We may need to be more explicit about the claim under discussion. I intended to say that if you partitioned students by race and economic status, ran a regression, then also added in some gene markers and ran another regression, you would find the race coefficient decreased more than the economic status coefficient.

The data you are providing suggests that genes become a better way to differentiate students at higher economic status from each other, which does not appear to disagree with my claim.

I will also note that the mechanism behind your data- increased uniformity of parenting styles- might also be strongly noticeable when looking at race instead of economic status.