How We Build Our Tests

Every test on PsyMed is built to a written method, and this page is that method. It covers where each test comes from, how it is scored, what we do about safety, what we record when you finish one, and the things a screening test cannot tell you no matter how carefully it is built. People make real decisions after taking these tests — whether to raise something with a doctor, whether to take a feeling seriously — so the process behind them should be readable, not assumed.

Two things are true at once here: the questionnaires we reproduce are serious clinical instruments with decades of research behind them, and a questionnaire you fill in alone on a website is not a diagnosis. Both facts shape everything below.

What we publish, and what we will not publish

We publish self-administered screening tests for mental health, personality, neurodivergence and relationship patterns, together with the explanatory writing that surrounds them. We do not review apps, rate therapists, run a practitioner directory, or accept payment to feature anything. There is no commercial ranking to influence, because there is nothing on this site that a company could buy a place in.

We will not publish a test that claims to diagnose. We will not publish one that gives a number without explaining what the number means, and we will not publish one whose result could be read as medical advice. Where a topic is real but the construct behind it is informal — rejection sensitive dysphoria is the clearest example — we say so on the page itself rather than letting the format imply a clinical status the idea has not earned.

Where our tests come from

Our tests fall into two groups, and the difference matters enough that it is stated on each page rather than left for you to work out.

Faithful reproductions of published instruments

Some tests are established clinical questionnaires reproduced as their authors wrote them. These include the Insomnia Severity Index, the PHQ-9 and GAD-7, the EAT-26, the Zung self-rating scales, the Edinburgh Postnatal Depression Scale, the Rosenberg Self-Esteem Scale, and the Adverse Childhood Experiences questionnaire.

Reproducing one faithfully means three specific commitments. The instrument keeps its prescribed length — the Insomnia Severity Index is seven items, so it is seven items here, never padded out to look more thorough. Each item keeps its own response anchors, including where those differ from item to item within the same questionnaire. And the score bands are the author published cutoffs, not boundaries we invented to make the output look tidier. A total that does not map onto the published scoring is not the instrument any more, whatever it is labelled.

Where an instrument has a named author or owner, the page credits them and the page says plainly that reproduction is for education and self-screening.

House-built scales, labelled as such

For some experiences there is no public, validated questionnaire to reproduce. In those cases we write our own items, informed by the clinical literature and the diagnostic criteria where criteria exist, and we label the result for exactly what it is: an original educational scale, not a validated instrument. A house-built scale can still be useful for recognising a pattern and for deciding whether to raise something with a professional. It cannot be compared against published norms, and we do not pretend otherwise.

The steps every test goes through

Before a test is published, it passes the same sequence of checks. Each one exists because it is a way tests go wrong.

1. Source the items

Items are either transcribed from the published instrument or written from the clinical description of the construct. Nothing is paraphrased from another site. Where an item is reproduced, it is reproduced verbatim rather than reworded, because rewording a validated item quietly changes what it measures.

2. Fix the scoring before anything else

The response anchors, their point values and the maximum attainable score are settled and checked arithmetically before the page is written. The bands then have to cover the whole range with no gap and no overlap, so that every possible total returns exactly one result. A test that can produce a score with no matching band, or two, is broken in a way most readers would never notice.

3. Decide the safety routing for this specific topic

Crisis information is matched to the subject and to the severity band, not pasted onto everything. A suicide and crisis line belongs on a depression or insomnia screen at the higher bands; it does not belong on a test about dyslexia, where it would alarm someone who came to find out why reading is hard. Abuse screens carry a domestic violence line and carry it above the questions, where someone can reach it without finishing. Around forty of our tests carry crisis routing for this reason, and the rest deliberately do not.

4. Check every claim against the source

Statistics, criteria and citations are checked against the literature rather than against secondary summaries. Where we cite a study we verify that the paper exists, that the authors, journal, volume and year are right, and that its identifier resolves. References that cannot be verified are removed rather than softened.

5. Verify the links and the surrounding claims

Internal links are resolved against the live site before publication, so a related test we point to is one that exists and is published. This sounds trivial; it is the single most common defect we catch, and a broken recommendation wastes the moment when someone is most willing to keep going.

6. Publish with dates and keep it maintained

Pages carry a publication and a last-updated date. When guidance on an instrument changes, when a link rots, or when we get something wrong, we change the page and move the date rather than leaving a stale version standing.

What decides whether a test ships

Different tests raise different questions, but the ones behind every publishing decision are the same five.

• Is the instrument intact? For a reproduction: right number of items, right anchors on every item, the author published cutoffs. For a house-built scale: is it honestly labelled as one?
• Does the arithmetic hold? The stored scoring is re-read after the test is created and the maximum recalculated from the database, because a scale that looks right in a draft and is wrong in the database is worse than no scale at all.
• Is the safety response right for this topic? The correct crisis routing, in the correct bands, and nowhere it does not belong.
• Is every factual claim traceable? Citations resolve, criteria match the manual, and nothing rests on a statistic we cannot source.
• Does it stay inside its limits? No diagnosis, no treatment instruction, no dosing, and no implication that a total settles a question only a clinician can settle.

