Such formulations seem rather deceitful. If the vast majority of the human genome is invariant between populations, then the average difference between populations will be tiny, while the differences within populations can be large, even if different populations have distinct signatures (drowned out in the average by the overwhelming amount of similarities).
You only need to change a single nucleotide (e.g. an A to a T) in the human genome to get a big change in phenotype.
Sickle-cell disease is one such example.
Many single-nucleotide variations (SNPs)
specific to certain populations are also known, as one would expect. And apparently: