Victor, that's why they aren't being compared to their effect on just one student, but they're entire student population, which is adjusted based on their student populations aptitude across all studies. If you have a bunch of students who pass highly in most of their other courses, so that your twenty students have an average overall of 3.2, and your class' average is 2.7, and another teacher has a class that has an overall average of 2.7, but receive 3.0's, you can adjust the value, it doesn't work as well when you only have 2 to compare, but as you add more sources, it becomes easier to determine who has a lower effect, or is fudging their numbers. You also need a tolerance, to help for inaccuracies. - Jimminy, CoG of FF

And what you'll find is what we already know, that GPA has more to do with SES and other similar external factors than anything else. This is no different a system from what the LA Times used to compare teachers in the LAUSD and it's just as flawed for the same reasons. - Scoble, Alex Scoble

Alex, it's way different, than the LA Times comparison, its not measuring cross grade effectiveness, it's a comparative analysis tool, to measure the differences in teachers grades, during one period, there is no delta, here. - Jimminy, CoG of FF

The LA Times comparison doesn't measure cross grade effectiveness either. Oh, forgot one thing also about your model, usually teachers collaborate with other teachers on children, particularly in elementary school. So any sort of metric based on standardized tests or grades is absolutely ineffective for measuring teacher performance in that type of scenario. - Scoble, Alex Scoble

Alex, yes they collaborate, but not all the time, it was 1-2 times a week for maybe an hour a day, when I was in elementary school, and we stopped doing that in 4th grade. So effective outcome is still on the main teacher. - Jimminy, CoG of FF

Nope, it's every day now. Even in 4th grade. At least in Cassie's school. - Scoble, Alex Scoble

Even proponents to the value added model admit that it has some weaknesses. The one that stuck in my head is the fact that while it seems pretty reliable when you're looking at the extreme ends, the data points in the middle are difficult to interpret. Is a teacher who increased the average percentile from the 60th to the 70th more effective than a teacher who increased the average percentile from the 90th to the 95th? What if the average percentile goes from 95th to 94th? Does the teacher get fired then? I think it's a better method of evaluation than just having administrators randomly audit teachers' classes, and it can certainly help in pinpointing what is effective and what isn't, but I feel like people want to use the methods and the resultant data for purposes it wasn't really intended for. - Victor Ganata

I wold say it is a rough tool, but a teacher who improves his students scores from 70 to 80 is probably better than one who's scores go from 70 to 55. - Alex Scrivener from iPhone

I don't disagree that it can be used as a rough guide, and can help determine what things work and what things don't, but people intend to use this data to fire teachers or determine how much money they earn. It doesn't seem fair that someone who starts with a class at the 90th percentile and only increases it to the 95th percentile ends up earning far less than someone who increases their class's percentile from the 50th to the 65th. - Victor Ganata

Sure, I agree. - Alex Scrivener from iPhone