In this video of 23 minutes, you will get a synopsis of why value-added modeling doesn’t work.
The video is a preview of a collection of research papers that will be available online in a few months, and published in 2014 by Teachers College Press in 2014.
The more we learn about the real consequences of VAM, the more we understand that it has perverse consequences.
We know already that it puts too much emphasis on test scores, and we saw what that produced in Atlanta, DC, El Paso, and other districts.
We know it narrows the curriculum, as only the tested subjects count.
We know it encourages teaching to the test.
We know that it is unstable and unreliable.
We know that good teachers may get low ratings because they teach kids who have high needs.
We know that the composition of the class has a greater effect on the teacher’s rating than the teacher’s “quality.”
We know that the test scores are affected by many factors, not just what the teacher does.
When the papers are available, I will post a link.
Diane Ravitch's blog 
The NYC Data Skeptics Meetup
We seem to be going through a regressive period in intellectual and scientific history. All the dullest and most pin-headed bean-counters have taken control of all the biggest machines, electronic or political, it’s all the same. All the rules I learned in school about making data meaningful and keeping scientific inference honest have been tossed out the window by the new breed of agenda-driven double-think tanks.
I remember the 90s and all the dreams we had for this century.
All dead now …
Teaching kindergarten in FL, my VAM was based on a beginning of the year pre-test and end of the year post-test in reading called the FAIR test. (Yes, just reading. No math or any other subject.) I administered the test with no supervision. Compare that to the teachers whose VAM is based on the highly-secure and supervised FCAT. Oh, and the FAIR test had not changed at all in the 3 years I gave it: same vocabulary words, same sight words, same stories, same comprehension questions. Compare that to the FCAT: which is changed regularly and the teachers can’t even legally read it. Also, the 1st-12th grade teachers don’t get the benefit of a pre-test FCAT or a pre-test FAIR, their VAM is based on last year’s scores compared to this year’s scores. Can you see any problems here? Nah, this makes perfect sense.
Our high school is using the national standardized tests called STAR tests for reading and math. However, science and social studies teachers are to write their own tests. VAM seems fair to me.
Too many variables to control for to effectively implement a VAM.