Reply #2: These sorts of tests use the exact same test [View All]

over and over; at least they are supposed to. They are "standardized" and scores can be compared because it is the same test, given the same way, every time. When we proctor a test, we have to read a script; we are not allowed to modify the scripted instructions or procedures in any way.

If we were to give tests that just scored kids on how much they had learned, we wouldn't have to give the same test every time because the point of the test wouldn't be to compare children to each other, but to determine how much they had learned. Publicizing the tests would still be expensive, because the test questions would have to vary from year to year.

I think we could compromise by having tests generate questions randomly from an available bank of questions; that way, no one person would see all the questions, and no tester/test taker would know which particular questions would appear on the test.

If we have to compare and rank kids/schools (which I don't think we should do), we could test sample groups; sort of like a poll to guage what's going on; not every kid needs to take the test, just like not every American needs to answer every poll question. That would give you a reasonably accurate picture, and save hords of $$$$ to spend on actual teaching!