Can a Few Years’ Data Reveal Bad Teachers?

Sotto riporto un assaggio dell’articolo scientifico che dovrebbe dimostrare la tesi dei tre economisti. Il tecnicismo è esasperato. L’analisi sembra poggiare su una quantità di ipotesi statistiche difficilmente verificabili. Il tema in compenso è caldissimo negli USA perché è in atto il tentativo di nascondere gli effetti del disagio sociale e dei tagli al sistema scolastico, sostenendo che le colpe ricadono sugli insegnanti “fannulloni” e iperprotetti dai loro sindacati. Molto istruttivo il “class-size debate”:

“does decreasing the number of students in a classroom improve learning?”

“According to recent statements by Education Secretary Arne Duncan and Bill Gates, for example, great teachers do just fine with oversize classes.”

In particolare, un economista di Stanford (E.A. Hanushek) sosteneva che non valeva la pena di ridurre le dimensioni delle classi, ma A.B. Krueger (Princeton) gli ha smontato la meta-analisi mostrando che aveva pesato i risultati della letteratura in modo anomalo (http://www.avongrove.org/district/Newsroom/ClassSizeTaskForce/ArticlesResearch/ClassSizeDebate.pdf#page=16). Gli interessi economici e politici sono alti.

Dall’articolo “The Long-Term Impacts of Teachers: Teacher Value-Added and Student Outcomes in Adulthood.”

“Value-added estimates consistently measure teacher quality only if they are uncorrelated with unobserved components of student scores. A natural first test of this identifying assumption is to examine the correlation between our estimates of VA and variables omitted from standard VA models.We use two sets of variables to evaluate selection: parent characteristics and prior test scores.

Parent Characteristics. The parent characteristics from the tax data are ideal to test for selection because they have not been used to fit value-added models in prior work but are strong predictors of student achievement. We collapse the parent characteristics into a single index by regressing test scores on mother’s age at childís birth, indicators for parentís 401(k) contributions and home ownership, and an indicator for the parent’s marital status interacted with a quartic in parent’s household income.”

Interessante questo commento di un lettore del NYT:

“I don’t think the Times or PBS are ever going to acknowledge that for which they BOTH should be ashamed. They both pushed the results of this study – one that was not published in any journals, nor was it peer reviewed whatsoever. This study, although deriving from researchers at two of our top institutions, was never sent out for external review, and it was actually never internally reviewed by the founding organization’s editorial board. This is embarrassing, not only for the research community but for NYTimes and PBS as well.

Just because the institutions from which these researchers came are noteworthy, this does not mean that the researchers got their analyses correct, particularly given the theoretical and ideological frameworks with which they approached these analyses, in particular, as econometricians. Thanks to Jesse Rothstein, also an econometrician, for countering.

In my professional opinion, this article will never be published in any well respected peer-reviewed journal because of its faults and the gross negligence of the econometricians who, by default, simplistically measured inputs and outputs regardless of the issues (e.g., the VA method used, the assumptions made there, the longitudinal linking of data, the statistical controls used and whether they wiped out extraneous variables, the test scores used pre-NCLB, etc.).

We need to hold our media accountable, possibly more than our teachers in this case.

Audrey Amrein-Beardsley
Arizona State University”

Vedi anche:

http://www.nytimes.com/roomfordebate/2012/01/16/can-a-few-years-data-reveal-bad-teachers/lets-not-rush-into-value-added-evaluations

1 commento

Giuseppe De Nicolao 28 Febbraio 2012 At 02:41
Sotto riporto un assaggio dell’articolo scientifico che dovrebbe dimostrare la tesi dei tre economisti. Il tecnicismo è esasperato. L’analisi sembra poggiare su una quantità di ipotesi statistiche difficilmente verificabili. Il tema in compenso è caldissimo negli USA perché è in atto il tentativo di nascondere gli effetti del disagio sociale e dei tagli al sistema scolastico, sostenendo che le colpe ricadono sugli insegnanti “fannulloni” e iperprotetti dai loro sindacati. Molto istruttivo il “class-size debate”:
“does decreasing the number of students in a classroom improve learning?”
“According to recent statements by Education Secretary Arne Duncan and Bill Gates, for example, great teachers do just fine with oversize classes.”
In particolare, un economista di Stanford (E.A. Hanushek) sosteneva che non valeva la pena di ridurre le dimensioni delle classi, ma A.B. Krueger (Princeton) gli ha smontato la meta-analisi mostrando che aveva pesato i risultati della letteratura in modo anomalo (http://www.avongrove.org/district/Newsroom/ClassSizeTaskForce/ArticlesResearch/ClassSizeDebate.pdf#page=16). Gli interessi economici e politici sono alti.
Dall’articolo “The Long-Term Impacts of Teachers: Teacher Value-Added and Student Outcomes in Adulthood.”
“Value-added estimates consistently measure teacher quality only if they are uncorrelated with unobserved components of student scores. A natural first test of this identifying assumption is to examine the correlation between our estimates of VA and variables omitted from standard VA models.We use two sets of variables to evaluate selection: parent characteristics and prior test scores.
Parent Characteristics. The parent characteristics from the tax data are ideal to test for selection because they have not been used to fit value-added models in prior work but are strong predictors of student achievement. We collapse the parent characteristics into a single index by regressing test scores on mother’s age at childís birth, indicators for parentís 401(k) contributions and home ownership, and an indicator for the parent’s marital status interacted with a quartic in parent’s household income.”
Interessante questo commento di un lettore del NYT:
“I don’t think the Times or PBS are ever going to acknowledge that for which they BOTH should be ashamed. They both pushed the results of this study – one that was not published in any journals, nor was it peer reviewed whatsoever. This study, although deriving from researchers at two of our top institutions, was never sent out for external review, and it was actually never internally reviewed by the founding organization’s editorial board. This is embarrassing, not only for the research community but for NYTimes and PBS as well.
Just because the institutions from which these researchers came are noteworthy, this does not mean that the researchers got their analyses correct, particularly given the theoretical and ideological frameworks with which they approached these analyses, in particular, as econometricians. Thanks to Jesse Rothstein, also an econometrician, for countering.
In my professional opinion, this article will never be published in any well respected peer-reviewed journal because of its faults and the gross negligence of the econometricians who, by default, simplistically measured inputs and outputs regardless of the issues (e.g., the VA method used, the assumptions made there, the longitudinal linking of data, the statistical controls used and whether they wiped out extraneous variables, the test scores used pre-NCLB, etc.).
We need to hold our media accountable, possibly more than our teachers in this case.
Audrey Amrein-Beardsley
Arizona State University”
Vedi anche:
http://www.nytimes.com/roomfordebate/2012/01/16/can-a-few-years-data-reveal-bad-teachers/lets-not-rush-into-value-added-evaluations
Entra per lasciare un commento

LASCIA UN COMMENTO Cancella la risposta

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.