The case for $320,000 kindergarten teachers

[ERD50]

Is the school board going to question every family as to their income, their drinking habits, their marriage fidelity, their dedication to their kids, and then assign a quality number to a kid?

No, not any more than I would expect them to measure my limb length, lung capacity, body mass, etc to determine how fast I can run the 100M dash. Make me run it and measure my time.

No question in my mind that family environment affects student performance. But that will show up on the entry tests.

Do we need better evaluation criteria? Very likely yes. I could easily be convinced of that. And I also think that smart people could develop some good (enough) tests. I can't be easily convinced that this is unmeasurable to the point of having no overall benefit. That extraordinary claim will require extraordinary evidence.

I'd be happy with small steps. It doesn't have to be a disruptive all-or-nothing thing.

One thing I learned at MegaCorp (it took me a long time), is that when apparently intelligent people are taking what seems like an unreasonable stance on something, there are two steps to take:

1) Think long and hard about your position. If it holds up from a few different angles, and you have conviction, move to step 2...

2) Do they have an agenda to support, and your data conflicts with their agenda? If so, you need to approach the problem differently, because it is now a 'political' problem, not a logical/technical one.

-ERD50

 

[FinallyRetired]

No question in my mind that family environment affects student performance. But that will show up on the entry tests.
-ERD50

Maybe, but I wouldn't bet on it. Entry tests can differentiate between knowledge levels, but not necessarily between capacity to learn. Education has many counter-intuitive issues. Let me give you one example based on a short time when I was a teacher/administrator.

We used student questionnaires to "grade" teachers at the end of courses. We figured that teachers with better student ratings would be better teachers, and that could be somehow reflected in their appraisals. From my experience, I suspected that was a faulty way of measuring teacher performance, since "easy" teachers would get high ratings even though they didn't impart much knowledge.

So we set up a small experiment in which we correlated teacher questionnaires with student performance on subsequent courses, and adjusted for their performance on previous courses. What we found was a negative correlation -- that is, teachers with good ratings tended to have students perform worse in subsequent courses than teachers with bad ratings. That's not to say that teachers should have sought bad ratings, but when a teacher is demanding, and requires students to learn the material, they tend to get worse ratings than teachers who let them slide by with minimal learning.

As a result, we threw out teacher ratings in the appraisal process. We still had ratings, but made them available only to the teachers for their own use, and with appropriate warnings.

 

[BTravlin]

I'm saddened that an educator does not see that the logic is flawed. A may be true, but it does not lead to B.

C is also not true. These things can be controlled for. It isn't perfect, but it can be pretty good and over the long run it should average out. Again, the alternative to not having a perfect measurement seems to be to not measure it at all? I don't think our children are served by that thinking. I guess I should rip out the thermostat connected to my A/C, it's useless as it is isn't NIST traceable

Again, it 'works' in the private sector and they have the same constraints (sales people don't get exactly equal sales regions, engineers don't get exactly equally difficult projects, etc). And it would make sense to look for changes over the course of the year, not the absolutes. That probably accounts for most of the 'objections' to differences in the students assigned to one class or another. With a little tweaking, you can have a pretty good measurement system. At least one good enough to point out that maybe one teacher needs some help from another who seems to be doing better. It'll never be perfect.

-ERD50

She didn't say that A always leads to B just that it is one of the reasons but that wasn't necessarily clear the way I related it.

Measuring the student's current situation regarding their home environment would be nearly impossible.

Are parents interested and involved in their child's education, do they monitor homework and studying?

Or are the parents alcoholics who express little concern regarding the child's education?

Has the child been sexually abused or still is being abused? Same goes for physical abuse?

Is the child more worried about homework or whether or not their will something to eat today?

Does the kid expend most of their time and energy toward athletics or other extracurricular activities vs studying?

Do you expect the Dept of Education to get a handle on all of this?

 

[ERD50]

Education has many counter-intuitive issues. Let me give you one example based on a short time when I was a teacher/administrator.

We used student questionnaires to "grade" teachers at the end of courses. We figured that teachers with better student ratings would be better teachers, and that could be somehow reflected in their appraisals. From my experience, I suspected that was a faulty way of measuring teacher performance, since "easy" teachers would get high ratings even though they didn't impart much knowledge.

... What we found was a negative correlation -- that is, teachers with good ratings tended to have students perform worse in subsequent courses than teachers with bad ratings.

Like you, I'm not surprised at those results at all. I am surprised that your associates expected positive correlation.

That doesn't mean there are not good measures to use.

Side note here: DD uses one of those 'rate your professor' web sites when working out her course selections. I've glanced at them, and I think that in a selective college with motivated students, the reviews are probably a reasonable indicator - these kids are serious about getting a good education and it is reflected in their critiques. I've looked at one that includes our High School - as expected, there is a much higher level of 'noise' there, but I'd say that most were serious evaluations. I'd expect a pretty steep drop off as you get into middle-school and elementary level.

Measuring the student's current situation regarding their home environment would be nearly impossible.

Are parents interested and involved in their child's education, do they monitor homework and studying?

etc, etc, etc....

Does the kid expend most of their time and energy toward athletics or other extracurricular activities vs studying?

Do you expect the Dept of Education to get a handle on all of this?

No - but YES, if you look at the question differently. Just as I stated in my 100M example, don't measure the influences, measure the results. If the kids have a sub-par home life, I'm sure it is going to show in their initial test scores too. Even entering Kindergarten, aren't they going to be behind (on average)?

So we measure improvement.

Take 40 kids from a poor environment, test them and split them evenly between two teachers. If one teacher is incompetent, and one is talented, I'd expect that the proper tests could show the difference as the year progresses. It surprises me that this would seem controversial.

Sure, it is possible for one teacher to get saddled with two bad apples that disrupt the others. But on average this is going to work out. And their administrators should be there to notice and help out.

The same things happen in the private sector - you get assigned to a lousy program and even if you work miracles it may be hard to get that noticed. A successful product tends to make almost everyone look good, even if that success was external to the talent in the group. But then there is another assignment, and another chance, and over the long run the talented are rewarded and the less talented aren't.

The more I think about this (and I think this came out in an earlier thread), the more convinced I am that measuring teachers is easier than measuring people in many other professions. A teacher has 20 different data points (students) in a year, and many opportunities to measure them during the year. An engineer may be on a single product for 18 months or more, and maybe only one final measure of success or failure. The more data you have, the faster averaging works in your favor to smooth out the noise.

-ERD50