Monday, May 24, 2010

Teacher Evaluation

It was my unfortunate experience to have to write an essay last weekend on school voucher programs. The problem with writing such an essay is that most of the justification for vouchers is US government structures of (a) local school district property taxes being the funding source and (b) the constitutional and philosophical objections to any Government aid to private schools.

Once these institutional factors are taken out the argument revolves around choice and effectively the supposed efficiency of markets and the perceived superiority of private (and hence competitive) schools.

The Australian discussion is pointless becaue of the different institutional structure and the information problem in deciding what good looks like. After all, if we can solve the latter problem then choice applied to the public system should generate the same outcome.

An article today goes to the core of the problem - the failings in existing programs for teacher evaluation. This is not a uniquely Australian problem, nor is it confined to public schools.

Ultimately more than test scores need to count. To the extent test scores do count they need to be adjusted to refer to "test score value added" - how much the individual score has changed from one test to the next. They possibly also need to control for other socio-economic factors (the child who moves school from a less performing to a better performing school would be expected to improve scores even with the same teacher due to some form of Westinghouse effect).

The article and the underlying report from the Grattan Institute unfortunately focus on why evaluation is important, but not perhaps on how to do it.

The problem is the same shared by all employee evaluation systems. If there is no immediate consequence from the evaluation then neither the employee nor the manager takes the process sufficiently seriously. The problem is that once a consequence (usually pay) is added then you get distortions from the consequence outweighing the purpose. Thisbecomes particularly bad in the corporate model where the emphasis is on "at risk" remuneration rather than benefits.

So here are my simple suggestions;

1. Start gathering the data to rate teachers on their test score value added, but the first cycle is only to work out the model that corrects for socio-economic effects.

2. Introduce into principal remuneration a performance based score on how well they implement teacher evaluation in their school. One key score needs to be teaching staff assessment of the quality of the evaluations.

3. Introduce peer ratings of teachers as part of the evaluation process including evaluation of overall contributions.

4. About three years into the process start permitting principals to reward the best performing teachers with pay rises beyond the scale, not bonuses. The principals salary pool needs to be based on the schools overall "rating" based on score value added, and parent evaluation.

5. At the same time schools with underperforming teachers to be given additional budgets for both "remedial teacher training" and the cost of casuals to replace the teacher being trained. The teacher needs to understand that the remedial training is just that.

6. Three more years later start to introduce rules to allow teachers to be managed out for underperformance.

No comments: