Thursday, January 1, 2015

"Value Added Model" for Teachers is Poor Modeling

Thank you, Anthony Cody, for prompting me to send a message before the January 2 comments deadline to the Department of Education protesting their proposed use of so-called "Value Added Models" to evaluate teacher preparation programs. Here is what I sent, slightly edited to remove details of a specific case (the teacher in question was not me, by the way; I knew her, but not well).

Subject: "value-added model" for teachers is poor modeling

To Whom It May Concern:

I am a public school parent and have been a math and science teacher since 2009, when I attended a teacher education program after a previous career in the scientific software industry. I am writing to register my objection to the idea of using student standardized test scores to evaluate teachers or teacher training programs. Although I would be in favor of putting teachers, schools, or even education programs under closer scrutiny in cases where scores plunge or where consistent losses are documented over multiple years, the test score data on student growth is nowhere near precise enough to use fairly for these evaluations. It would be like basing the Consumer Price Index only on strawberry jam prices. I agree with many of the objections listed in this article:

Data on student performance also inevitably reflects far more than the learning experience the student had in the classroom. For instance, availability of computers at school could have a huge effect. For another example, [horror story redacted in which test scores plunged for some classes one new teacher had because of events the previous year, when she was not even at the school]. Should the teacher or her education program be punished for [redacted]?

Furthermore, the SBAC and PARCC tests are at a "beta" stage. We all know there are going to be problems with the rollout; how could there not be, with such major changes? By putting so much pressure on teachers to "achieve" on inevitably faulty first-round tests, you risk alienating some of your best allies. Many of us want to improve our teaching with Common Core standards, and we want high-quality tests that will give us some information about what our students are learning and what we still need to improve in our instruction. Help give us a way to give feedback on the tests and improve them without the horrendous conflicts of interest you introduce when we and our education programs are punished for poor test questions or design that are likely to disproportionately impact our ELL, SpEd, and non-white-middle-class students.

I do believe that over time, education programs that produce the most valued teachers (and, of course, the most value for their teacher students) will be those that have the best record of placing teachers in schools. I completely approve of efforts to increase reporting requirements for programs of education on teacher placement, and could support programs gaining an advantage if they consistently supply teachers to high-needs schools. 

I hope you will carefully consider all points of view and ensure that any data-driven analyses are actually using high-quality, relevant data.


Julie Wright
Portland, Oregon