ofchurches / naplan_explorer Goto Github PK

At the moment the error bars can extend to impossible values (outside the range of Scale Scores that map to Raw Scores and even below 0! This is misleading making the error at the extremes look even bigger than it really is (even though it is quite big).

Requires me interpolating these from the limited data ACARA gives out

Writing would be slightly different in that there is only one scale that extends across the year levels

This would be useful to show the effect of density in the normal distribution and helpful because ACARA use 90% confidence intervals in their published work which is unusual in the statistical literature but we should be able to replicate their approach.