Grading Teachers

I blogged recently about how teachers have a huge amount of influence over the outcomes of our children. Because of this leverage, we should take steps to ensure we have the best teachers, the ones who lead to the best outcomes. One of the comments on that post asked how we would actually evaluate teachers, taking into account both social inequalities and influences on the “whole child”.

These are reasonable questions. Children in richer neighborhoods often have better test results and overall outcomes than children from poorer neighborhoods, and school is about more than just test scores. So, how to do you ensure each school offers roughly the same opportunities and how do you evaluate the performance of teachers?

Dealing with inequalities

To deal with inequalities, we can start by eliminating private schools. Or rather, you can have private schools, but not ones that operate during school hours. Every child goes to public school. Rich people who would prefer something else can further educate their kids after school hours. By doing this, I expect it will increase the incentives for public school funding.

Second, any donations or fund-raising for any school should be “taxed” 50%. In other words, the school will get 50% of the income, but the rest is returned to the school district to be redistributed to other schools. Thus, a school in a wealthy area will still be better than a school in a poor area, but the inequality will fall. I think a strategy like this is a reasonable way to ensure a reasonable trade-off to encouraging fundraising without creating an extremely skewed education system.

Ask the experts

Almost every teacher at higher grades evaluates their students through testing. Thus, they clearly believe that testing is a reasonable way to evaluate students’ abilities, so should have no problem with tests being used to evaluate their abilities. What’s more, because of all the testing they do, one could argue that teachers are the experts at creating tests to evaluate people’s performance.

Thus, a simple (though maybe a bit cynical) way to grade teachers is to ask them to come up with a way to grade themselves, to distinguish the best teachers from the worst teachers. First, they would have to determine the important things they should be teaching, and then how to evaluate whether the students are learning those concepts well.

If they come up with a system where every teacher gets an A+ grade, then they will have proven that they are unable to evaluate someone’s performance. But I view this as the most core component of being a teacher. If you can’t even determine how someone is performing, you can’t know on which areas you should focus your teaching. At best, that means you’ll be a grossly inefficient teacher and at worst completely useless. If you’re that ineffective, you should find another job.

Standardized grading

If people are reluctant to allow teachers to set up their own grading system for themselves, then I’d suggest creating a standardized grading system.

With a committee involving teachers, administrators, and parents, come up with a list of what we hope to achieve in each grade. Then, create a huge database of questions related to each topic. Have a bunch of kids in each grade answer the questions so that you have a rough idea of how difficult each question is. Then, build random adaptive online tests that use a small subset of the questions.

By adaptive, I mean giving easier questions when the student gets the question wrong, and harder questions when the student gets the question right. By having the questions randomized, you minimize the chance for teachers to cheat, such as by giving the students the answers to the questions in advance. They’ll still be able to focus on the topics that will be on the test, but, since the test evaluates the dimensions we care about, that doesn’t seem like a bad thing.

In this way, in a relatively small number of questions, you’ll be able to evaluate the student’s abilities in various topics. At the beginning of the year, give the students the test, and then give them another test at the end of the year. The difference between the results is the impact of that year of school, which, in aggregate, should largely be the effect of the teacher.

But what about….

Of course, there could be the temptation for teachers to encourage their kids to sandbag the beginning of the year test so that they have a low hurdle. However, if this is occurring, it should be relatively obvious by comparing year over year results. (“That’s odd. Every one of Ms. Bingly’s students lost 80% of their knowledge over the summer, while everyone else’s students only lost 25%. And this has happened for 3 years straight.”)

Of course, teaching has a human component. So what do you do to ensure that teachers who have a good “soft side” get rewarded?

Nothing. If it’s important, it should be on the tests. If the teacher is actually abusive, then they shouldn’t be teaching anyway, regardless of test scores. Otherwise, put in the test…

The bottom line

I don’t imagine this strategy would be perfect, and it may be completely wrong. But teaching is about evaluating gaps in knowledge and skills, and filling in those gaps. If we believe that teachers are able to evaluate students at all (and I would find it hard to believe many people truly think most teachers can’t do this), then you also ought to be able to use similar evaluation techniques to judge the performance of teachers.