We separate two questions that are easy to blur. Does the result faithfully summarise the answers you chose? And does the quiz measure the preference named by its labels? A result can do the first without the second being established.

Accuracy stops at your answer pattern. It does not show that our questions cover every version of an interest, that everybody reads them in the same way or that the resulting dimensions exist as fixed traits outside the quiz.

We have not psychometrically validated these quizzes. We do not publish validation samples, factor analyses, comparisons with external measures, measurement-error bands, diagnostic-accuracy figures or clinical cut-offs.

The joint Standards for Educational and Psychological Testing treat validity as evidence for a particular interpretation and use of scores. Evidence for using a result as a reflection prompt would not also justify using it to classify identity, predict behaviour or make a clinical decision.

A fuller validity case would ask whether the questions represent a clearly defined preference, whether people understand them as intended, whether response data support the proposed dimensions and whether relationships with other measures fit the interpretation. The standards' account of validity evidence also asks researchers to consider rival explanations for scores.

Our results organise what you told us through the choices available in one quiz. That is the whole claim.

Getting the same result twice is not, by itself, proof that a quiz is reliable. Reliability research asks how consistent or precise scores are when people repeat a measurement under defined conditions.

The testing standards' discussion of reliability and precision makes the conditions of repetition part of the claim. A stable trait, a temporary mood and an interest tied to a particular situation do not call for the same expectation. If you picture a different activity or discover a new variation between attempts, a changed answer may tell you something real.

We do not publish reliability coefficients or standard errors of measurement for our quizzes. We would not treat a small score difference between attempts as precise evidence that you changed.

Every answer starts with an interpretation. If one person reads "public" as a private party and another reads it as a busy street, they are responding to different imagined situations even if they choose the same option.

The UK Government Analysis Function's questionnaire design guidance explains that vague wording and question order can affect responses. It describes cognitive interviewing as a way to examine how people understand a question and reach an answer. We have not published that evidence for every question in our quizzes.

Words such as "rough", "formal", "control" and "intense" have no universal unit. A more specific example narrows the possible readings but may miss the version you have in mind. We read your answer as a response to the words on the page, not as a context-free quantity.

We record what you choose at one point in time. Your mood, recent experiences, privacy, familiarity with the vocabulary and the scene you imagine can affect that choice. You may answer about fantasy on one attempt and wanted experience on another unless the question fixes the frame.

A methodological review of sexual-behaviour surveys describes error connected with sampling, participation, questionnaire language, administration, recall and socially desirable responding. It also distinguishes stability over repeated answers from checks that related answers fit together. Neither makes self-report an objective observation.

We do not treat every changed answer as an error. Your preferences can move, a word can start to fit differently and the context you care about can change.

Each quiz has a finite set of questions and dimensions. We cannot name every role, sensation, intensity, emotional tone or private association that might make an interest appealing. A low result can mean that our examples missed the version you like.

We cannot infer motive from a label. Two people might choose the same impact-play answer for quite different reasons, and one person may want different roles in different activities.

We use a short result profile to make the answers easier to navigate. Your strongest individual responses usually contain more detail than the profile name.

Comparing groups requires evidence that questions and dimensions carry comparable meanings for those groups. Researchers call this measurement invariance. Without it, a score difference can partly reflect wording, interpretation or how particular questions behave.

A 2023 study of three sexual-attitude scales across sexual-orientation groups found strict invariance for one scale, only weak invariance for two and differential item functioning in two items from one scale. That study does not tell us that any question in our quizzes is biased. It shows why identical wording alone does not establish comparable meaning.

We have not published invariance studies across gender, sexuality, age, culture, language, disability or experience level. We do not use our results to rank demographic groups or claim that one group is more interested in a kink than another.

The testing standards define a percentile as a person's standing within a specified score distribution. It is not the percentage of a trait they possess. An 80th percentile result, if we had one, would not mean "80% kinky" or "80% submissive".

We have not established a representative reference population or published norms. The standards' guidance on norm-referenced scores calls for a sound, representative sample of the intended population and notes that norms can lose relevance over time. A large group of self-selected website visitors would not satisfy that requirement by itself.

When we use words such as low, moderate, strong, leading or curious, they describe a result within one quiz. They are not population ranks and do not have one standard meaning across every quiz.

When you skip a question, we do not read the skip as disinterest. We simply have no answer for that part of the quiz, so the result rests on less information.

If several questions do not fit, completing them anyway may create a neater result but not a truer one. The more useful clue may be that our wording, examples or categories did not describe your experience.

When a result surprises you, look first at the statements you answered most strongly and the questions you skipped. Ask what you pictured, which detail appealed and whether you were answering about fantasy, curiosity or something you would like to try.

Treat the result as a draft description. Keep the parts that help you name an interest and discard the parts that do not fit. A profile label should never overrule your own account of yourself.

A result cannot decide whether you should adopt a label, try an activity, disclose an interest or make it part of a relationship. It cannot assess readiness, skill, compatibility, mental health, capacity or future behaviour.

If you retake a quiz, compare the answers and the situations you had in mind. We would not read a tiny score movement as measured change, and a different result can be useful precisely because it prompts you to notice what changed.