[WalterBright]

In Washington State, teacher salaries were raised substantially, with no change in educational results.

The thing is, just giving everyone raises doesn't work. It needs to be based on merit.

I've proposed a system where teachers get a base salary, plus a substantial increment for every student in their class that meets grade level expectations at the end of the school year.

[ejb999]

>>plus a substantial increment for every student in their class that meets grade level expectations at the end of the school year

Don't be surprised when all of a sudden almost every student magically 'meets grade level expectations' and teachers get their large bonuses, and yet many of the students can't actually read when they graduate. Unfortunately, this is just an incentive to inflate grades and fudge results to make them look better.

[WalterBright]

Naturally, the teacher won't be in charge of evaluating their own students.

[ejb999]

But there is no real firewall between teachers and administrators - most administrators are former teachers - it's a pretty cozy club in most schools.

I am not opposed to the idea in theory, but don't see any way to honestly make it work. Also fairly certain almost every teachers union will oppose it - and I say that as someone who has the experience of negotiating teacher's union contracts from the other side of the table.

[WalterBright]

Do what other organizations do. Have a separate organization to the evaluations.

> Also fairly certain almost every teachers union will oppose it

Of course they will. Those unions are absolutely, irrevocably committed to the idea that teacher merit is totally determined by length of service and having a master's degree.

[idontpost]

And of course the people who are won't have any conflicts of interest of their own.

[WalterBright]

There are well known techniques to deal with this. It's hardly a new problem.

[citizenkeen]

There are factors other than teacher's merit that affect outcome. You're saying that a teacher who gets two learning-disabled students (who need to be paired due to sibling issues) is less deserving than their peer who didn't have the disruption of two learning-disabled students?

The moment you tie income to results, you incentivize teachers leaving behind students who won't perform.

[bombcar]

It's amusing (and sad) that almost everyone involved with a school can quickly point out the best and worst teachers there, but there's no way to "programmatically" encode it so that the bureaucracy can do something about it.

[idontpost]

Because it's not that hard to measure teacher quality. But once a measure becomes a target, it ceases to be a good measure. It gets gamed and manipulated. When it was just a measure, no one bothered to game it, so it wasn't that hard to tell who was good and who wasn't.

[troupe]

Imagine a system where of 500 teachers, 25 that have a track record of doing much worse than average are let go. You then roll the dice to get new teachers.

Could you pay more to attract better teachers in that situation and get better overall value for the kids being taught?

So how do you identify poor teachers? I would guess that if you let each member of the faculty vote for the 25 top teachers and then have parents do assessment of their children's teachers and then average the results over 3 years, you could come up with something that has virtual no false positives as to who the worst teachers were.

[ALittleLight]

Evaluating teacher quality is difficult and impossible to do perfectly. Even a simple effort to evaluate quality would be vastly superior to the current state where quality isn't even a consideration.

I agree with the ensemble of evaluations approach - student tests at the start and end of year show student attainment, student and parent feedback, peer feedback, and administrator feedback. Come up with a weighted average and experiment with it. Retain average and better teachers, reward rockstars, train underperformers, and let the bottom ranks and those who don't improve with training go.

[WalterBright]

Yeah, I've noticed that, too. It's the same in every office in every organization.

[WalterBright]

The idea is to randomly assign the students to teachers. Sometimes teachers will get students who will never reach grade level, and sometimes students who will effortlessly achieve. By being random, it evens the opportunity out.

[citizenkeen]

That sounds horrible. Right now, schools try to give each teacher an even number of high-performing, average, low-performing, and “difficult” kids. But the numbers are so small it’s hard.

An average sounds like a shitty, shitty system. Imagine finding out your kid is in a classroom with every low-performing and difficult kid in the grade just because it was random.

[glmdev]

The concept sounds good in theory, but I think it's going to be nigh unworkable in practice. The NCLB/high-stakes testing era exposed many problems with tying educator pay to student outcomes -- chief among them that student outcomes didn't improve.

[WalterBright]

As I recall they had the teachers themselves graded those tests, so naturally they cheated.

[softwarebeware]

> In Washington State, teacher salaries were raised substantially,

Can you cite where you saw this data, what time period it refers to, what "substantially" actually means, and include a comparison against inflation, please?

> ... plus a substantial increment for every student in their class that meets grade level expectations at the end of the school year.

I think basing teacher pay on merit is a great idea in theory but I have some problems with it. Most of all, student performance is influenced to a greater degree by things outside the teacher's control like the student's socioeconomic status, their attendance (or lack thereof), their parent's education, and even the air quality in their school.

