What has to be true before a teacher's mark can count?

By Catherine Wang

Ask who assesses a French secondary pupil, and the answer is: their own teacher. Ninety-six per cent of collège teachers report managing their own assessment process frequently or always, against an OECD average of 77 per cent. That is the highest share among the thirteen countries surveyed, and it means French teachers carry more responsibility for grading their own pupils’ work than teachers in any other country in the sample.

Ask those same teachers whether they feel well prepared to assess, and 26 per cent say yes, against an average of 53 per cent. That is the lowest of the thirteen. That gap doesn’t correspond to a general sense of low-confidence,it is specific to assessment: 82 per cent of teachers in France feel well prepared when it comes to subject content..

Meanwhile both French national credentials, the baccalauréat and the DNB (Diplôme National du Brevet, the lower-secondary school-leaving certificate), now place 40 per cent of the final mark on teacher-marked ‘contrôle continu’ (ongoing assessment set and marked by teachers throughout the school year, rather than in a single final exam).

France is the most stark example of a widespread mismatch between teachers’ responsibility for assessment, and  their confidence in assessing. Many education systems are moving marks out of the examination hall and into the classroom. CES recently reviewed fifteen jurisdictions to ascertain whether assessment literacy is embedded in teacher training and professional development. It turns out that the answer to that question is less interesting than what sits underneath it.

The mandate is not the problem

Nine of the fifteen jurisdictions embed assessment structurally in initial teacher training, according to our coding, though several of these have different rules for different routes into teaching or different providers.

In England, assessment is a mandated core area of the Initial Teacher Training and Early Career Framework and is linked to an explicit Teachers’ Standard tied to qualified teacher status. In Scotland, assessment is a named competency in the Standard for Provisional Registration, the benchmark all eleven accredited teacher-training providers must meet. In Ontario, "Student Observation, Assessment and Evaluation" is a compulsory element of accreditation. Singapore goes furthest: a dedicated, compulsory assessment course sits in every training track, the only such arrangement in the fifteen.

The Netherlands writes its requirement into law. Dutch teachers are statutorily required to be able to develop tests, judge and analyse results, assess the quality of tests and examinations, and select those most reliable. Schools may only appoint teachers who meet these requirements.

So on the headline question, whether assessment sits in the training mandate, the answer across most of the fifteen jurisdictions reviewed here is yes, and that has been the case for years. That makes two further questions even more pertinent: what is contained in the mandate, and what happens once a teacher qualifies.

Everyone teaches assessment for learning

Where our review could see what initial teacher training actually covers, one pattern recurred. Formative assessment is ubiquitous. But there are three measurement concepts that are not: validity, reliability and freedom from bias.

In Wales, two independent sources confirm that validity, reliability and bias content is absent from the professional standards. In Scotland, the training standards do not mention validity, reliability, grading and bias. New Zealand’s coverage is formative-dominant and thin on summative assessment, moderation, reporting, validity and reliability. Even in Singapore, where teachers take a compulsory standalone course on assessment, that course covers reporting, moderation and bias only sparsely. However, one caution applies throughout. These documents show what the standards require, not what training providers actually teach, so absence from the standards is not proof of absence from the classroom. Hong Kong is the exception. There, a survey of practising secondary teachers found weak command of validity and reliability specifically, rather than simply an absence of published requirements.

The divergence in emphasis on these two facets of assessment training matters. Validity, reliability, comparability and freedom from bias are not academic niceties. They are what makes a grade mean the same thing in two different schools. A teacher trained thoroughly in assessment for learning has been trained to help a child improve. That is essential, but it is not the same as being trained to award a mark that an admissions tutor or employer can trust, and which provides the basis for fair decisions. 

A focus on moderation rather than ongoing training

There is very little evidence in any jurisdiction of national expectations for post-qualification training in assessment skills and knowledge. Instead, almost all formal requirements focus on moderating grades, not on training teachers to award them in the first place.

New Zealand’s arrangement is framed as an obligation rather than an entitlement. The ‘standards’ that collectively make up New Zealand’s main secondary qualification (NCEA) must be moderated at a school level, and the qualification authority (NZQA) moderates a sample of these, with assessor tools and workshops built around that process. The focus is therefore on verifying marks, not building teachers’ assessment literacy. Meanwhile New Zealand’s initial teacher training content is dominated by formative assessment and is thin on the three measurement concepts. It is worth noting that NCEA itself will be replaced from 2028.

Hong Kong’s arrangement is a standardisation and moderation strand attached to the school-based assessment component of the Hong Kong Diploma of Secondary Education (HKDSE). Teachers take part in this standardisation and moderation process directly, through within-school standardisation and through marked work shared across schools, with the examinations authority and the education bureau providing training built around that process. Capability there is developed by taking part in moderation rather than delivered separately, though the strongest evidence for that effect dates from the 2005–11 rollout.

The Netherlands runs a statutory two-corrector system on its central examination, in which every script is marked independently by two examiners and any disagreement is resolved before a final grade is issued. This process is universal, annual and cross-school, and appears to be one of the more robust moderation arrangements among the fifteen jurisdictions. We do not count it as professional development at all, because it is an operational examining duty with no development design attached.

None of those three arrangements amounts to a training offer. Each checks one teacher’s judgement against another’s, repeatedly, as a condition of the awarding process, rather than building teachers’ assessment skills and knowledge in advance. Moderation of this kind is no substitute for building expertise directly. 

Leaning hardest, holding least

The systems that rely most heavily on teacher judgement are not, on the whole, the systems that have built structured arrangements for checking one teacher's mark against another's, in the way New Zealand, Hong Kong and the Netherlands do.

The Netherlands shows the split inside a single credential. The centrally examined half of the Dutch diploma involves a two-corrector moderation process. The school-examination half, of exactly equal weight, carries none. The Inspectorate found in 2023 that 43 per cent of schools had skipped the required annual reflection on examination quality, and judged school-based assessment insufficient in almost a third of the 127 departments it examined.

Ontario and Japan both award credentials that are wholly teacher-assessed with no external moderation body. France has responded to its own preparedness gap by tightening national rules around how contrôle continu is used, rather than by building individual teachers’ assessment skills. Finland splits the same way: its matriculation examination is centrally graded, but the lower-secondary certificate that governs entry to upper secondary is set by teachers alone, with no moderation. England provides a marked contrast: exams dominate, with awarding bodies carrying most of the burden for comparability.  Estonia made a similar choice to England, deliberately concentrating qualifications-facing expertise in central examination bodies rather than spreading it across the workforce, which looks less like a gap than a design choice.

Several systems are also now moving toward more external, written assessment: Ontario will reintroduce written examinations in 2026-27, and Singapore will restructure its certification model from 2027.

So the questions worth putting to anyone contemplating moving towards teacher assessed qualifications are these. 

• What has to be true before a teacher’s mark can carry credential weight: teacher capability, verification of the marks themselves, or both? 

• Can verification substitute for capability? 

• Should assessment expertise sit across the workforce, or be concentrated centrally as  Estonia and, through its exam boards, England have chosen? 

• When a system shifts the comparability burden for reliability, validity and freedom from bias from an awarding body to an individual teacher, what has it actually handed over: trust, or technical work?

Catherine Wang is a Research Assistant with the Centre for Education Systems and is completing her MSc in Comparative and International Education at the University of Oxford.

This blog draws on a review she is conducting of assessment literacy in initial teacher training and continuing professional development across fifteen jurisdictions: England, Wales, Northern Ireland, Scotland, Ireland, Ontario, New Zealand, Singapore, Japan, Hong Kong, Finland, Estonia, Poland, the Netherlands and France.