I went looking, and found that "unincorporated self-employed" is exactly as you say, and is 2/3 male. Farm employees are roughly 3/4 male. Soldiers (~90% male) are also excluded.

"Nonfarm payrolls" was never meant to be a gender-balanced stat, it just excludes the hardest to track jobs. Which is fine for comparison over time, but makes 50% a meaningless metric.