Who gets measured

No - the professionally measured studies behind the standard averages mostly recruited men already attending a clinic for something else, not a random draw from the population. Their means describe that population first. That fact about where the data came from is worth carrying alongside every figure Veale et al. (2015) reports.

Why a clinic, and not a survey

A clinician can press a ruler to bone and read a tape correctly. A stranger filling in a web form cannot be checked at all. So the studies that measured properly - rather than asking people to report a number - almost all did it inside a clinical setting, because that is where a trained measurer and a willing, undressed participant can both be arranged at once.

The trade-off is recruitment. A general population sample is drawn to represent everyone. A clinic sample is drawn from whoever walked into that particular building for whatever reason brought them there, which is a narrower and non-random slice.

What that does to the sample

Two effects follow directly from recruiting through clinics, and both are about who shows up rather than about anatomy.

Age skews toward adulthood in clinical care, not across the whole lifespan. Men attending urology and sexual-health services cluster in the age ranges that use those services most, which is not evenly spread from eighteen to eighty.

The reason for attending is unrelated to the measurement, but the population attending for reasons is not the general population. People who need any kind of clinical care differ systematically, on average, from people who do not, in ways that have nothing to do with size and everything to do with why a clinic sample is a clinic sample rather than a population sample.

Neither point is a claim about direction or magnitude here - the literature does not settle that, and this site does not invent a number where none has been checked. The honest statement is structural: a sample recruited this way is not interchangeable with a sample recruited by knocking on random doors.

How studies try to correct for it

Mostly, they do not correct for it - they disclose it. A typical example is Habous et al. (2015), The Journal of Sexual Medicine: a retrospective cohort of 778 men attending urological outpatient clinics in Saudi Arabia, stated as such in its methods. Veale et al. (2015), BJU International, filtered on how men were measured and why they attended, excluding samples with a complaint of small size or erectile dysfunction, and named its own limitation plainly: "relatively few erect measurements were conducted in a clinical setting".

Some individual studies restrict further, drawing only from urology outpatients rather than sexual-health clinics specifically, which narrows the recruitment channel without fixing the underlying issue. Few, if any, of the pooled studies recruited through a method designed to be population-representative, because that is a much harder and more expensive study to run than measuring people who are already present for another reason.

The honest way to read the resulting mean is as the mean of men who attend this kind of clinic, with the population figure sitting somewhere nearby and unmeasured. What a meta-analysis actually pools and inherits is worth reading with that caveat attached, because a bigger sample size does not fix a sampling method - it only makes the flawed estimate more precise.

This is a separate problem from the two it gets confused with

Clinic recruitment is not the same issue as who volunteers once asked - that selection effect is its own topic and applies even inside a clinic sample. A review by King (2021), Journal of Sex & Marital Therapy, put researcher-measured erect length between 5.1 and 5.5 inches and judged it "probably toward the lower end of this range" once volunteer bias is allowed for. It is also not the same as which men a study excludes after recruitment for reasons like an existing condition affecting the measurement - exclusion criteria are a separate filter, applied downstream of who walked in the door. Clinic recruitment, volunteering, and exclusion stack, one on top of the next, and each one trims the sample a little further from a true population draw.

None of this makes the Veale figures useless. It makes them what they are: careful, professionally taken measurements of a specific and disclosed population, not a census. That is a genuinely different question from what a rating service like Rate Cock is answering, which is a subjective judgement about an individual photograph rather than a population estimate at all, closer in kind to browsing a public distribution of scores than to reading a study - and different again from what an image model can and cannot see in a picture, which is proportion rather than a distance in centimetres. If you want a human opinion rather than a population statistic, that is a different service entirely.

Read the mean, read the caveat, and do not treat either one as more precise than the sampling method behind it allows.