How we build our questionnaires

Every item was written for Brain Quiz, in French first, from the published literature on the field concerned. We read what exists, identify the registers the work describes, then write sentences about lived situations rather than about symptoms. What is taken from the literature is the registers. What is not taken from it is the sentences.

In practice, a questionnaire has between 24 and 30 items, is read across four or five axes, and takes seven to eight minutes. Items are phrased in the present tense, without judgement and without an implied gender, and eight to fifteen of them are reversed depending on the questionnaire: a high answer there lowers the score. That reversal exists to stop anyone answering on autopilot all the way through.

• What we name in our sources: the publications documenting the field, with their DOI, on every methodology page.
• What we do not claim: that those publications are the origin of our sentences. They document a field, they do not supply the items.
• What we rule out: any rewording of a published item, which would only be a reproduction in disguise.
• What we adapt in translation: the concrete examples, because an everyday example does not carry across word for word from one language to another.

The arithmetic is deliberately simple, because a calculation we could not explain in five lines would have no place here. Each answer is worth 0 to 4 points. Those points are multiplied by the weight of the item, a whole number between 1 and 5. The resulting sum is divided by the maximum reachable, then rescaled to 100. An unanswered item is removed from both the numerator and the denominator.

Each axis is computed separately, with the same method restricted to its own items. It is the breakdown by axis that carries the useful information: two people scoring 60 out of 100 can have opposite profiles depending on which axes carry that score. The overall score is a summary, not a conclusion.

Because a published scale belongs to someone. The best-known instruments are protected by the copyright of their holders, whether an international organisation, a university or a test publisher. Reproducing them requires permission, sometimes a paid licence, and that permission also covers translations. We have not sought it, so we reproduce nothing.

A second reason is translation. A validated instrument is validated in one language and on one population. Translating an English version yourself does not produce a validated French version: it produces a translation whose measurement properties are unknown. Publishing such a translation while presenting it as the original scale would borrow a credibility the translation does not carry.

Not every scale is in the same position. Goldberg and colleagues describe, in a 2006 article, a pool of personality items deliberately placed in the public domain by its authors, precisely so that anyone may use them. That is a documented exception, not the rule. We examined it for our personality questionnaire and did not adopt it, for lack of translations whose accuracy is verified by the project itself.

The reading band the score falls into, and nothing else. Each band carries a level set in advance, which can never weaken as the score rises. There are three levels, and they are the same across every questionnaire in the serious universe.

These messages encourage; they do not dramatise and they conclude nothing. A high score means your answers describe difficulties that come up often in that register. It does not mean you have anything, and the warning attached to every result says so in the same terms.

The free scope is deliberately wide: everything needed to understand your result is available without paying and without creating an account. What is paid for is depth and follow-up over time, never access to the result itself.

No advertising is served on the serious universe. That is a decision of principle: a page dealing with mental health that also sells advertising space sends two contradictory messages. Your answers are neither sold nor rented, and the privacy page sets out what is kept, for how long and who has access.

This list matters as much as the ones above. It describes things we would be technically able to do and have chosen not to.

• We do not compare your score to a reference population, because we have none: not one of our questionnaires is calibrated.
• We do not conclude on your behalf. No result confirms or rules out a disorder, and no wording suggests otherwise.
• We display neither an average rating nor user reviews on the questionnaires: we collect none, and marking up reviews that are not shown would be deceptive.
• We do not require an account to see a score, and we do not place the result behind an email address.
• We do not publish a questionnaire whose three languages are incomplete: an unfinished version stays invisible rather than being filled in with machine translation.
• We do not use the vocabulary of medical practice to describe our own content, because an online questionnaire is not medical practice.