Spain Certified How Universities Evaluate Teaching — Not the Survey Itself. The DOCENTIA Lesson.

Spain made a quietly radical choice about course evaluation: it did not impose a national questionnaire. Instead, through ANECA's DOCENTIA programme, it certifies each university's own model for evaluating teaching. The lesson for the rest of Europe is uncomfortable for anyone still shopping for the "right" survey: quality in teaching evaluation does not come from a common instrument. It comes from a defensible, transparent, actually-used evaluation system — and that is what a regulator can and should certify.

If your institution is still debating which five Likert items to standardise across every faculty, DOCENTIA is evidence you are optimising the wrong layer.

What DOCENTIA actually is

DOCENTIA (Programa de Apoyo a la Evaluación de la Actividad Docente del Profesorado Universitario) was launched by ANECA in 2007 and substantially revised in 2021. It is not a survey. It is a framework of guidelines that each university uses to design its own model for evaluating the teaching activity of its academic staff — after which ANECA, working with Spain's regional quality agencies, audits and certifies that model.

Three features make it distinctive:

• The university designs the instrument; the agency certifies the design. ANECA does not hand universities a questionnaire. It sets criteria — the model must be valid, must combine sources of evidence, must feed into recognition and improvement — and then evaluates whether the university's home-grown model meets them.
• It is a coordinated, multi-agency system. DOCENTIA is run by ANECA together with regional agencies such as AQU Catalunya, ACSUCYL in Castilla y León, and others, through a joint follow-up committee. A regional agency like ACSUCYL certifies the model for institutions in its territory.
• Certification is time-limited. An implementation certificate is valid for a renewable period of five years, so a university cannot certify a model once and let it ossify.

Adoption is not marginal: around 90% of Spanish universities participate in DOCENTIA. That is close to a national consensus that teaching should be evaluated systematically — reached without a single national form.

The design principle everyone else keeps missing

Most European debates about course evaluation are debates about the instrument: which items, which scale, which benchmark. DOCENTIA moves the object of quality control up a level, from the questionnaire to the evaluation model — the whole apparatus of what evidence is collected, from whom, how it is weighted, and what happens to the result.

This matters because the instrument is the least important part of a teaching-evaluation system. A perfectly worded five-item student survey attached to no consequence, read by no one, and triangulated with nothing is worthless. A modest survey embedded in a model that combines student feedback with peer observation, self-report, and outcomes data — and that visibly feeds recognition and development — is valuable. DOCENTIA certifies the second thing.

It also implicitly concedes a point that the research literature settled long ago: a student satisfaction score is one input, not a verdict. Averaging Likert responses into a single number and treating it as "teaching quality" is a well-documented category error — the distribution matters more than the mean, and the mean is contaminated by class size, discipline, and grading leniency (why averaging Likert scores misleads). By certifying a model rather than a number, DOCENTIA structurally discourages the single-metric fallacy.

Spain is not alone in standardising the process rather than the form. France mandates evaluation through the HCÉRES process without dictating the instrument; Italy's ANVUR/OPIS model took the opposite route with a common national questionnaire. Reading DOCENTIA against ANVUR is the clearest way to see the trade-off: a common instrument buys comparability at the cost of local validity; a certified model buys local validity at the cost of comparability. There is no free lunch, only a choice about which you need more.

But doesn't a certified model just let universities mark their own homework?

This is the strongest objection, and it deserves a straight answer. If the university designs its own evaluation model, what stops it from designing a lenient one that flatters its staff?

Three things, in principle. First, the model is externally audited against explicit criteria before it is certified, and re-audited every five years — self-design is not self-approval. Second, DOCENTIA is embedded in the wider Spanish accreditation architecture aligned to the European Standards and Guidelines (ESG), which require that internal quality assurance actually be used, not merely possessed. Third, a certified model that produces no differentiation and no improvement is itself evidence of a weak model, which the next audit can catch.

But the objection is not fully answered by process, and honesty requires saying so. Certification checks that a model is well-designed; it cannot guarantee the model is well-run every semester in every department. A framework that certifies the apparatus still depends on the quality of the evidence flowing through it. If the student-feedback component is a low-response-rate, end-of-term satisfaction survey, the certified model inherits all of that instrument's non-response bias — it just inherits it inside a nicer frame. A good model with poor inputs is still poor.

That is precisely where the instrument layer re-enters — not as the thing to standardise, but as the thing to improve.

Where Koji fits: better evidence inside a certified model

DOCENTIA tells you to build a defensible evaluation model. It does not tell you how to make the student-feedback input worth trusting. That is the gap Koji for Education is built for.

Instead of a static Likert battery that produces a contaminated mean, Koji runs AI-moderated conversational interviews that probe why a student answered as they did — turning "3 out of 5 on assessment" into a usable account of what specifically was unclear and when. It supports six structured question types (open-ended, scale, single- and multiple-choice, ranking, yes/no), applies automatic thematic analysis to open-text at programme scale, and attaches a quality score to each response so that careless or duplicate answers do not silently distort the evidence. Because the AI moderator is standardised, it removes the human-moderator inconsistency that undermines interview-based evidence — every student gets the same bias-aware probing. And because it supports formative, mid-cycle collection, the feedback can inform teaching this semester, not just next year's certificate renewal. Its action-tracking and programme-level reporting produce exactly the "evidence of use" that a DOCENTIA audit — and the ESG behind it — actually looks for.

Koji frames this carefully: it mitigates the weaknesses of the student-feedback input and surfaces evidence a mean hides. It does not eliminate bias, and no responsible vendor should claim it does. What it does is make the most fragile component of a certified model — what students actually tell you — strong enough to carry weight. (Institutions that also run general user or staff research use the same AI interview engine on the main Koji platform.)

Spain got the architecture right: certify the system, not the survey. The universities that get the most out of that architecture will be the ones whose survey is no longer a survey at all.

Building or renewing a DOCENTIA-certified evaluation model? See how Koji for Education strengthens the student-feedback evidence inside it.

Frequently asked questions

What is the DOCENTIA programme?

DOCENTIA is ANECA's programme, launched in 2007 and revised in 2021, that helps Spanish universities design their own model for evaluating the teaching activity of academic staff. ANECA and Spain's regional quality agencies then audit and certify that model against common criteria. Around 90% of Spanish universities participate.

Does DOCENTIA impose a national course-evaluation survey?

No. This is its defining feature. DOCENTIA certifies the university's own evaluation model — the framework of evidence sources, weighting, and consequences — rather than dictating a single national questionnaire. Quality control sits at the level of the system, not the instrument.

How long is a DOCENTIA certification valid?

An implementation certificate for a university's teaching-evaluation model is valid for a renewable period of five years, so models must be periodically re-audited rather than certified once and left unchanged.

How is DOCENTIA different from Italy's ANVUR/OPIS model?

Italy's ANVUR/OPIS uses a common national student questionnaire, buying comparability across institutions. Spain's DOCENTIA certifies each university's bespoke model, buying local validity. The trade-off is comparability versus contextual fit; neither is free.

How can Koji strengthen a DOCENTIA-certified model?

Koji improves the student-feedback input inside the model. Its AI-moderated conversational interviews, automatic thematic analysis, quality scoring, and action tracking produce richer, more actionable, and more traceable evidence of use than a static Likert survey — the kind of evidence a certification audit and the underlying ESG look for.