Why do you say that? If you look at the graph in the article, that's not true at all.

The comment I was responding to said that "the variations in privilege by gender for the white university-educated middle and upper-middle classes are barely a rounding error compared to the gulf in privilege between them and the working class."

But that's not true. Men got 1.8x as many callbacks as women, and upper class candidates got 2.6x as many callbacks as lower class candidates. Even if the class divide as a somewhat greater impact, the gender divide is not a "rounding error."

The other way to look at it is that the callback rate for upper class women was the same as the average callback rate for lower class applicants, and both rates are much lower than for upper class males.

The whole point of breaking the results down by class is that your statement of "Men got 1.8x as many callbacks as women", while true, is very misleading, because the sex difference only benefitted upper class men. Lower class men got the fewest callbacks of any group.

It's equally wrong to say that it's about class and not gender, because the class difference only benefited upper class men. Upper class women did not get a statistically significant number more callbacks than lower class applicants.

Again, I'm responding to a comment that called the gender discrimination a "rounding error." It clearly is not--there is both gender and class discrimination happening and upper class women aren't better positioned than lower class applicants.

By choosing to group the data in male/female as you do and omitting the class variable, consciously or not, you can reach wrong conclusions about the interpretation of the data. See https://en.wikipedia.org/wiki/Simpson%27s_paradox

I didn't, I grouped it both by class and by gender.