The assumption that conviction rates by a demographic match actual criminal acts by that demographic is the Occam's Razor outcome, and thus since it by definition can't be proven is either way automatically is the one assumed to be most likely true without evidence indicating otherwise. There is good evidence to indicate black males accused of crimes are convicted more often then others accused of crimes but I've never seen anything credible to indicate the rate is so high as to close that gap, and Occam's Razor rules in absence of that, in this case that black males commit more violent crimes then other racial/gender demographic groups. Discard that and you must essentially discard all other criminal demographics, including that men re more likely to commit rape then women, because the same logic applies.

Most rapes are committed by men is unprovable, or at least not practically provable currently. But it is a logical inference of most convicted rapists being males and of most claims of rape accusing males of perpetuating them. It is not practically provable, but many things aren't and yet must still be discussed. I can't prove that anyone will die of old age in the next year, but I can strongly support that conclusion.