Research done by 3 economists to narrow down the effects of teacher characteristics. Didn't use a randomized dataset, used a highly specific – 3 Chicago HS, all underpriviledged areas, matched student w/teacher characteristic, tried to control for everything possible.
If teachers move 2 SD on a scale of quality student outcomes improve 45%
Controlled for SO MUCH that whatever was left over was attributed to "unobservable factors" –
To say that "unobservable factors" are casual in student achievement is a little weird…They CAN see that some teachers are better than others, but hard to find what you can attribute it to.
SL: Doesn't mean it's not observable – just means they don't have it in this data set that they have.
Loeb & Lankford
Value added – link students and student outcomes to teachers over time to try to measure the correlation between teacher quality and students. States don't collect this other than Texas, some districts do.
Training for Teachers
Cost-effectiveness and training, key findings:
Teacher training for teachers who worked in nonreligious branches led to an improvement in math & reading scores, religious schools not so – no significant improvement.
Teacher training may be as cost-effective as lengthening the school day and cheaper than reducing class size.
SL – what's the applicability? Generalizability? It's nice to see something concrete that works, but how do we take these little studies and apply them at another level?