I find it disturbingly odd that such assessments continue to coexist with the understanding we have for how magnificently messy real learning really is.
I think that is why I love this quote from Cathy Davidson:
One reason item-response testing (the twentieth-centurys dominant method of testing) is so deficient is that it tends to reduce what we teach to content (especially in the human, social, and natural sciences) or calculation (in the computational sciences). Think of the myriad ways of knowing, making, playing, imagining, and thinking that are not encompassed by content or calculation.