Remember, these are percentiles we're talking about here - a normalized measure of a distribution. If a student increases their percentile, then they necessarily bump someone else out of that percentile. It's no surprise to me at all that if there is a classroom that has a huge increase in percentile then there will likely be another classroom that sees a huge decrease in percentile.

The fact that a particular classroom saw the decrease tells us nothing about why. Sure, part of it may be the teacher, but this does not tell you how much of it can be contributed to their value as a teacher. In my experience, kids tend to be placed in classes with other kids that are going in the same direction. So the best students will end up in classes that will improve, and the worst students will end up in classes that decline. The fact that this is surprising to some people, is well, surprising.

Parents are allowed to evaluate the performance of their children by any metric they choose. If a parent wants to congratulate their kid for passing band, but doesn't care if they failed English, then they are free to do so.

What's being proposed is that we use quantifiable measures that have questionable correlation with teacher performance, and that contain biases that are very very difficult to quantify.

People seem to think that teachers unions don't represent the majority of teachers, but this is wrong. The vast majority of teachers support the fight that the unions are waging on their behalf. No teachers want to be judged based on these test scores because they simply do not correlate with their value as a teacher. Of course, value means different things to different people. If you think that the value of a teacher is based purely on how well their students perform on a handful of standardized tests, then you have the right to that belief. This doesn't make it true, of course - any more than my belief that the value of a teacher is based on how much they inspire their students to love learning in general.

The only difference is that you think that your metric actually measures what you think it does, whereas I accept that there is no quantifiable metric for what I believe is important.

Students are not assigned 'practically randomly'. Bad students are placed in classes with other bad students. Two teachers can be teaching the exact same course and consistently have widely differing batches of students.

It wouldn't surprise me in the least if the two classrooms that were being used as comparison were the extreme opposite ends of the distribution of teacher 'ability' according to this metric. Honestly, without looking at this distribution, it's very hard to draw a conclusion that anything needs to change. For all we know, the teacher with the high increase in test scores was an extreme outlier - and while outlying teachers on the good side of the distribution should be praised and perhaps used as a model, this still does not allow us to draw a conclusion that anything needs to change.

My biggest concern is that both the best and worst students in any distribution tend to be where they are because of their own ambitions (or lack thereof). If you toss out the teacher who has to teach the worst students because of failing test scores, and you replace that teacher with the best teacher in the school, then you'll probably end up firing that teacher as well when the test scores don't improve.

We have to focus on the middle 90% of students and teachers, and do what's best overall. The best will be the best regardless of any metrics we choose to measure them by, and the worst will always be bad apples. So why would anyone argue to remove the bad apples from the teacher pool without also arguing to remove the bad apples from the student pool?

This makes me sad...

True, but while cubicles and factory workstations are usually interchangeable, classrooms are often shared, and not suited for the subjects they are being used for. Physics labs being taught in English classrooms? English classes being taught in the auditorium? This happens quite often.

The school only provides a minimum of necessary equipment. Most teachers use a considerable percentage of their salary just buying pens and paper out of their own pockets so their students can do their assignments.

And who don't show up...

As long as by 'occasional' you mean 'almost every day', then sure.

The world of plumbing is one of the most over-standardized industries in the world. Plumbers almost always work in highly controlled environments, because municipalities tend to have very strict building codes for what materials and construction practices are allowed to be used. The types of things that can go wrong in plumbing are always fixable by a reasonably skilled plumber - as long as the money is there to perform the necessary repairs. In fact, the world of plumbing is so standardized now that the biggest problems plumbers have is when working on fixtures that are so old the parts can't be purchased at the local Home Depot.

Teachers rarely have access to the sort of regular problems that plumbers are faced with.

And teachers do? Plumbers aren't likely to have to fix 30+ pipes that are all interconnected in mysterious ways, where some pipes are running slow because they're too busy getting high to let water flow through them, and other pipes are empty because their parent pipes don't feel like it's their responsibility to keep them pressurized.

Scrutiny is incredibly difficult. My company still thinks that LOC is a good metric for measuring developer skills.

And why on earth would you say that the selection of subjects is 'limited'? You think that Math, History, Language, Science, Literature, Government, Art, and all the other subjects taught in public school is limited? I would call the pretty damned comprehensive...

There are dozens if not hundreds of different heuristics that make up whether a teacher is performing well. The problem with optimization is that once you select a function to optimize, it will do that no matter what your actual goal is. Once you 'decide' that a handful of standardized test scores are what we will use to optimize, then you will see those test scores rise and rise and rise to some mysterious limit - and they will still have no correlation to how 'good' a teacher really is.

Here's a counterargument that I don't expect you to give a crap about, but I'm going to include it because I think it's important:

There will always be a distribution of teachers in the workforce. A small handful of really bad ones, a small handful of really good ones, and a vast majority of decent ones.

There will also always be a distribution of students in the classroom. A small handful of really bad ones, a small handful of really good ones, and a vast majority of decent ones.

How would you decide who is a bad student? Would you use standardized test scores?

Now can you think of a metric that measures teacher performance based on standardized test scores that can be normalized to their students' performance on standardized test scores?

It's no more fair to measure a teacher's performance without evaluating their students than it is to measure a student's performance without evaluating their teachers.

Are you trying to make the case that the ambient noise of their rooms does not effect their ability to do their jobs?

Does it affect yours? It certainly affects mine...

Learning is dependent on focus. If there are distractions, kids won't be able to focus, and that will affect their education to a greater or lesser extent. I would say that 'studies' into the ambient noise of teachers' classrooms are performed not so much to measure how well teachers are able to do their jobs, but rather to measure how well students are able to learn in that environment.

Meh to your math...

Teachers normally get paid for a 10-month period, and don't receive paychecks for 2 months during the summer.

If teachers are paid year-round, that same salary is distributed over a 12 month period instead. If you wanted to find out how much LESS per MONTH a teacher would make if paid year-round, you need to multiply by 10/12, not 12/10.

However, this argument is a red herring because teachers make 40-50k a year no matter how many months they work. Salaries are generally determined based on number of days in the school year, not by number of months worked.

In fact, having natural ability hurts far more than it helps. More often than not, the students that display a lot of natural talent early on become intensely frustrated when they reach the extent of their natural gifts, and have to start practicing like their less gifted counterparts. The students that worked hard to achieve their talent are far more likely to advance their skills to the highest level, and be able to cope with the pressures involved at the level of professional performance.

I'm sure that things are similar in any creative field.

Many people say that too much study kills spontaneity in music, but although study may kill a small talent, it is a must to develop a big one. By concentrating on precision, one arrives at technique, but by concentrating on technique one does not arrive at precision. - George Gershwin

All that shows is that 'smartness' has little to do with competence.

The purpose of education is not to be competent at a job; it's to keep from reinventing the wheel every time a previously solved problem comes about.

As an audio engineer, it would certainly be my opinion that these tracks were highly processed after mixing, and possibly before. The amount of noise and distortion from hundreds of cheap laptop/webcam microphones would be horrific. I'm certain they used a noise reduction filter, and an awful lot of additional ambiance/reverb to mask the sonic artifacts.

I wonder if they moved to Chicago...

I disagree. The government is just made up of people. And people, pardon me, are assholes...

Oops, that was supposed to be a response to the GP. Anyone who thinks that this country is theirs and theirs alone is delusional. Same goes with political parties that believe that they represent all of America with their political beliefs.

I think if you're one of 300 million Americans, it's only 1/300,000,000 YOUR country.