OK, if you want a metric that might work ... testing a year after the teacher is gone.
If you test while the teacher is in charge, they will encourage the student to game the test. This has short-term gain and long-term pain. If you reward teachers for how well the student retains their knowledge (long term), it should work better.
That won't stop teachers pressuring their co-workers to teach to the test, or principals from telling teachers to teach the test, but it helps reduce it. You could rate schools on how well students do 1 year after leaving, to keep the principals in line. I'd guess teachers will just fob off their co-workers.
You can't argue that metrics should be the be-all and end-all of everything, and a teachers' intuition should count for nothing, when there are all these problems with the metrics we have. Here's a good story: http://www.washingtonpost.com/local/education/creative--moti...
This teacher was by all accounts a phenomenal teacher, but she was fired because her batch of kids came from a school that had suspiciously high scores the year before, and is currently under investigation for cheating. She didn't maintain that high level of scores so she was fired.
Our police system is dominated by metrics and we got The Wire as a result. In education, you hear about cheating scandals in DC, good teachers being fired due to unfortunate statistical conflagrances.. if you're gonna advocate turning your brain off and delegating all decision-making to your metrics, they have to be good.
...good teachers being fired due to unfortunate statistical conflagrances..
How do you know they were good? Because they said so, and one of their cronies agreed?
This link is relevant: https://en.wikipedia.org/wiki/Principal-agent_problem
I'll note that you want to scrap objective metrics because they have some problems, but you don't seem to realize "teachers' intuition" also has problems. One of the biggest is comparability - how I can compare the intuition of teacher A with the intuition of teacher B?
With standardized tests, the answer is simple arithmetic.
Metrics are fine, they just shouldn't be fetishized, and we shouldn't allow them to make us fire good teachers. They also shouldn't dominate what's taught to the extent they do. Nation of multiple choice test-takers? That's what we want out of our school system?
If the students didn't perform, but her superiors liked her, why are you assuming the superiors know what they are talking about?
Again, how can I compare the opinions of heterogeneous superiors, and why do you believe the opinions are correct and objective metrics are wrong?
Opinions fail on many counts - comparability and objectivity being the most obvious. Why do you fetishize opinion, and assert it as being more accurate than objective metrics?
They also shouldn't dominate what's taught to the extent they do.
All standardized tests do is measure whether a school is meeting their goals. If standardized tests are preventing schools from ignoring their goals and teaching whatever they feel like (or perhaps nothing at all), why is that a bad thing?
Further, standardized tests don't really measure much, except for how well student can take tests and regurgitate information.
Client - parents Contractor - admin Sub - teacher
Now imagine that both the client and the sub contractor are influential in the final product (student ability). How do you isolate an incident like divorce from the overall performance measurement of the teacher?
In the UK, governments have set up a system whereby schools can choose which exam board they want to have for each subject. Each exam board makes and marks its own exams.
Exam boards are motivated to be cheap, generous with the marks, and give workshops you have to pay to attend where they give you information about what will be on this year's exams.
The exam boards are not motivated to make decent exams. They are riddled with mistakes, and do not adequately test students' understanding of the subjects.
Look at some of the GCSE IS/IT/ICT exams; you can find them online. You'll see vaguely-phrased mini-essay questions, where students get marks based on how many of the required points they make, where the points are listed in the examiner's marking guide. If you make a point that wasn't anticipated by the people who set the exams, you're not getting any points.
Problem is, the questions are so open and vague that students will be fumbling around in the dark if they don't know specifically what examiners are looking for. It really is just a matter of teaching to the test. You need to have learned specifically what the examiners are looking for. It doesn't matter whether you know much about ICT or not.
So what's the point? It's supposed to be an ICT exam, not a jumping-through-examiners'-hoops exam.
Something interesting that I've observed is that most of the people complaining about metrics are really complaining about the goals of the system. I.e., many people want the school to increase mean(student scores), and then complain when the school attempts to increase count(student score > CUTOFF). If you want the school to have different goals, say so. Don't blame the goals of the system on the existence of metrics.