Doctoral & PGR Programme Evaluation: Turning Supervision and Research-Experience Feedback into Accreditation Evidence

Most course-evaluation guidance assumes a taught module with dozens of students. Doctoral education breaks that assumption: cohorts are small, the "curriculum" is a supervised research project, and the relationship that most shapes the student experience — supervision — is one-to-one and identifiable. Yet doctoral education is a full cycle of the European Higher Education Area, and reviewers increasingly expect the same closing-the-loop evidence for it that they expect for taught programmes. This guide shows how to turn doctoral and postgraduate-research (PGR) feedback into accreditation-ready evidence without collapsing under the small-cohort and confidentiality constraints that make doctoral evaluation genuinely different.

The short answer: treat the doctoral experience as a distinct quality object with its own feedback instruments (supervision, research environment, training and development, progression, wellbeing), collect it at the right moments in the candidate journey, protect confidentiality in small cohorts by reporting at the doctoral-school or aggregated level, and document the actions you took. The internal quality-assurance principles of the ESG apply to doctoral education — the challenge is method, not mandate.

Why doctoral education is a distinct quality cycle

Doctoral education "rests on the practice of research" and is fundamentally different from the first and second cycles — a point at the heart of the Salzburg Principles, endorsed by ministers in 2005 alongside the adoption of the Standards and Guidelines for Quality Assurance in the European Higher Education Area (ESG). In 2010 the European University Association's Salzburg II Recommendations pointed to the need to "develop specific systems for quality assurance in doctoral education," and the EUA's position is that the internal quality-assurance principles in the ESG are applicable to doctoral education — the indicators Salzburg II proposes do not contradict them. In 2025, the 20th anniversary of the Salzburg Principles, the EUA Council for Doctoral Education (EUA-CDE) revisited the landscape, underlining how much doctoral QA has professionalised.

The practical consequence for a QA director: an institutional review or accreditation panel will look for evidence that you systematically gather the doctoral candidate voice, that supervision quality is monitored, and that you act on what you learn — the same enhancement logic as ESG Standard 1.9 (on-going monitoring and periodic review), applied to a cycle where the standard survey format fits poorly.

What reviewers actually want to see

Across ESG-aligned systems, panels look for four things in doctoral QA evidence:

1. Systematic collection of the candidate voice across the journey, not a single exit survey.
2. Supervision quality monitoring with a route to act when it falls short.
3. Evidence on the research environment — training, resources, integration into a research community, wellbeing.
4. Closing the loop — documented changes to doctoral-school policy, supervisor development, or resources that followed from feedback.

Mapping accreditation requirements to doctoral feedback and Koji outputs

The table maps common doctoral QA/accreditation expectations to the feedback source that evidences them and to the concrete output a modern evaluation platform can produce.

Accreditation / QA expectation	Doctoral feedback source	Evidence output (e.g. Koji)
Candidate voice gathered systematically (ESG 1.9)	Stage-based check-ins: induction, annual progress, pre-submission, exit	Longitudinal cohort record showing collection at each stage
Supervision quality is monitored and acted on (Salzburg II)	Confidential supervision-experience feedback + probing on specifics	Aggregated supervision themes with action tracking, not raw identifiable comments
Research environment and training reviewed	Feedback on training needs analysis, resources, research culture	Thematic analysis of open responses across the cohort
Wellbeing and integration monitored	Experience feedback on belonging, isolation, support	Standardised, bias-aware prompts applied consistently to every candidate
Enhancement demonstrated (closing the loop)	Follow-up on prior actions	Insight-to-action log linking a change to the feedback that prompted it
Benchmarking against sector	Alignment to sector instruments (e.g. PRES)	Comparable, structured outputs that sit alongside external survey data

The UK's biennial Postgraduate Research Experience Survey (PRES), run by Advance HE, is the best-known sector instrument — the 2022 cycle drew nearly 14,000 PGR students from 62 institutions, and supervision has consistently been the scale respondents rate as most important and are most positive about. PRES is valuable for benchmarking, but its cadence and standardised items are not enough on their own for continuous doctoral-school enhancement; institutions supplement it with their own stage-based collection.

The small-cohort confidentiality problem (be honest about it)

This is where doctoral evaluation is genuinely hard, and where honest method matters more than tool marketing. In a research group of three candidates and one supervisor, "anonymous" feedback is not anonymous — the supervisor can often infer who said what, and candidates know it. That fear suppresses candour precisely on the supervision relationship you most need to monitor.

