Ethnicity

Veale et al. (2015) considered whether its pooled data could support conclusions about differences in length by ethnicity, and reported that it could not. That is the finding this post covers, in full, and it is worth stating precisely rather than rounding it into something stronger or weaker than what the paper actually says.

What the paper's own limitation says

A meta-analysis is only as good as the studies it pools, and Veale 2015 pooled studies that were not designed, individually or collectively, to answer an ethnicity question with the rigor that question requires. The component studies varied in where they were conducted, who they recruited, how consistently ethnicity was recorded or verified, and which method and state each one measured. Several of the studies contributing to the pool did not report participant ethnicity systematically enough to be broken out and compared with confidence. Put together, those design features mean the paper's authors judged the available data insufficient to draw a conclusion about ethnic differences, and said so, rather than reporting a comparison that the data could not actually support.

Why "insufficient data" is different from "no difference" and different from "a difference exists"

This is the point that gets lost whenever this finding is repeated informally. "The data could not support a conclusion" is not the same claim as "there is no difference" - that would require adequately powered, methodologically matched samples across groups, which the pooled data did not provide. It is also not the same claim as "a difference exists but wasn't captured" - that would require evidence the paper does not present either. The honest position, and the only one the paper's own limitation section licenses, is a null in the specific statistical sense: the question was asked, and the available data were not adequate to answer it either way.

Why this deserves care rather than a quick line

This is a subject where folk claims exceed the evidence by a wide margin, and where an absence of adequate data gets filled in, in casual conversation, with confident claims running in whichever direction someone already believed. Neither direction is what the paper supports. The responsible thing a reference site can do here is describe exactly what was tested, exactly what limitation the authors named, and stop there, rather than speculating about what a better-designed study might eventually find. This site is not the place to relitigate that question with anecdote or with country-level figures pulled from unrelated sources, and it will not do so.

A separate and much shakier practice, worth naming precisely because it gets confused with this one, is the circulation of country-by-country league tables built from a mix of self-reported surveys and differently-measured clinical studies - that comparison has its own problems that are distinct from what this post covers, and conflating the two makes both worse.

Any figure claiming to show a size difference by ethnicity, wherever you encounter it, is not resting on this paper, whatever it says. That includes figures drawn from self-report surveys, from country tables, or from any subjective source - a rating produced by an AI model, a human judge, or a photo-scoring platform is not population data of any kind, adequate or otherwise, and none of them is answering the ethnicity question this post describes. If what you want is a subjective assessment rather than a population claim, that is a different service entirely, and it carries no evidentiary weight on this specific, careful, unresolved question.