blog roll
|
unburying the lead...,
2011-02-24 18:53:33
| Main |
who supports "peer review"?...,
2011-02-25 17:30:24
capturing the benefits of unobserved differences in instructional quality:
This study of teacher influences on student outcomes is really quite interesting. The amount of variation in student outcomes you can actually pin on teachers is, at most, 20%, and usually much, much lower. The strongest correlation between teaching and student outcomes occur in the most disadvantaged districts.
So let's talk about the free lunch we're being offered as regards those: the study figured that if you had some hypothetical means of accurately assessing teacher performance apart from all the other confounding variables, and you replaced the worst teachers with the best teachers - whom you would apparently entice to the poorest districts by offering them lower pay and a hostile attitude towards their right of free association - you could improve outcomes by a third of a standard deviation. Just to put it in terms we might be familiar with, I know the SATs have a standard deviation of 100 by design, so you're talking about all of 30 points on that exam. Which would be a pretty good improvement if you could capture it year after year, could actually assess teachers accurately, could pull great teachers out of your hat, and could do it all for free. Good luck with that plan, Indiana.
:: posted by buermann @ 2011-02-25 14:25:34 CST |
link
|
|
|
|