[maus42]

While the article is interesting, I think some of the more curious part of HN crowd might find it also worthwhile to look at the actual dataset available here: http://www.gss.norc.org/ . Looking at the codebook, some of the relevant variables for analysis of this kind would be PARTNERS, SEXFREQ, NUMWOMEN, NUMMEN, PARTNRS5 (starting at p.2586).

The codebook includes nice summary tables that might alone answer some the questions you are thinking about (like here https://news.ycombinator.com/item?id=13823544 ; I myself didn't download the data yet but could write that comment).