Take the difference in pay for black and white men. That difference narrows to just 4% after adjusting for years of schooling and it reduces to zero when you factor in test scores on the Armed Forces Qualification Test (AFQT), which is basically an intelligence test. In other words, adjusting for just two factors that cause people to be different, the pay gap between black and white men disappears entirely. Among women, the gap actually reverses after adjusting for education and AFQT scores. That is, black women get paid more than white women.
Among Hispanic and white men, the pay gap narrows to 8% after adjusting for years of schooling and disappears altogether with the addition of AFQT scores. Among the women these two variables cause the pay gap to reverse. As in the case of race, Hispanic women are actually paid somewhat more than white women.
What about men as a group versus women as a group? In addition to years of schooling and test scores, men and women differ in the amount of work they do. Men are more likely to work full-time; and among full time workers, men work 8%-10% more hours than women. Also, men typically accumulate more continuous work experience and therefore acquire higher productivity in the labor market. In fact, the gender gap shrinks to between 8% and 0% when adjustments are made for work experience, career breaks and part-time work. As professor O'Neill writes:
The most important source of the gender wage gap is that women assume greater responsibility for childrearing than men. That influences women's extent and continuity of work, which affects women's skills and therefore wages. In addition, women often seek flexible work schedules, less stressful work environments, and other conditions compatible with meeting the demands of family responsibilities. Those come at a price — namely, lower wages.
And here's one more telling statistic from Professor O'Neill. Among middle-aged adults who have never been married and who have no children, women actually earn 8 percent more than men, although the pay gap is not statistically significant.