I received a reply from Harvard's Dr. David Reich to the mail I had sent (check previous post comments).
Hi Ashish,
The Z-scores are formal tests for model fit – appropriately calibrated by a block jackknife – and the tail of large Z-scores are simply reflecting the reality of model failure.
What the analyses are showing is that the “Indian Cline” is in fact best modeled as a two-dimensional cloud not a one-dimensional gradient: a mixture of three source populations that cannot be simplified to a mixture of two source populations as we had done in previous modeling for example in Reich et al. Nature 2019.
As such, what our enrichment analyses are showing is simply that the groups of traditionally priestly status tend to be enriched for the Steppe-related source population after controlling for the proportions from the other two.
Yours, David
I will address these points in this analysis post. This will be final evidence to disprove the confident Steppe--->Brahmin causality claim made in the study.