Free school headlines – stuffed full of selected stats (Part 1)
The New Schools Network (NSN) has published a blurb called The Case for Free Schools.
It would be tempting, but wrong, to write a case against free schools. They have a role to play in meeting the need for school places. And some like School21 are truly innovative. It would be perverse to argue against free schools for the sake of it.
But this doesn’t mean that puff pieces claiming free school superiority should be left unchallenged.
NSN presented four ‘headlines’. I’ve commented on the first two (in italics) below:
Free schools are more likely to be rated Outstanding than ‘council-run schools’
The UK Statistics Authority (UKSA) said comparing the proportion of Outstanding free schools with all other schools isn’t ‘materially misleading’ (email to author 21 April 2017). Nevertheless, UKSA had concerns about such comparative statements which it’s shared with the Department for Education.
If it’s not materially misleading to say free schools are more likely to be Outstanding then it’s also not materially misleading to state:
Free schools are less likely to be Good
Free schools are more likely to Require Improvement
Free schools are more likely to be Inadequate
The data in bold above compares free schools with all other schools. But NSN compared them only with ‘council-run schools’. Does this comparison paint free schools in a more favourable light?
The answer is ‘No’.
Analysis published on local.gov.ukusing Ofsted data up to 31 December 2016 looked at overall grade distributions. This showed:
91% of council maintained schools were good or better compared with 89% of converter academies, 84% of free schools and 65% of sponsored academies.
The analysis recognised that Ofsted changed the way it inspected schools in 2012. Many Ofsted reports pre-date this. Using post-September 2012 inspection data only, the figures show council maintained schools still have the largest proportion of good or better schools:
88% of council maintained schools are good or better, 87% of free schools, 83% of converter academies and 64% of sponsored academies.
The analysis comes with several caveats, however. Pre-conversion ratings for converter academies are excluded. Ofsted’s adopted a ‘risk based’ approach. Schools moving from LA supervision to academy status change the data for both types of school.
In any case, the number of free schools inspected is too small to come to reliable conclusions about their group performance. Figures comparing free schools with other schools should be used with caution (a warning that’s usually missing from statements praising free schools).
Free schools are popular. In 2016, secondary free schools attracted an average of 3.6 applicants per place, compared to an average of 2.4 applicants per place for local authority schools.
These figures are not first-preference only. Including second, third or later preferences gives an inflated idea of a school’s popularity.
UKSA says first preference are ‘reasonable proxy’ for popularity. First preference figures for non-selective secondary schools in 2016 show free schools are indeed the most popular of non-selective secondaries. But the point* difference between free schools and non-selective community (LA maintained) schools is just 0.16. It’s unclear whether such a small point difference is significant or not.
These first two headlines contain no warnings about the use of free school data. The statistics are selectively used. In particular they are used to malign ‘council run schools’ as if they are the lowest type of school. It should be remembered, however, that the majority of schools are still LA maintained despite seven years of aggressive promotion of academies and free schools combined with constant sneering at ‘council run’ schools.
Mercifully there are just four headlines in The Case for Free Schools. I will write about the next two shortly.
*Using the secondary phase pupil preference data “times put as 1st preference” was divided by the “total number of places offered” to find the 1st preference per place offered value for each grouping