Tests, tools & research
Sex toy statistics: which numbers exist, and why we quote none of them
You have read that some percentage of adults owns a vibrator. That figure came from somewhere, and almost every article repeating it has dropped the part that would let you judge it.
Search for sex toy statistics and you will find the same handful of percentages, attributed to each other. Follow them back far enough and one of three things happens. The trail ends at an academic survey whose question was narrower than the claim built on it. It ends at a retailer or a market research firm whose figure describes its own customers and whose method is not published. Or it ends nowhere at all, at an article citing an article citing an article.
That is not a reason to be cynical about the whole subject. Real data exists, and some of it is good. It is a reason to know which family a number came from before you let it change your mind, because the three families answer completely different questions and are routinely quoted as though they answered the same one.
The four families of numbers, and what each one can honestly tell you
- Probability-sample population surveys. A sample drawn so that every adult in a defined population has a known chance of being selected, asked a fixed question. This is the only family that can support a sentence beginning "x per cent of adults". There are few of them, they are expensive, they are usually national, and they are usually several years old by the time you read about them.
- Convenience-sample surveys. A questionnaire answered by whoever saw it: a magazine readership, a mailing list, a panel, a university’s undergraduates. Useful for comparing groups inside the sample and for generating hypotheses. It cannot tell you what a national population does, however many thousands answered, because the people who answer a survey about sex toys are not a random slice of anybody.
- Commercial sales data. What a retailer or a platform sold, or what a market research firm estimates an industry turned over. It describes purchasing, not use, and it describes that retailer’s customers. A category that grew in one shop’s sales may simply have moved there from another shop.
- Search and traffic data. Query volumes, trend lines, keyword exports. It measures what people typed, which is a genuine signal about curiosity and language, and no signal at all about ownership or behavior. It is also the family most often dressed up as the other three.
How to check a sex toy statistic in four moves
Find the original, not the citation
Click through until the links stop. If the trail ends at another article rather than at a study, a dataset or a methodology page, you have found a rumor with a decimal point. That happens more often than not in this subject.
Ask who was asked
A national probability sample, a panel, a shop’s own customers and a magazine’s readers support four different sentences. If the population is not stated, the number cannot be generalized to anybody, and nobody can tell you how wrong it is.
Read the actual question
"Have you ever used a vibrator", "do you currently own one", "have you used one in the past year" and "have you used one with a partner" produce wildly different figures. Articles routinely swap one for another while keeping the number. Where the wording is not published, treat the figure as unusable rather than approximate.
Check the date and the place
This market changed shape over the last two decades, and attitudes differ enormously between countries. A figure with no year and no country attached is not a small gap in the citation, it is most of the information missing.
The specific mistakes that produce most of the bad numbers
Four errors account for nearly everything wrong with the statistics in this market, and all four are easy to spot once you know their shapes.
Sales presented as prevalence. A retailer reports that a category doubled. An article reports that use of that category doubled. Those are not the same claim, and the first one cannot support the second even in principle, because one shop’s growth may be another shop’s loss.
Ever-use presented as current use. "Has used at least once in their life" is the most generous question a survey can ask, and the resulting figure is always the largest one available. It is therefore the figure that travels, usually rewritten as what people do rather than what they have ever done.
A number given a false precision. A figure reported as a range with a confidence interval in the original becomes a single decimal in the retelling. The decimal is not more accurate. It is the same estimate with its uncertainty deleted.
A score invented to fill a column. Not a survey problem but a reviewing one, and the version we are most often asked for: a table of toys rated out of ten on power, noise or rumbliness, by people with no instrument. Those numbers are not measurements of anything. We do not publish them, and we say so on every page where the gap is visible.
Questions people ask
- So how many people actually own a sex toy?
- Nobody can tell you that with confidence for a global population, and we are not going to pretend otherwise. Good national surveys exist for a few countries and a few years, and they disagree with each other partly because they asked different questions. Any single worldwide percentage you see is an extrapolation whose assumptions are not shown.
- Are the figures in news articles just made up?
- Usually not invented outright. Usually they are real figures that have lost their population, their question and their date somewhere along the chain of citation, which leaves them technically true of something and useless for the claim they are attached to.
- Why does market research get quoted so much then?
- Because it is written to be quoted. It arrives as a press release with a headline number and a growth rate, it is free to repeat, and the method is behind a paywall most people citing it never cross. That makes it convenient, not reliable.
- What would make a statistic good enough for this site?
- A stated population, a stated sampling method, the full question wording, the field dates, the sample size, and a way for a reader to reach the original. If a figure has those six things we will cite it and link it. If it has five, we will say which one is missing.
- Do you at least have your own traffic data?
- We have search volume estimates, which is how the pages on this site were prioritized. We treat them as a measure of what people type, never as a measure of what people own or do, and we do not dress them up as behavioral data.