Is the LA Times Correct?

At this date, it is well known that the LA Times published a number of articles on the quality of LAUSD teachers as measured by their "value-added" calculations of student tests.  In the aftermath, I am confronted by the apparent reaction to the Times.  It seems that people fall generally into 2 camps:

1. Educators -- Who while not resistant to the concept of accountability ARE concerned by the LA Times approach.
2. The Rest of the World -- Who are predominately of the opinion -- "Well absolutely, is there any question on the matter?"

My opinion is probably someplace in the middle.  Clearly, we are and must move for all of us being more publicly accountable for our work and progress.  The "value-added" approach certainly provides an improvement over the other existing notions of "performance" based on API and AYP.  I detailed some of this in one of my earlier blogs.

My problem with the Times approach is that while the approach may be correct, the information can be very suspect.  Why not use the Sanders {Tennessee} approach that is proven over time / multiple locations and been reviewed a number of times?

My concerns about the Times approach are:

1. I think we have the wrong unit of analysis.  We should from a data / statistical / pragmatic view probably be more focused on the school rather than the teacher -- which is how Sanders approached the matter.  Often teachers have unique student populations (e.g., GATE, English Learners, Special Education, etc.) that can skew the results.  Additionally, teachers would lack incentive to undertake instruction of these classes for obvious reasons.  However, schools are generally large enough to handle the variations mentioned -- AND would address the low-performing educators and low-yield instructional practices that are limiting the value-added performance of the school.  I DO like how the value-added notion (ala Sanders) appropriately accounts and controls for the demographic variable.
2. At a normal class size of 30 (and there are certainly those that either exceed or are vastly less than 30) -- it at strains the normal notions of required sample size to make the conclusions valid.
3. The normal test instrument we use in California (CST -- California Standards Test) and therefore by the LA Times was and is designed to assess the effectiveness of programs NOT students.  Therefore, when it is applied to looking at students as the "outcome" and performance of teachers (or any of the rest of us) -- it leaves a lot to be desired AND ultimately does not stand the test of statistical validity.  This has been a long-standing issue in California.  If we appropriately change the test, it will work.

What is your view on this matter?  Should we be taking about some value-added approach (not necessarily the LA Times / LAUSD approach) to improving our public accountability with an easy-to-understand metric?