Post: 2 May 2017
Following is a recommendation in yesterday’s report on primary assessment from the House of Commons education committee:
The Government should change what is reported in performance tables to help lower the stakes associated with them and reduce issues of using data from a small number of pupils. We recommend publishing a rolling three year average of Key Stage 2 results instead of results from a single cohort. Yearly cohort level data should still be available for schools for use in their own internal monitoring.
In its coverage SchoolsWeek has interpreted this to mean that yearly cohort level data would be available for school internal use only. I’ve read it the same way.
This is the wrong approach, and runs against the principle that non-personal public data should be open by default.
There are sound arguments for adding a rolling three year average to the KS2 performance tables and giving that primacy over the annual data. In its report the committee quotes two witnesses, Russell Hobby of NAHT and Catherine Kirkup of NFER, both of whom recommend moving to rolling averages for assessment of school performance. However the idea of reserving yearly cohort level data for the internal use of schools seems to be the committee’s own notion.
Yes, annual data is a poor basis for judging the quality or performance of a primary school. And yes, parents and the general public (and sometimes journalists) don’t always understand that.
There is potential for misrepresentation and misuse, and focusing on a rolling average may well reduce those problems.
But it does not follow that the public should no longer have access to the annual data. A rolling average is not a direct substitute – if the committee accepts that schools will continue to have an internal need for the annual data, why does it assume there are no legitimate external uses? Parents and analysts in the education field may well have ancillary uses for cohort level data that cannot be fulfilled by a rolling average.
Most open data has potential for misuse. Where that misuse is deliberate, open data acts as its own safeguard: it is readily accessible and can be employed as the basis for a counter interpretation. Where misuse is inadvertent we should treat that as a problem of data literacy; explain and educate, rather than hide the data.
I also question whether withdrawing yearly cohort level data from publication is likely to be effective. If the Government published a rolling three year average instead wouldn’t it be possible to calculate the annual results from movement in that average?
Image credit: Great Sankey Primary School by Guy Hatton (CC BY-NC 2.0)