Monday, March 05, 2007

Tuesday Column: Paying teachers for performance

Who’d want to be a teacher? Whether it’s back to basics, a national curriculum, or the latest: payment by results, politicians are forever drawing up new schemes designed to help teachers teach better.

Payment by results, apparently up for discussion at an education summit convened by the minister Julie Bishop next month, is particularly insulting.

The inference is that teachers aren’t being paid to deliver results at the moment.

Of course not everyone finds such an idea insulting. Australia’s chief executives, among them those whose performance has been the worst, seem to love it.

A few years back I was at the annual general meeting of Bluescope Steel, at the time chaired by Peter Smedley then lorded as a business genius for his stewardship of the Mayne health and logistics group, since broken up.

An elderly shareholder sitting near the back asked him why he felt the need to pay his chief executive a performance bonus on top of his salary.

Smedley replied in what I took to be a condescending tone that that in fact the chief executive received three salaries: a wage, which he said was a payment “for turning up”..., a short-term bonus for keeping the share price high in the short-term, and a long-term bonus for keeping it high in the long-term.

I thought it disturbing that one of Australia’s major public companies would pay its chief executive a six or seven figure salary “for turning up” – Smedley’s exact words – and I think that’s what disturbs me now about the idea of paying teachers extra for “performance”.

That, and the fact that performance is hard to measure.

Auditors love to tell this story: When Sydney had a problem with its trains running on time it collected statistics on every arrival and departure from every platform and compared it with the timetable. Drivers were instructed to more closely match the timetable. So they shot past stations without picking up passengers. I know, it happened to me. But it improved the drivers measured “performance”.

Auditors call this a misspecification – picking the wrong target.

But it wouldn’t happen in teaching would it? After all, the results of teaching can be measured quite precisely at the end of each year in standardized tests.

Well, yes it would.

The state of Iowa in the United States has been measuring primary school teaching outcomes for decades now as part of the Iowa Test of Basic Skills.

In 2002 a University of Chicago economist Steven Levitt played a hunch. Examining the test results for the eight years to 2000 he noticed that some teachers seemed to always to improve the academic performance of every class they taught. But those students seemed to regress to where they had been as soon as they moved on to different teachers.

Using high-level statistical analysis he concluded that 4 to 5 per cent of teachers in Iowa had cheated by “adjusting” the answers of their students after those students had completed the exam papers. (It helped that students were given pencils with which to complete the exam papers and that pencil is easily rubbed out.)

I have experienced a lesser form of teacher cheating myself. Many universities now ask students to grade their tutors at the end of each semester. Sitting in on a tutorial for a student who was ill I saw the tutor hand around biscuits and tell the class how much he loved each and every one of them before distributing the evaluation forms.

One of mantras chanted by those in the United States administering the No Child Gets Left Behind program which links school funding to test results is that “What gets measured, gets done”.

Critics of this type of program say that that is its weakness.

Its supporters in Australia may be surprised to know that it has been tried here before. I certainly was.

Last week the ABC’s Background Briefing radio program searched back into history to outline the results of the Victorian education department’s “Payment By Results” program introduced in the 1860’s.

Education historian Professor Richard Selleck explained that each Victorian teacher received a base salary, and then an extra salary dependent on the results that their children got (in much the same way as would Australian CEO’s more than a century later).

Inspectors would come into primary school classrooms and set and mark exams. The percentage that each class got in the exam would feed directly into the teacher’s performance pay.

As Selleck said: “In one sense when you went into a classroom, you were looking at your financial future, if you were a teacher.”

So teachers worked to ensure that their students did well.

As Selleck put it: “Teachers studied minutely the questions each inspector was likely to ask in reading, writing and arithmetic, whatever the course was. They passed them on to each other, they reduced their curriculum to the narrowest one possible, abandoning subjects that were not going to be examined and concentrating almost entirely on what was examinable.”

On AM last month the Education Minister Julie Bishop described the evidence to support performance pay for teachers as "overwhelming". She promised to release that evidence ahead of her April summit with state and territory ministers.

The most persuasive-sounding evidence to date comes from the Department of Education Reform at the University of Arkansas. It has examined the fledgling merit pay scheme in Little Rock, Arkansas funded by the Walton family which owns Wal-Mart. It finds that it has boosted maths results by 4 percentage points above those of schools without performance pay.

But it has come in for criticism. The associate editor of the Arkansas Times, Warwick Sabin put it this way on Background Briefing:

“Well I'll be very blunt with you. The Department of Education Reform at the University of Arkansas is only about a year old and it was established with money from the Walton family, which has a very public interest in pushing merit pay… They did not pick schools that already were performing very well, because those schools would not produce the improvement that the merit pay advocates would like to see.”

Merit pay may well have some merit. But until Julie Bishop releases the evidence she has assembled, we will have to be content with the publicly available evidence, which is far from conclusive and points to pitfalls. Many of us would be reluctant to entrust our children to teachers who were focused almost exclusively on measurable results.

The ANU economist Andrew Leigh is one of Australia’s leading proponents of performance pay for teachers. But, in apparent contrast to the minister, he is aware of its potential limitations. He says that designing a good merit pay system is hard but not impossible.

He told me that if he was advising the minister he would be suggesting setting up a series of different trials, using different methods of rewarding different types of performance in different schools. He says then we would know what education authorities in Little Rock and elsewhere still do not – whether rewarding teachers for performance works as well or better than rewarding CEOs, or whether it is impossible to reward the right things.