Safety rules we apply to every test

These are fixed, and they override editorial preference.

• Crisis lines are matched to the topic and the score. In the United States that is 988, the Suicide and Crisis Lifeline. Readers elsewhere are pointed to an international directory rather than a number that will not connect.
• Help comes before the questions when the topic demands it. On abuse screens the hotline sits above the test, not only in the results.
• No medication guidance of any kind. No drug names presented as options, no dosing, no comparisons. Where treatment is discussed we name the approach and point to a professional.
• No directives about leaving a relationship. On abuse screens we route to safety planning with an advocate, because leaving is the most dangerous moment in an abusive relationship and a website is not positioned to time it.
• Low scores are not dismissals. On screens where a low total can still coexist with a real problem, the lowest band says so instead of telling someone nothing is wrong.

What a screening test cannot do

A screen takes a structured measurement of what you report about yourself, at one moment, in your own words. That is genuinely useful: it turns a vague feeling into something specific enough to describe to a professional, and it can tell you whether that conversation is worth having. It is also the whole of what it does.

A screen cannot diagnose a condition, rule one out, distinguish between conditions that present similarly, or detect what you have not noticed about yourself. It cannot see the context a clinician would ask about, and it cannot account for the ordinary reasons a score might be high this fortnight and not the next. Where a result surprises you in either direction, the result is not the authority.

How the site is funded, and what that does not buy

PsyMed is funded by advertising. Ads are served by Google and are selected automatically; we do not sell placements inside our content, and an advertiser cannot buy a test, a score band, a recommendation or a position in any list. We run no affiliate links on our tests, so there is no product we earn more from when you click it. The separation is structural rather than a promise: there is no commercial inventory in the editorial layer to sell.

What we record when you take a test

Being accurate about this matters more than sounding reassuring, so here is what actually happens. When you finish a test, the completion is recorded: which test, your score, the time it took, and the IP address the request came from. That record is what lets us count how often each test is finished, which is how the “most taken” ordering on our test index is produced.

If you are not signed in, no name, email or phone number is attached to that record — the overwhelming majority of records hold none of those. If you do have an account, or you choose to enter your details to have a result emailed, that information is stored with the record because that is the only way the feature can work.

We do not sell this data, and we do not use it to build profiles of individuals. Your individual answers to specific items are not what we analyse; the score and the completion are. If you would rather leave no record at all, taking a test will always create one, so the honest answer is that we cannot offer that today. Our privacy policy is the formal statement of how data is handled across the site.

Tell us when we get it wrong

If a score band looks wrong, a citation does not check out, an instrument is misattributed, or a page contradicts your own experience of a condition in a way that seems like an error rather than a difference, write to info@psymed.info and say which page and what is wrong. Corrections to substance are made on the page and the updated date is moved, so you can see that it changed. We would rather be corrected than consistent.

Frequently Asked Questions

Are PsyMed tests scientifically validated?

Some are and some are not, and every page says which. Where we reproduce a published instrument — the Insomnia Severity Index, PHQ-9, GAD-7, EAT-26, the Edinburgh Postnatal Depression Scale and others — you are taking that validated questionnaire at its prescribed length with its author published scoring, and the validation belongs to the original instrument. Where no public validated questionnaire exists for a topic, we write our own items and label the result an original educational scale, which means it has not been through validation and cannot be compared against published norms. The distinction is on the page, not hidden in a disclaimer.

Can an online test diagnose a mental health condition?

No, and that is true of every online test, not only ours. A diagnosis requires a qualified professional who can take a history, see context a questionnaire cannot capture, distinguish between conditions that look similar on paper, and rule out physical causes. What a screening test can do is measure what you report about yourself in a structured way, give it a name, and tell you whether it is worth raising with someone. That is a genuinely useful first step, and it is the only step a website can take.

Are my test results private?

Mostly, and here is the precise version. Each completion is recorded with the test, the score, the duration and the IP address the request came from. If you are not signed in, no name or email is attached to it. We do not sell the data and do not profile individuals with it; the records exist to count completions and to order our most-taken list. We cannot currently offer a way to take a test without creating a record, so if that matters to you, that is the honest position rather than a reassurance.

Who writes and reviews the tests?

Tests and the writing around them are produced and reviewed by the PsyMed editorial team against the method on this page. We do not claim individual clinical authorship we cannot stand behind, and we do not put a clinician name on a page to borrow authority. Where a test is a reproduction, the instrument author is credited by name on that page, because the credit belongs to them and not to us.

A note on scope. Everything on PsyMed is educational. Our tests are screening and self-reflection tools, not diagnoses, and nothing here replaces assessment or treatment by a qualified professional. They are not crisis services. If you are in crisis, call or text 988 (Suicide & Crisis Lifeline, US, 24/7), contact your local emergency number, or find an international helpline at findahelpline.com.