s Commentary “Accountability That Works,” (May 14, 2008):
I was impressed with the general tone and content of Ms. Weingarten’s proposition for a four-pillared school evaluation system. Her thoughts are penetrating and appropriate, particularly with respect to administrative accountability.
I take issue, however, with a couple of her details. To begin with, she writes about “absolute” academic achievement. Perhaps she has forgotten: The bell curve extends to infinity. We have no way of knowing where the average score resides. When scoring based on the total number of correct responses, we compare the results for known “marker items.” This information tells us what is, not what could be. The numbers are based on the test items’ histories and are in no way absolute.
Second, Ms. Weingarten says that “much of the variation in test scores over a single year is random.” My organization’s research shows that this commonly held notion is false. Students give the best answers they can, based on what they know and how they think. This information comes from the answers they give or choose. It is lost when we score items “right” or “wrong.”
The “right” answers are subject to an additional problem, that they can be supplied from memorization in the absence of understanding, or from procedures less mature than those the markers had expected.
Finally, when we add up the number of correctly answered items, we no longer know which questions were missed. In short, almost all the helpful information for teaching is gone before we start interpreting test results.
Although Ms. Weingarten’s accountability model is excellent as a general approach, until the fundamental flaw in our test-scoring procedures has been corrected, her admirable efforts and those of hundreds of thousands of other dedicated educators will be to no avail.
Jay C. Powell
Better Schooling Systems
A version of this article appeared in the June 11, 2008 edition of Education Week as The ‘Fundamental Flaw’ In Test-Scoring Systems