Is this because (by definition) those millennials who've already had children must then have had them at a relatively young age? That might be indicative of a more traditional attitude towards adulthood and, by extension, towards gender roles.
If the oldest of the group had to have children under the median, then only those who had children young have them already.
Looking at millenials with children is a huge selection bias:
> Mothers of newborns are older now than their counterparts were two decades ago. In 1990, teens had a higher share of all births (13%) than did women ages 35 and older (9%). In 2008, the reverse was true — 10% of births were to teens, compared with 14% to women ages 35 and older. Each race and ethnic group had a higher share of mothers of newborns in 2008 who are ages 35 and older, and a lower share who are teens, than in 1990.
http://www.pewsocialtrends.org/2010/05/06/the-new-demography...
But I see there is some overlap (Wikipedia):
Milenials : Researchers and commentators use birth years ranging from the early 1980s to the early 2000s.
GenX : Demographers, historians, and commentators use birth dates ranging from the early 1960s to the early 1980s.
Incidentally I have 2 kids.