25 May 2009

Chris Woodhead

Chris Woodhead, former head of OFSTED, has said that OFSTED is an irrelevance and part of the problem in education.

The quote that caught my eye is that inspection is "an exercise driven by analysis of the data"

I think that is a wonderful point (not that his other arguments are bad). One of the key changes of the last twenty years or so is the large advance that has been made in analysing data, due to cheap computer power and modern statistical techniques.

These techniques are powerful enough to quantify and compensate for a degree of inaccuracy in the original data. However, I fear that demand for data has been met by abandoning quality to a degree few people realise.

This is not just about education. This, after all, is the story of the credit bubble. It is the story of the endless senseless health scares. It is also, I would suggest, the story of climate modelling. In each case there is a machine which demands data, day in, day out. If you don't have good data, you give it whatever you've got.

Never before has so much data been collected with so little concern for its real value. The statistical techniques were initially used on data that had been collected by enthusiasts the way philatelists collect stamps, and they were so successful that they led to the modern hunger for datasets. Surveyors and inspectors now amass raw data by the megabyte, and whether the data is good or bad, their pay is just the same. Once it's in the computer, and I speak from experience here, it's always easier and more satisfying to refine the processing than to check the input, but if the inputs are bad enough, the processing is worthless.

No comments: