How exactly do you accurately assess teaching? Test scores? I've worked at very tough schools. High unemployment, high crime area. Good teaching in this context meant the teacher showed up, made it through the year without having a mental breakdown and student attendance was > 50%.
On the flipside I've worked at top tier private schools where success was gauged by how many of our students were in the top n% for the state.
Teachers fear outside assessment from bureaucrats. Here in Australia we even have league tables for schools, and there has been talk of tying teacher pay to student performance. Obviously a worry if you are a classroom teacher who understands the human component of teaching (ie it's more than test results in a spreadsheet).
In my opinion to truly assess a teacher you'd need to observe their teaching for weeks to see how they deal with the myriad of situations that arise in the classroom. What we end up with though is standardised testing which ends up telling us what we already know.
Basically I think all this "it can't be measured" stuff is a bit dishonest. My wife is a high school teacher, and amongst her department everyone knows who is good and who isn't, same as any job. The only difference is that that knowledge can't possible have any effect on your career.
In any case, the point you are making about standardized testing is completely beside the point. No one who advocates using testing advocates using only testing for teacher evaluation. And no one advocates using the raw scores. Of course teachers with poor students in bad schools will have worse raw scores then teachers with rich students in good schools. The proposed measurement is virtually always some kind of "value added" score that attempts to control for these variables and measure the teachers impact.
I never said that it cant be measured. It's just really hard to do it right.
The solution is not to simply use private sector type performance reviews since correcting for differences within the school is difficult let alone correcting for differences between schools.
Measuring teacher performance is a hot area of debate around the world, and a consensus for how to "do it right" hasn't yet emerged.
Lots of questions need to be fairly addressed:
-- Will testing be able to differentiate between school-based and non-school-based influences on
-- Will testing be able to differentiate between the influences of different teachers on individual students or groups of students? What might be the ultimate extent of increased standardised testing?
-- What criteria are intended to form the basis of a performance-based pay scheme?
-- If the predominant criterion used is to be student progress as measured by standardised testing, what measures will the Government take to ensure the implementation of valid and appropriate testing regimes and instruments?
These questions come from an issues paper published by Australian Primary Principals Association.
Also from the paper:
In jobs where pay is linked to performance:
The criteria for determining the payment of additional rewards are to be objectively determined: whether in volume of product or sales, increase in profits, or additional hours worked. More accurately put, the context of the industries in which systems of this kind work well are those
where outputs and outcomes are easily, and objectively, quantifiable. This quantification can usually (although not always) be reduced to monetary terms.
So how exactly to we quantify and reward the teacher who helps her students develop skills that 'are not quantifiable'?
PS: In Australia primary and secondary teachers can't get tenure, however given a powerful union it is difficult to sack teachers, and this admittedly is a big problem.
Second, judging a programmer solely (in an atomic fashion) by their code is the exact same thing as judging a teacher solely by test scores. How do you know the code was not written at 2am the night of a release to stop vital information from being deleted every time a user submits a form.
On the flipside how do you know that the elegant code wasn't the combination of effort and discussion of several coders all considered to be in the top n% of some mailing list.
I think my point is really that we all tend to rush to judgement of professions outside of our own, when in reality we share the same human problems that are incredibly hard to formalize and hence provide a systematic solution for.