3 ms·

Imagine two groups of students in two different environments. If students in the bad environment are incapable of realizing significant gains regardless of thei

Imagine two groups of students in two different environments. If students in the bad environment are incapable of realizing significant gains regardless of their teachers whereas students in the good environment are likely to manifest gains no matter how bad their teachers are, then comparing deltas won't help. What you'd need to do is rotate teachers into a variety of different environments. I mean, even a mediocre teacher should produce significant gains for a class of wealthy students: the more mediocre the teacher is, the more private tutoring her students will get to compensate. Consider graduates from selective engineering schools like MIT. They tend to be very good. Does that mean that MIT is really good at undergraduate education? Of course not. In fact, MIT is pretty bad at undergraduate education. But that doesn't matter because the world's best engineering students are desperate to attend MIT and MIT can pick and choose. I think this method holds promise, but like the GP, it is hardly a silver bullet.

yummyfajitas 16y ago
If students in the bad environment are incapable of realizing significant gains regardless of their teachers whereas students in the good environment are likely to manifest gains no matter how bad their teachers are, then comparing deltas won't help. In that case, good teachers will avoid such students, and bad teachers will seek them out. This is actually quite a highly desirable result - we won't waste a good teacher on hopeless students, and a bad teacher can't harm them much. As I said, VAM gives teachers an incentive to target the students that they can help the most. Now that I think about it, it also gives bad teachers an incentive to target students they can't harm. The more I think about it, the more I like it.
- MichaelSalib 16y ago
  In that case, good teachers will avoid such students, and bad teachers will seek them out. This is actually quite a highly desirable result - we won't waste a good teacher on hopeless students, and a bad teacher can't harm them much. That might be the case in a world where there was perfect information, no barriers to entry, and no significant asymmetries. In the real world though, a brilliant young teacher may get her start working with children damaged by, say, lead poisoning, and get no results. As a result, she'll be branded as incompetent. Since she's been branded as a bad teacher, she won't be able to get jobs in high performing schools with better students and will never get an opportunity to demonstrate her brilliance. Or she might be so discouraged by her failure that she leaves the profess altogether. You can imagine the equivalent problem faced by teachers of well-off children.
  - yummyfajitas 16y ago
    Nonsense. The students are hopeless and she has no power to help or harm them. Hence, her students perform exactly as well as they are expected to and her VA score is 0 (average). That's the beauty of VAM. If student sensitivity is measured as well, she will simply be branded as "inadequate information to evaluate performance". If you are concerned that she might be really unlucky and get a class full of people worse than their statistical predictor, that's what confidence intervals are for. As for barriers to entry and imperfect information, VAM also doesn't provide green energy or solve P=NP. All it does is improve incentives and give administrators a better measure of teacher performance.

yummyfajitas 16y ago

If students in the bad environment are incapable of realizing significant gains regardless of their teachers whereas students in the good environment are likely to manifest gains no matter how bad their teachers are, then comparing deltas won't help. In that case, good teachers will avoid such students, and bad teachers will seek them out. This is actually quite a highly desirable result - we won't waste a good teacher on hopeless students, and a bad teacher can't harm them much. As I said, VAM gives teachers an incentive to target the students that they can help the most. Now that I think about it, it also gives bad teachers an incentive to target students they can't harm. The more I think about it, the more I like it.

MichaelSalib 16y ago
In that case, good teachers will avoid such students, and bad teachers will seek them out. This is actually quite a highly desirable result - we won't waste a good teacher on hopeless students, and a bad teacher can't harm them much. That might be the case in a world where there was perfect information, no barriers to entry, and no significant asymmetries. In the real world though, a brilliant young teacher may get her start working with children damaged by, say, lead poisoning, and get no results. As a result, she'll be branded as incompetent. Since she's been branded as a bad teacher, she won't be able to get jobs in high performing schools with better students and will never get an opportunity to demonstrate her brilliance. Or she might be so discouraged by her failure that she leaves the profess altogether. You can imagine the equivalent problem faced by teachers of well-off children.
- yummyfajitas 16y ago
  Nonsense. The students are hopeless and she has no power to help or harm them. Hence, her students perform exactly as well as they are expected to and her VA score is 0 (average). That's the beauty of VAM. If student sensitivity is measured as well, she will simply be branded as "inadequate information to evaluate performance". If you are concerned that she might be really unlucky and get a class full of people worse than their statistical predictor, that's what confidence intervals are for. As for barriers to entry and imperfect information, VAM also doesn't provide green energy or solve P=NP. All it does is improve incentives and give administrators a better measure of teacher performance.