How the Leigh article was reported by Justine Ferrari

http://www.theaustralian.news.com.au/story/0,20867,21765738-13881,00.html

I have this article, and have read it deeply. Justine Ferrari's reporting of this publication is incorrect. In it she says: "The study examined the literacy and numeracy test results of more than 90,000 students with more than 10,000 teachers in Years 3, 5 and 7 between 2001 and 2004, tracking the same group as it advanced through the school system".

Actually, the study examined literacy and numeracy test results for that many students and that many teachers, but not tracking the same group between Years 3 to 7 as she suggests. There were three cohorts' results in the study. The results of the first were from Year 3 (2001) and Year 5 (2003), the second from Year 3 (2002) and Year 5 (2004), and the third from Year 5 (2001) and Year 7 (2003). For each cohort, two test result were examined, two years apart.

Further, Ferrari says, "The study, to be released today by the Australian National University, debunks claims by teacher unions that teacher performance cannot be confidently measured by looking at the results their students achieve in universal tests.". This is actually not true. The study shows that teacher 'performance' can be measured with many limitations. If Ferrari read more deeply she would realise that there are data missing, and that there are many other factors that affect students' achivements missing, too. The author does mention these limitations, and does not use the word 'confidently', 'confident' or 'confidence'.

AND we must be careful about saying that the teachers of students who score well in these standardised tests are 'better'. The students did better, suggesting that the teachers were better. Don't ignore that this analysis was based on 'fixed effects' where there is an assumption that things have not changed with students, teachers, the schools, etc. between the years of the first and second tests.

"Education Minister Julie Bishop said the results made a mockery of claims by the Australian Education Union and Labor that a teacher's performance could not be measured. " She would, of course: she is not unbiased! We must remember that this type of measurement is NOT considered to be the correct or most accurate measurement of teacher performance. It is ONE attempt at figuring out a way to do so. Leigh makes that quite clear, even though he obviously has an agenda.

"The study, by economist Andrew Leigh, also finds that demographic differences between teachers account for less than one-hundredth of the variation in their student scores, suggesting other factors such as a teacher's IQ or classroom skills are much more important." Er, no, he didn't say the latter at all!!

Come on, Justine. This is a paper about statistical measurement, and the possibilities of using it for assessing teacher 'performance'. Although Leigh uses it to back the government's push for differential pay, the content is far from convincing as a model to be adopted even if the stats are excellently executed. Education writers with no idea should perhaps be more careful in what they write and what they convey to the public. They should also not make stuff up. It might be your job, darling, but this subject is far too important for mucking with.