Could you refer us to the "work" results (studies) that conclude (mathematically) that the variation you refer to, are "not adequeate" to explain the variation

Could you refer us to the "work" results (studies) that conclude (mathematically) that the variation you refer to, are "not adequeate" to explain the variation in "gendered employment"? I'm particularly interested what traits were measured between the groups and how (the hell) did those traits have anything to do with programming.