I've seen write ups with credible references, but it would take a sort of "meta-analysis" and definitely take into account the groups having their levels tested. Healthy men, unhealthy, regional diets, lifestyle ect.
Sometimes we can be surprised by information and other times it's really just a bunch of puzzle pieces that don't fit.
Sample 1,000 labor workers and 1,000 office workers levels. There may be a difference, but to narrow it down to that disregards huge amounts of variables. Lifestyle, cultural, psychological. Anomalies like the skinny trans office worker has the highest TT levels of the entire group. A huge laborer has low TT compared to the statistical distribution. Comparing SHBG, LH/FSH, diet, activity level, personal history.