Our wonderful reader Laura Chapman reports here on the origins of the laws that purport to measure teacher quality by the test scores of their students. The founding father of this methodology was the late William Sanders, an agricultural statistician who believed that the same productivity used to measure cows could be used to measure teachers. His ideas were adopted and promoted by Arne Duncan’s Race to the Top, which required states to adopt “value-added methodology” if they wanted to compete for a share of billions of federal dollars. The Gates Foundation also embraced test-based accountability. These methods proved to be ineffective at measuring teacher quality; they are inaccurate and demoralizing.


According to a 2019 report coauthored by Audrey Amrein-Beardsley, 15 states are still inflicting teacher evaluations by VAM (value added measures) and 28 are using the equally invalid process of writing up Student Learning Objectives (SLOs). SLOs require you to predict the end-of-year (or end of unit) achievements of students, among other ridicule-worthy feats. https://kappanonline.org/mapping-teacher-evaluation-plans-essa-close-amrein-beardsley-collins/

Vamboozled, the website of Audrey Amrein-Beardsley, is a great resource for anyone still being a victim of this method of estimating the “value you have added” to the test scores of your students.

But there is also a deeper and little known origin story for VAMs. That story was exposed to view in April, 2020, by Gene V. Glass, a Senior Researcher at the National Education Policy Center and a Regents’ Professor Emeritus from Arizona State University. Glass released a treasure trove of correspondence about VAM (value added measures), first used in education by the late William Sanders, an agricultural statistician. http://ed2worlds.blogspot.com/2020/04/an-archaeological-dig-for-vam.html

In his blog post “Archeological Dig for VAM,” Glass reveals how William Sanders borrowed statistical methods for calculating VAM, then began using those calculations to judge teacher productivity/quality, based on the test scores of their students, specifically in the Tennessee Value-Added Assessment System (TVAAS).

It turns out that Sanders’ TVAAS process (VAM) was “built on the formulation of the late C.R. Henderson, a Cornell statistician, a fellow in the American Statistical Association, known for his pioneering work in breeding animals, specifically herds of dairy cows. Henderson’s statistical methods of producing a “genetic evaluation of livestock have been widely accepted, utilized, and enhanced by animal breeders and statisticians.”

Until Henderson’s 1953 publication of “Estimation of variance and covariance components” in Biometrics,” no one had tackled the difficult problem of “estimating variance components from unbalanced data of cross-classified models, e.g., of milk production records of daughters of A.I. (Artificially Inseminated) sires in different herds – where sires are crossed with herds, and, for a large group of herds, each sire has daughters in many herds and each herd has daughters of many sires.” https://ecommons.cornell.edu/bitstream/handle/1813/31657/BU-1085-MA.pdf;sequence=1

If you have a background in statistics (mine is minimal and vintage), you may enjoy reading the extended “defense” of VAM/TVAAS by the late William Sanders who cites his debt to Henderson’s work. Sanders’ defense of using VAM with teachers and the test scores of their students is revealed in his answers to numerous questions from William Robert Saffold, Vanderbilt Institute for Public Policy Studies, who is well-informed about the results in TVASS in Tennessee and wanted more information to interpret the results of TVAAS for educators. The extended discussion reveals the many unwarranted assumptions Sanders made in constructing TVAAS.

I think the hoopla over the specifics of VAM (and SLO’s) is too often disconnected from the fact-based origin story on “how to cull herds of dairy cows to maximize their productivity.” VAMs and SLOs are designed to cull teachers based on their productivity in raising the test scores of their students.

Almost all of the accountability structures in education based on standardized test scores are designed to cull–select and discard–teachers who are not producing gains in test scores. In VAMs, test scores of students are not much different from measures of milk production, whether of individual teachers or the whole herd (school).

Some supporters of VAM’s are acting as if education geneticists. They seem to think that some teachers are destined to be more productive than others. They insist, for example, that Teach For America graduates with high GPA’s from selective colleges are good breeders of above average test scores in their students. Moreover, these potentially good breeders only need is a brief course in summer before they are ready to produce students who are high scorers on tests. That brief summer dose of instruction is analogous to providing artificial insemination in breeding females… or for males, perhaps a dose of Viagra.

VAMs and SLOs are flawed measured pushed by the Obama/Duncan administration’s Race to the Top. These measures are still present in many state ESSA plans. That may explain why Race to the Top testing resources are still available, even if developed under contract for Race to The Top by members of a “Reform Support Network.”

The Reform Support Network was nothing more than a huge marketing campaign for these flawed measures. Here, for example, is how they marketed SLOs as a substitute for subjects and grade levels for which there were no statewide standardized test scores for calculating VAM. One is the infamous collective measure where, as Diane notes, teachers “are assigned ratings for students they never taught in subjects they never taught.” https://www.engageny.org/sites/default/files/resource/attachments/rsn-slo-toolkit.pdf