Wednesday, 11 May 2016

People who design tests don't care if the fact is right/wrong - it's the child that has to be!

Letter to me earlier:

"I was involved once in designing educational tests at [xxxx] was even then producing curriculum packages with tests included, but for adult learners in the USA. Users could choose what sort of tests they wanted in terms of what sort of distribution of grades would be acceptable. They were never designed to measure attainment of criteria but to produce a pattern of winners and losers -- you suspect as much, I gather, for school SATs at the moment? But it was all quite explicit in those past examples -- I wonder if Pearson have also offered tests with different distributions of results to the UK Government, and if so, which ones they chose. Would an FoI request be useful, do you think?

The other thing that emerged from [xxx] debates was that test items could still be 'good' even if they did employ faulty reasoning or ambiguity, as long as they helped produce the required distribution. Some [xxx] colleagues argued once that just such an item should be left in the menu of questions offered to students, even though the 'right' answer was actually wrong. "