About this blog Subscribe to this blog

Thompson: First Do No Harm

Cep2 I can’t deny that I enjoying writing a spoof of the Center for American Progress'’ latest paper on Value Added Models, and parts were awfully easy to lampoon. It was a serious paper, however, and balance is necessary. For instance I have long argued that growth models that misfire 15 to 20% of the time may be valid for incentives. But who would invest in a career that has a one in six chance per year of being destroyed by an invalid statistical model? Similarly the CAP wrote "few would dispute, for example, that the decision to terminate a teacher’s employment is a terribly serious one. In contrast, few would maintain that the decision to award a $600 bonus to especially effective teachers is anywhere near as serious."

I am willing to support local efforts to use VAMs to complement or supplement other evaluations, especially when that data is in the hands of a peer review committee, and not just an administrator. And the CAP admits "indeed, some empirical evidence of misattribution seems to militate against using value-added estimates for any purpose..."

"Teachers can be forgiven for worrying about the imposition of crude, unfair policies characterized by the phrase ‘sort and fire,’ given public education’s history of half-baked, fads, and fleeting reforms."

It was nice to read the acknowledgment that "The work of teaching is complex, but test-based accountability has focused attention on one dimension of practice: ... The term value-added belies a sense of bottom-line certainty that is not justified. A particularly troubling one is encapsulated by the finding that value-added estimates are sensitive to the choice of achievement test. ...

Concern that estimates may be biased rears its head because student-teacher matches are not random. ... Whether students or patients live in concentrated poverty, for example, affects the nature of the instructional or medical challenges that they present."

The comparison with medicine was constructive, even if the CAP was much too quick to deemphasize the maxim of "First Do No Harm." Prudence is not negligence, and even if the CAP sees a political reason to move "full speed ahead," it did not give an educational reason for that risky strategy. Improving teacher quality must move ahead on all fronts. Even if we perfected the evaluation and termination process (which in itself would require a revolutionary change in the culture of schools) we will not come close to recruiting and retaining enough new talent until we also transform the learning cultures in poor schools.

The use of Value Added Models for evaluation purposes is comparable to using medicines for purposes for which they have not been approved. Doctors take that sort of risk, but they know to be very careful when doing so.


TrackBack URL for this entry:

Listed below are links to weblogs that reference Thompson: First Do No Harm:


Permalink URL for this entry:


Feed You can follow this conversation by subscribing to the comment feed for this post.

The comments to this entry are closed.

Disclaimer: The opinions expressed in This Week In Education are strictly those of the author and do not reflect the opinions or endorsement of Scholastic, Inc.