As my sibling poster indicates, the SO survey is a convenience sample not a census of coders. If you look at the 2016 survey, the 0-5 years experience cohort is larger (50.3%) than than it is in 2019, so I am not sure we can make those kinds of conclusions about the wider industry.