> Arnie Duncan's financial aid package to public education has some
> seriously bad points. The fed Race to the Top program has strings that
> include provisions to rate teachers by their student's scores on
> academic achievement levels. State's without this `accountability'
> provision, can't apply for the money.
There's a paper forthcoming in one of the major economics journals which absolutely hammers the idea of using econometric 'value-added models' to assess 'teacher quality', especially at an individual level. Someone forwarded it to me at work but I can't for the life of me remember the author, publication, or track it down with Google Scholar or anything (I guess because it hasn't actually been published yet). I'll forward the details when back in the office on Monday.
'Value added models' attempt to strip out the effects of socio-economic status, 'prior achievement', etc., everything but the variation between teachers, to show the effect of particular teachers on student test scores. The basic idea in the paper is to demonstrate that substantial 'effects' of teachers show up on student performance _in years before they taught them_, indicating that something is wrong - their conclusion is that student allocation to teachers is somehow biased in ways that are not captured by the other variables.
Cheers, Mike Beggs scandalum.wordpress.com