Tests: the more 'reliable', the less 'valid'.

The history of testing is full of examples which show that the more 'reliable' they try to make them, the less 'valid' they become.

Here's the latest one I've heard:


"We once did a trial for SATs papers when it was first introduced. It was computer marked and entailed colouring in a small area with a graphite pencil from a choice of four small boxes for each question. One pupil just coloured a pretty pattern down the page without even reading the questions. The results came back showing this particulat pupil above average."