How are you really meant to do it? Is the grade you assign intended to represent the level at which they are 'comfortably' working (how subjective is that!), or the very maximum level you have ever seen them achieve (across a reasonably wide range of topics), or.... I can obviously judge if a student gets a correct answer to a given level question(!). But it's one thing getting the answer correct in a supportive clasroom environment, immediately after they've been briefed and have practiced on a whiteboard for 10 mins, and for another 10 mins with a partner - but it can be another thing doing it totally alone, and not after intense preceeding practice. [Fractions are notorius. Many students can appear to 'do' them at the very end of a lesson - where the AfL and plenary suggests the lesson has been completely effective, and yet the next week the students are vitually back to their prior lesson's level]. The variability in 'context' in which the assessment is made seems, to me, to be huge. Overwhelmingly, I see teachers just increase a students previous grade by a little, or lot - depending upon how they feel they've done - rather than in any way coming to an independent conclusion. Any to have student levels recorded to the nearest sub-level seems excessively precise, given the inherent inaccuracies. How is the system really *intended* to work, and how - best practice - should we *actually* be using it? I'm forming the opinion that it's largely a mechanism to make fairly innumerate, un-scientific, but managerially-inclined SLT's feel 'in control' by being able to quote figures to high precision, in the misplaced belief that they are actually accurate. But perhaps I'm not using the systems as well as I might.