I am wondering the same thing. It's interesting that they didn't report at all on the median and only the average. Also find the timing interesting, as I can't help but suspect that both the justification and incentive for self-reporting a higher number of friends materially changed for some people in the early days of social media. They didn't seem to acknowledge this at all.
The model they built that draws a causal relationship between graph density and polarization is interesting, but these gaps leave me skeptical.
There are just so many other reasons I can see for polarization.
1. Late-stage of civilizational monetary cycle (bretton woods - petrodollar) -> historically leads to polarization
2. Dramatic increase in access to information / wide range of things to know and care about
3. Attention economy (novel upsetting news is best at getting attention, not nuance, not truth)
4. Habits of instant gratification diminishes patience for nuance
5. Maybe foreign state interference/bias towards polarization to destabilize rivals?
6. Several more maybe??
So I buy the graph density correlation, and I'm curious about contributing to causation, but I'm extremely skeptical that it's the primary or sole cause.