Village primary schools and specialist settings send us extracts where a ‘group’ is three names. Publishing 67% for that group is mathematically tidy and educationally misleading. One pupil’s absence on results day moves the figure by more than any teaching change could.
Our default is to suppress percentages below a threshold agreed with the school — often five or ten pupils, depending on how identifiable the group is in a small community. The cell shows an em dash. The footnote says why.
Banding is the middle path: ‘fewer than five’ or ‘five to nine’ instead of a point estimate. Training providers face the same problem on niche apprenticeship standards. A board that insists on a ranking of six learners is asking for a chart that will not survive a complaint from a parent or employer.
We still show counts. Governors can hold the idea ‘four of nine pupils eligible for pupil premium met the expected standard’ without converting it into a percentage that will be quoted out of context in the village Facebook group.
If your last pack printed every percentage regardless of n, ask the chair whether those figures were actually used in discussion. In our experience they are quoted once, then quietly ignored because everyone in the room already knows the names.