Just as a thermometer can't measure blood pressure (even though both assess something about health when taken in context); a standardized test given annually cannot measure teacher performance.
Yes, we can build tests that measure student academic progress on a short term basis. It's way too expensive to build many, many tests for all the situations, students, and curriculums in today's schools. As such, the standardized tests now used are subject to all sorts of confounding that is almost impossible to "subtract out" with value-added (structural equation modeling) formulas. This is the conclusion of the majority of folks with the exception of those selling the snake oil.
Many of us have tried it and studied it, and soooooo....if you want to evaluate teachers your best bet is to use teacher made tests at the school level for pre-post testing multiple times a year combined with observations of teaching by peers and administrators, and qualitative discussions of the context around all the evidence.
Having student work samples is often useful. Long term evidence (like kids who win American Idol and credit their elementary school teacher), and also surveys of parents and kids can be tossed in the mix. Sometimes affective evidence like the kids enthusiasm for a subject is more important than vowel sounds and multiplication tables.
We know what methods work best...and it can't easily be put in a bottle! In fact, the methods change with the cultural and subjects. Different kids and situations need lots of different methods. Veteran teachers are actually (if you watch them) more versatile and have a larger tool kit of methods in most cases (contrary to popular opinion). In 1980 one famous research published articles on 40 years research on "what works" to produce exceptional adults. Yes, some great teachers were identified, but even a great teacher works with some students but is a poor fit for others. It's the same now as 75 years ago...no one number will accurately work.
= new reply since forum marked as read
