[michaelt]

Reliably measuring cheating is of course difficult; we can measure /detected/ cheating, we cannot measure /undetected/ cheating.

And who can blame students for cheating considering the incentives - I was never faced with the choice to cheat or fail, but you can bet if I'd gone $70,000 into debt for school I'd cheat before I'd fail.

I once sat in a computer lab and watched a group of four guys completing an online take-it-when-you-like multiple choice exam for first year mathematics for engineers. The software attempted to prevent copy-and-paste and changing windows to Google, and of course there were instructions saying the exam should be their work alone, and they shouldn't look up answers. One of the guys was on the exam system and would read the question; his three peers on computers next to him would google for and calculate answers. Needless to say, by the time the fourth guy was doing the exam his results had little to do with how much mathematics he had personally learned!