I also don't know how "teacher" raises based on "meeting grade level expectations" would work past elementary school when students are cycling through seven teachers a day? Just because one child excels at math do you give the math teacher a greater raise?

Finally, what if a student does NOT meet grade level expectations, but shows the greatest improvement year-over-year against any other student. Do you fail to recognize the achievement of the teacher who improved this student's outcome because the student does not meet grade-level expectations?

These are just some of the problems which make this a much thornier issue and worth greater consideration. It sounds good when you say teacher pay should be based on merit, but it oversimplifies things quite a bit.

[WalterBright]

> Can you cite where you saw this data

Not offhand. It was the topic of the Seattle Times for months.

Let's take a look at the private sector. Pay is based on accomplishing goals. It works well. Sure it is imperfect.

> it oversimplifies things quite a bit.

It would be hard to be worse than the current system, which simplifies merit as "has a masters degree". I answered your other points in other replies in this thread.

[softwarebeware]

This is just muddy thinking and hand waving. “Pay in the private sector” is not based on accomplishing goals at all. If it were, there wouldn’t be a huge disparity between men and women, or between those who live in Kansas City vs those who live in San Francisco. Or between those who are over six feet tall and those who are under. And you also say this as if accomplishing goals isn’t figured into teacher pay.

[WalterBright]

> “Pay in the private sector” is not based on accomplishing goals at all.

Where does one start with such an egregiously wrong statement? Companies go out of business if they have employees that cost more than they produce. They can't just raise taxes like the schools do. They can't run deficits like the government can. What do you think regular performance evaluations of private sector employees are for? Why do you think a lot of employee compensation is commission based?

> those who live in Kansas City

Think about it. The only reason pay is higher in SF is because it is more productive to have people in SF, i.e. people accomplish more being in SF.

> as if accomplishing goals isn’t figured into teacher pay.

It isn't, other than getting a master's degree. Which teachers in the government sector are paid based on how well their students do?

[xur17]

> The thing is, just giving everyone raises doesn't work. It needs to be based on merit.

Agreed - to me it seems like paying teachers more is a tool. This enables you to offer incentives to improve student performance, but it also enables you to raise the bar (in some way) when hiring to get better teachers. If you just raise existing salaries across the board, you'll see no immediate change other than more people applying to be teachers. In theory if you have a good way of filtering for the "best", you might be able to then slowly overtime replace your existing teachers with (on average) better teachers.

[WalterBright]

Once hired, good people still need incentives to perform. This is well understood in the private sector.

[TimPC]

So all the teachers want to work in rich suburban schools where students perform above grade level and no one wants to work in poor inner-city schools where they don’t? The only reasonable thing you can base merit pay on is a delta. You test a performance difference between incoming and outgoing students. If a student starts a year at grade level and makes no progress you shouldn’t reward the teacher for having them at grade level.

[WalterBright]

All the teachers want to work in a rich suburban school already.

Basing the bonus on the increment is probably a good improvement.

[TimPC]

Plenty of teachers are willing to work in an environment where they feel they could make the biggest impact. My mother spent her whole career working in schools that could be classified as inner city with high percentages of new immigrants. Not everyone wants the easiest job possible in their field, some people see rewards in being able to make a bigger difference in people's lives.

[WalterBright]

That doesn't seem to line up with your previous post?

[TimPC]

My point is don’t actively screw those people by coming up with a dumb compensation formula that punishes them for students underperforming before they ever entered their class. If you have merit based pay you need to evaluate how much a student improved under a specific teacher’s watch.

[WalterBright]

Even a blunt results based compensation would be far better than the current situation.

[bombcar]

That last has been basically tried, and you end up everyone always "exceeding expectations" unless you tie it to some sort of standardized testing (which has its own issues).

Your best bet may be removing the obstacles that people who want to do a good job encounter (for example, the best teachers seem to often leave higher paid public teaching jobs to go to private schools that pay less - investigate why?).

[WalterBright]

There are always going to be issues around evaluating student achievement. But by and large, compensation based on results works very well in the marketplace. And we certainly see the results in the public schools of no merit pay. It's hard to see how it could be worse.

BTW, MIT has gone back to SATs. The reason is simple - despite all the controversy about SAT validity, when the rubber meets the road the SAT does a better job than anything else at evaluating candidates.

[zehaeva]

Great Idea!

We just tell the teacher that "hey you'll get a bonus if everyone in your class gets an A"!

Who's in charge of giving their students grades again?

[WalterBright]

Grade inflation has been around a long time under the current system. It's also why the teachers union tries so hard to get rid of all testing.

[panzagl]

Cool, so no one will ever teach anywhere except a rich suburb.

[WalterBright]

Same as now.