To the Editor:
I agree with National Center for Education Statistics Commissioner Mark S. Schneider about the high quality of two recent “hierarchical linear modeling,” or HLM, studies of charter and public vs. private school achievement on the National Assessment of Educational Progress (“NCES Calls for Sticking to the Stats,” Aug. 30, 2006).
I also agree that it is inappropriate for a statistical agency, as opposed to a research agency, to initiate studies—no other federal statistical agency I’m aware of does so—though often there is overlap in the kinds of judgments that go into collecting and reporting raw data and in constructing analyses.
By that same logic, however, shouldn’t the NCES cease and desist from reporting NAEP results by achievement levels? Achievement levels, by their very nature, are subjective, far more so than the decisions that go into doing HLM analyses. Moreover, NAEP’s achievement levels have been strongly criticized as misleading by the Government Accountability Office, the National Academy of Education, and the National Research Council. Does the NCES think its integrity is covered by the teeny-tiny note it has been printing for years about how the achievement levels are “provisional”?
Finally, now that it’s clear for the umpteenth time that there are significant background differences between public and private school students that strongly affect school-sector averages, I would hope that the NCES could go beyond its simple reporting of public-private averages on NAEP. That would be every bit as objective, and enlightening to the public, as the breakdowns of test-score averages by race, ethnicity, and the like that the NCES has traditionally provided.
Bella Rosenberg
Washington, D.C.
