Measure to correct: census assessments and open data for every school in Spain

Democracia 21's launch speech asked for something every party says it wants: measure results in order to adapt policy. We have analysed the speech as we would any party's. In Spanish schools the measuring already happens. What the law forbids is using it to compare schools.

What Spain measures, and what it hides

Under the LOMLOE, every school must run a diagnostic assessment in the fourth year of primary and the second year of ESO. Eurydice's summary of the law is clear: the assessment is census-based, for every pupil, it covers at least language and mathematics, and it is informative, formative and guiding for schools, teachers, pupils and families. On top of that, the national education assessment institute (INEE) runs a sample-based general assessment of the system in the sixth year of primary.

And then the brake. The results of the diagnostic assessments may not be used, in any case, to rank schools. Article 140 of the Organic Law of Education adds that the purpose of system assessments cannot justify using their results for individual assessments of pupils or for rankings of schools, at state or regional level.

The result is a legal tangle. Madrid's 2022 law on freedom of school choice, the Ley Maestra, says the opposite in its article 5: each school's individual results in all the general tests its pupils take shall be public, broken down by stage and by every dimension assessed, including university entrance and the region's own assessments. One law says publish school by school, the other says never rank. In practice families get neither a clear right to the data nor an official, comparable picture of their schools.

So Spain tests every pupil and leaves the results in a legal grey zone. The early school leaving data shows what is at stake: regions with the same basic law range from 3.6% to 20.6%, and the children of mothers with only primary education leave at 32.9%. Those differences start in particular classrooms, years before the age of 18.

What Portugal does

Portugal publishes. Its Ministry of Education runs InfoEscolas, a portal with data on every school. One headline measure is the share of pupils on a "direct path to success": those who complete a cycle, say from the seventh to the ninth year, without repeating a year and with a pass in the national exams. The statistics office's technical note explains the key step: each school's figure is compared with the national average for pupils who had similar results in the exams at the end of the previous cycle, so a school that takes in pupils with difficulties can still score high if more of them succeed than similar pupils elsewhere.

The Portuguese government's 2020 report on the data showed why the comparison matters. Lower secondary direct paths to success rose from 45% to 47% between 2017-18 and 2018-19, and in upper secondary from 37% to 44%. Among pupils on the highest tier of school social support, only 21% in lower secondary were on a direct path, against 56% of pupils without support. The report also found schools in disadvantaged areas doing better than others in more favourable ones, and the government set out to study them. It announced a restricted area for schools with socio-economic data by class, access for the inspectorate to use the data in external evaluation, and an indicator of "overcoming the context" to measure each school's contribution to social mobility.

The proposal

The left is right that school data should send help where it is needed, never shame the schools that teach the hardest-to-teach children. The market-liberal tradition is right that families have a right to know how their school is doing, and that public money should answer for results. Karl Popper's piecemeal engineering adds the piece that fits neither camp: a reform is a hypothesis, and you only learn from it if you look at the results and correct.

So:

• Publish what each school adds. Amend articles 140 and 144 to allow, and require, publication of each school's diagnostic results adjusted for its intake, on the Portuguese model, together with the gap between pupils who receive grants or the minimum income and the rest. Raw averages and individual pupils stay confidential. A school is judged by what it adds and by how well its poorest pupils do.
• Attach help to every weak result. A school more than a set margin below expectation for two years in a row gets an automatic support plan: extra teaching hours for small-group tutoring, a mentor team from a school with similar intake and strong results, and the right to choose its own measures.
• Turn the fixes into evaluated pilots. Where several schools adopt a measure, the region assigns it in stages, with a comparison group, and publishes the result. Jennifer Pahlka's lesson on delivery applies: build the data pipeline once, with schools' input, so that results reach teachers within weeks.
• Open the anonymised data to researchers. The INEE and regions should release pupil-level microdata, pseudonymised, with the safeguards statistical offices already apply to microdata.

Spain already tests every pupil twice during compulsory schooling. Publishing what each school adds, and acting on it, turns that measurement into correction.