Every DNA report, from any company, is built the same way underneath: your letters are compared against what researchers observed in groups of studied people. That makes a quiet question the most important one in the whole document: who was studied? The answer shapes what any reading can honestly claim about you.
The reference problem
A genetic association is a statistical observation: people carrying this letter, in this studied group, showed this trait more often than people without it. The association travels with the group it was found in. How common the variant is, which other variants tend to ride along with it on the same stretch of chromosome, and how diet or environment interact with it can all differ between populations. The biology is universal; the statistics are local.
The studied world is narrow
In 2016, researchers tallied every participant in the global catalog of genome-wide association studies and published the result under a blunt title: “Genomics is failing on diversity.” At that point 81 percent of participants were of European descent, with African, Latin American and Indigenous groups together making up only a few percent, a skew the authors traced to where and how studies get funded and recruited, not to biology.
The consequences are measurable. A 2019 analysis in Nature Genetics showed that polygenic scores, predictions built by summing many small genetic effects, are systematically less accurate when applied to people whose ancestry differs from the studied group, and warned that using them clinically as-is could widen existing health disparities. A score trained on one population does not automatically describe another.
The same letter, a different story
Single markers show the problem in miniature. The European milk-drinking variant near the LCT gene is a reliable reading in people of European descent, but many people of East African or Middle Eastern ancestry digest milk through entirely different letters that standard chips don’t always include. A C/C result at the European position can therefore belong to a perfectly happy milk drinker. Read without population context, the genotype looks like an answer; read with it, it’s one observation with a stated blind spot.
What honest reporting looks like
None of this means consumer genetics is useless outside Europe. It means a trustworthy report has to do three things. First, say which populations the evidence comes from, next to the reading rather than in a footnote. Second, prefer findings that have been replicated across ancestries, and say so when they haven’t been. Third, admit blind spots: when the marker that matters for your ancestry isn’t in your file, the honest output is “not readable here,” not a confident number borrowed from someone else’s reference group.
This is how Evigenics writes its readings: population-relative, with the comparison group and the limits stated in the result itself. It is less dramatic than a universal-sounding verdict, and considerably more true.
The research world is working on its side of the problem: recruitment is broadening, and methods for cross-ancestry analysis are improving. But the skew accumulated over decades won’t vanish quickly. Until it does, the question to bring to any genetic claim stays the same: studied in whom? For the companion skill, read what your raw file can and cannot tell you.