Practical mitigations you should document in your QA framework:

• Report at an aggregated level — doctoral school, faculty, or discipline — never at a granularity that identifies an individual candidate. Set a minimum threshold below which results are not broken out.
• Separate the confidential channel from the developmental one: candidate feedback that feeds supervisor development should not be attributable in a way that risks the relationship.
• Do not over-quantify small cohorts. A mean score from four responses is noise; qualitative themes gathered consistently are more defensible evidence than a fragile number. This is also a measurement point — see the companion research-methods article on what supervision-experience measures actually capture before you report any doctoral metric.
• Be transparent with candidates about how their feedback is aggregated, stored (GDPR), and used, which itself raises response rates.

A conversational, AI-moderated approach can help here in a specific way: it applies the same standardised, bias-aware prompts to every candidate (reducing interviewer and item-order effects), probes vague answers for actionable specifics, and produces aggregated thematic output rather than a short list of identifiable verbatim comments — while you still control the reporting threshold.

Closing the loop for doctoral schools

Accreditation credit comes from the response, not the collection. A defensible doctoral quality cycle records, for each cycle: what candidates told you (aggregated), what the doctoral school decided, what changed (supervisor training, resource allocation, progression-monitoring policy), and how you told candidates. An insight-to-action log that links each change back to the feedback that prompted it is exactly the artefact a panel wants to see, and it is far easier to maintain in a tool that tracks actions than in a spreadsheet of survey exports.

When a simpler approach is the right call

Honesty serves this audience: you do not always need a dedicated platform.

• A handful of candidates and a strong supervisory college culture. Structured annual-review conversations, minuted and aggregated by the doctoral school, may be sufficient — the risk is consistency and evidence trail, not the format.
• You already run PRES and act on it visibly. If your enhancement loop from PRES is genuinely closed and documented, adding a second instrument only helps if it fills a cadence or qualitative-depth gap.
• Pure benchmarking need. For cross-institution comparison, a sector survey like PRES will always be the reference point; a platform complements rather than replaces it.

A dedicated, conversational platform earns its place when you need consistent stage-based collection across a large or dispersed PGR population, qualitative depth on supervision and research culture, confidentiality-aware aggregation, and an auditable closing-the-loop record.

How Koji supports doctoral QA evidence

Koji for Education uses the same AI interview engine as the main Koji research platform (koji.so) used for user and customer research. For doctoral QA it targets: stage-based, mobile-first collection across the candidate journey; standardised, bias-aware moderation applied identically to every candidate; automatic thematic analysis of supervision and research-environment feedback; aggregated reporting with thresholds that protect small cohorts; and an action-tracking log that evidences enhancement for institutional review. Use it to supplement — not replace — sector benchmarks such as PRES.

Related resources

• Evaluating Doctoral Supervision: What PRES and Supervision-Experience Measures Actually Capture
• Student Feedback as ESG Accreditation Evidence
• Annual Programme Monitoring: Turning Course Evaluation into Accreditation Evidence
• Programme Review & Revalidation
• Institution-Level Evaluation Reporting for Quality Audits
• Writing the Self-Evaluation Report

Frequently asked questions

The questions below address the issues doctoral-school directors and QA leads most often raise when building doctoral evaluation evidence.

Do the ESG apply to doctoral education? The EUA's position is that the internal quality-assurance principles of the ESG are applicable to doctoral education, and that the indicators proposed in the Salzburg II Recommendations do not contradict them. Doctoral education is a distinct cycle that rests on research, so the method differs even though the enhancement logic is the same.

What is PRES and do we need it? The Postgraduate Research Experience Survey is Advance HE's biennial sector survey of PGR students (the 2022 cycle covered nearly 14,000 students across 62 institutions), with supervision as its most valued scale. It is excellent for benchmarking but, given its cadence and standard items, most institutions supplement it with their own stage-based collection for continuous enhancement.

How do we protect confidentiality in small doctoral cohorts? Report only at an aggregated level (doctoral school or discipline) with a minimum-response threshold, separate confidential feedback from developmental feedback, avoid over-quantifying tiny cohorts, and be transparent with candidates about aggregation and storage. Qualitative themes gathered consistently are more defensible than fragile small-N averages.

What evidence do accreditation panels want for doctoral programmes? Systematic collection of the candidate voice across the journey, monitoring of supervision quality with a route to act, evidence on the research environment and wellbeing, and a documented closing-the-loop record showing what changed as a result of feedback.

Can course evaluation software handle doctoral evaluation? Standard Likert-survey tools fit doctoral cohorts poorly because cohorts are small and the experience is relational. A conversational, confidentiality-aware approach that standardises prompts, analyses themes, and aggregates results is a better fit — but it should complement, not replace, sector benchmarks like PRES.

How does Koji fit doctoral QA? Koji supports stage-based collection, standardised bias-aware moderation, automatic thematic analysis, small-cohort-safe aggregated reporting, and action tracking for closing the loop — producing evidence aligned with ESG-based institutional review while you retain control of reporting thresholds.