How the diagnostic was built — and what it does and does not claim.

Each candidate is placed on five spectrums: Motivational Drive (Intrinsic ↔ Extrinsic), Methodological Clarity (Anchored ↔ Drifting), Supervisor Resourcing (Resourced ↔ Abandoned), Research Workflow (Structured ↔ Exploratory), and Life Load (Protected ↔ Overloaded). Each dimension has 14 items in the question bank, plus 3 attention-check items distributed across the test.

These dimensions are not invented in the abstract — they are the axes that the empirical literature on doctoral attrition keeps surfacing as the strongest predictors of who finishes, who finishes well, and what kind of intervention moves people forward. We picked five because the research keeps converging on roughly that count and because five is the right number for a profile that is rich enough to act on but small enough to remember.

A candidate’s position on the five dimensions resolves to one of twelve archetypes, organised into four role groups (Builders, Strategists, Seekers, Resilients). The archetypes are research-backed but original — they are the academy’s synthesis, not borrowed from any existing typology. They are calibrated to resonate hardest for PhD students in business administration, economics, and the social sciences.

The archetype boundaries will be refined after pilot data — once roughly 50–100 candidates have completed the test, we will see which archetypes are over- or under-represented and adjust accordingly. The current version is v1.

When you complete the diagnostic, you receive a personalised report organised into 15 sections across six blocks: Identity (hero + introduction), Profile at a glance, the four-quadrant SWOT (Strengths, Weaknesses, Opportunities, Threats), four PhD-context sections that combine your dimension scores with your archetype (Supervisor, Motivation, Workflow, Life around the PhD), a Path Forward block, a Beyond-the-PhD block, and a wrap-up.

Each archetype contributes its own prose to the SWOT and to the four PhD-context sections, so the report you read is genuinely about your archetype rather than a generic template with your name pasted in. The Visionary report and the Architect report look structurally similar but read very differently because the underlying writing is archetype-specific.

Anonymous viewers can read the first seven sections (identity, profile, full SWOT) before being asked to register; sections 8–15 require a free account. The free account is genuinely free — no credit card, no upsell during sign-up, just an email address and password.

Each archetype profile recommends 3–4 trainings drawn from a central catalog of 19 named programs across five tracks: Foundations of method, Working with your supervisor, Shipping the thesis, Why you’re doing this, and Life around the PhD. The catalog is published openly, and the per-archetype assignments are designed to map directly to that archetype’s typical SWOT-level weaknesses and threats — not to recommend everything to everyone.

Each training has a self-explanatory name (for example, Find Your Method, Run Better Supervisor Meetings, Minimum Viable Thesis, The Long-Haul PhD), a one-line description, and a track classification. Future versions of the academy will deliver the trainings as workshops, coaching, or on-demand content; the catalog itself is finalized and stable.

Every test response is scored against three validity flags before the archetype is reported. Three attention-check items (statements with one obvious correct answer) are distributed pseudo-randomly through the test; failing two or more raises a validity flag. A straight-line check identifies candidates who selected the same response for an implausibly long run; this also raises a flag. A speed check compares total time to a reasonable lower bound; finishing the 73 items in much less time than the lower bound suggests skim-clicking and raises a flag.

When any validity flag is raised, the report still generates but a warning banner appears on the result page advising the candidate to consider re-taking the test in conditions that allow more attention. The flags are stored on the response row and are visible in admin views.

The scoring algorithm is fixed and published. Raw responses (1–7 Likert) are reverse-keyed where appropriate, averaged per dimension, and standardised to a [-1, +1] vector. The vector is then matched to the closest archetype centroid using a published distance formula with explicit tie-breakers and a confidence tier (high / boundary / low).

There is no machine learning, no random component, no personalisation engine on the test. Two candidates with identical responses get identical archetypes — every time. Determinism is a feature: it is how candidates can trust the result.

It is not a clinical instrument. It does not diagnose mental health conditions or learning disabilities. It is not a substitute for therapy, supervision, or institutional support.

It is not a career-decision oracle. The archetype is a description of how you are working today, in the context you are in today. It is not a prediction about whether you should finish, whether you will finish, or what your career should look like.

It is not validated as a psychometric instrument in the formal sense (no published reliability or convergent-validity studies — yet). It is a structured, transparent, research-grounded reflection tool. Treat it as such.

The full literature synthesis behind the dimensional model is internal at this stage and will be published once the v1 pilot data is analysed. The archetype profiles, dimension definitions, scoring algorithm, archetype-specific narrative content, and training catalog are all versioned in the project repository so any change to the instrument is traceable. Two candidates who take the test on the same algorithm version with identical responses receive identical archetypes and identical reports.

Current versions: scoring algorithm v1 (locked 2026-04-26); archetype framework v1 (12 archetypes, 4 role groups); report structure v2 (15 sections); training catalog v1 (19 trainings, 5 tracks). Any future change to any of these will increment the relevant version number and be noted at the top of this page.