In psychology and the social sciences, there is a strong tradition of qualitative, as well as quantitative, investigation. More emphasis on qualitative data would undoubtedly have resulted in better treatment of the teacher in question and of the students who will be deprived of her talents in the District.
Tom Holman, Montgomery Village
●
Those familiar with the methods of science generally appreciate the perils of measurement error, particularly as it affects the evaluation of complex phenomena. For example, when data from the CERN particle collider in Switzerland and Italy produced data that stood Albert Einstein and 100 years of quantum physics on its head, investigators wisely invited an open and careful review of their findings, methods and interpreted results. Months later, it was determined that their initially astounding findings were the product of measurement error.
However, D.C. school administrators apparently take a far less discerning view of their “value added” statistical measure of teacher performance, even when that measure, much like the CERN equipment, produces highly contradictory and dubious data.
But one hopes there is some useful “value added” learning here. If nothing else, the school district has produced interesting pilot data for a new measure of school administrator performance. Perhaps we could call it the “value subtracted” statistic.
David Fago, College Park
●
The No Child Left Behind approach isn’t working. If teachers’ jobs and paychecks depend on students’ test scores, then — lo and behold! — the test scores will be astronomical. Perhaps teacher Sarah Wysocki should have done exactly what it is suggested that educators at Barnard Elementary School may have done: Take out her trusty eraser and correct the students’ tests. That way the District would still have an excellent teacher, and the Education Department would be smiling because No Child Left Behind was working so well.
I’m grateful that I retired from public education — I taught in Prince George’s County for 25 years — before this insanity began. I may buy stock in an eraser factory with my retirement money.
Deni Foster, Berwyn Heights
●
Thank you for printing the article about the teacher fired by D.C. public schools because of her students’ tests scores. I am assigning it in my graduate-level statistics class at DePaul University as an example of how not to use statistics in management.
Independent of whether the students’ earlier scores were inflated, no social scientist would draw strong conclusions from only 25 students over just two waves of data, particularly using the methods described in the article. Random variation in student performance on each test day would be so large that there would be no way to evaluate how much of the difference between the two sets of scores could be attributed to the teacher.
If the District’s public schools really want to use testing to improve performance, I have a suggestion: Test administrators on their understanding of applied statistical methods and use the scores to determine whether they should remain in their jobs.
Christopher J. Einolf, Chicago