Methods · Kink Test v2.0
How the Cuffplay Kink Test works
Version 2.0 · launched October 12, 2026 · updated as data comes in
We publish how our test is built and how well it measures, because a self-exploration tool should earn your trust rather than ask for it. For the story behind this version, read How We Rebuilt the Cuffplay Kink Test, With the Data.
At a glance
| Questions | 36 rated statements (12 for a first look), plus 3 optional context questions |
| Answer format | Not at all · A little · Somewhat · Very · Extremely (0–4), plus Not sure yet / Skip |
| Scales | Lead, Yield, Sensation, Structure, Play, Explore, each scored 0–100 |
| Portraits | 7; everyone gets their closest one, labelled clear or leaning |
| Cost / account | Free · no account |
| Status | Not a diagnosis. Reliability for v2 is being measured on live responses (see below). |
What each scale measures
- Lead: How appealing it is to guide the pace and choices in an agreed moment.
- Yield: How appealing it is to hand those choices to someone you trust.
- Sensation: How appealing vivid physical sensations are, such as firm pressure, warm and cool, or a brief sting.
- Structure: How appealing agreed rules, rituals and titles are.
- Play: How appealing role-play, teasing and improvised stories are.
- Explore: How appealing it is to try ideas that are new to you.
Lead and Yield are measured separately. Someone who enjoys both scores high on both; someone with little interest in power scores low on both. A single dominant-to-submissive line can't tell those two people apart.
How the questions are written
- Each question is about one idea only and is scored on one scale only.
- Questions are non-explicit and written in plain English.
- You rate how appealing an idea is, not whether you would do it. The other person can be real or imagined, and you don't need a partner or experience.
- Not sure yet and Skip are recorded as missing answers, never as zero.
How scores are calculated
- For each scale, we average your answers (0–4) and multiply by 25, giving 0–100. Every question counts equally.
- A scale is shown only when you've answered enough of its questions: both questions in the first look, and at least 4 of 6 in the full test.
- Scores are described as low (under 40), some (40–69) or high (70+). Scores within 5 points of those lines are marked "close to the line".
- We don't show percentiles yet. We'll add them once at least 2,000 people have completed the new version, and we'll say who you're being compared with.
How portraits work
| Portrait | When it appears |
|---|---|
| Compass | Lead stands out |
| Current | Yield stands out |
| Mirror | Lead and Yield are both high and close together |
| Flame | Sensation stands out |
| Artisan | Structure stands out |
| Signal | Play stands out |
| Explorer | Explore stands out |
Every result gets its closest portrait: the scale that scores highest, or Mirror when Lead and Yield are both at least 55, within 15 points of each other, and among your top three scales. Each portrait carries an honest strength label: clear when the top scale is at least 70 and at least 15 points above the next (Mirror: Lead and Yield both at least 65), otherwise leaning. We also show your next-strongest scale. If every scale is under 35, your result is Explorer with a note that your interest is quiet across the board right now. A first look after 12 questions shows a provisional portrait. A portrait is shorthand for your answers, not a type you belong to.
Reliability
Version 2.0: being measured. We will publish, for each scale, internal consistency (McDonald's omega) with the number of responses; how closely the 12-question first look matches the full 36; how often each portrait appears; and test–retest reliability once enough people retake the test after 2–4 weeks. Our targets are omega of at least 0.80 per scale, a first-look-to-full correlation of at least 0.85, and test–retest of at least 0.75. If a scale falls short, we'll say so here and replace weak questions in the next version.
Version 1 (v1, until October 12, 2026): we analysed 6,259 completed tests from September 1 to October 12, 2026 (5,044 full, 1,215 quick). Internal consistency on the 5,044 full completions (Cronbach's alpha):
| Power | 0.67 |
| Sensation | 0.49 |
| Role | 0.43 |
| Intensity | 0.29 |
| Connection | 0.30 |
Scored from the same people's answers, the 12-question version matched the 28-question portrait 40.5% of the time, while dimension scores tracked closely (r = 0.62–0.97). Four of twelve portraits covered 86% of full results. These findings shaped version 2: one idea per question, separate Lead and Yield scales, a first look that previews rather than decides, and a closest portrait with its strength stated.
What this test can't tell you
- It isn't a diagnosis or a clinical assessment.
- It describes how appealing ideas feel to you right now, not what you have done or will do.
- Interest is not consent. Any activity needs specific, ongoing agreement, and anyone can stop at any time.
- People who take this test are not a random sample of the population, so comparisons are with other test-takers only.
Your data
Your answers are stored with an anonymous result link and are never sold. You don't need an account. See our privacy policy for how answers are stored.
Version history
- v2.0, October 12, 2026. Six separate scales, 36 single-idea questions, a first look after 12, seven portraits shown only when clear.
- v1, until October 12, 2026. 28 questions, five two-ended dimensions, twelve archetypes. Results taken with v1 still display as they were.
Sources
Clark & Watson (2019), Psychological Assessment 31(12) · McNeish (2018), Psychological Methods 23(3) · Simms et al. (2019), Psychological Assessment 31(4) · Haslam, Holland & Kuppens (2012), Psychological Medicine 42(5) · Holvoet et al. (2017), Journal of Sexual Medicine 14(9) · Weierstall & Giebel (2017), Archives of Sexual Behavior 46.