I don’t think this has anything to do with node selection, at least, this is not the main reason.
The main reason is that large and long established nodes will by definition have more pieces that can be sourced for repairs.
Therefore ingress and egress bandwidths scale differently: egress repair traffic correlates with node size, but ingress traffic does not; is distributed just like any other traffic in the network. So you’d expect that largest nodes will see way more repair egress than younger ones.