Yes, which is why the per capita number is not the proper way to measure this.
For example, imagine a company with 7 male employees and 3 female employees. That company has 9 engineers and one administrative assistant. All the engineers are underpaid. The male engineers are underpaid by $1000 and the female engineers are underpaid by $1200. The administrative assistant is a woman who is fairly compensated. That means the men are underpaid by $7000 total and the women are underpaid by $2400 total. Men make up 70% of the employee workforce and are roughly 75% of the underpaid total. More men are underpaid than women. The per capita underpayment of men is $200 more than the underpayment of the average woman. There are plenty of ways to frame this data that make it look like men are the biggest victims and yet woman engineers are all payed $200 less than their male counterparts.
I understand how that works but we're going to need to draw a line somewhere. Going by job, rank, and salary alone could ignore hours worked per week (statistically men tend to do more in the same role), bonuses for high quality work (who knows how this plays out?), etc. I could choose whatever filters I want to frame the data how I want it to, it's one of the biggest tricks of the "women are paid X less than men" camp.
You asked a direct question and I gave you an answer why that question isn't the right one to ask. You are now shifting the debate to be about something completely different.
As I have said in other comments in this thread, there is simply not enough information in this article to say definitely whether there is any gender discrimination in pay at Google and if there is, what gender benefits from that discrimination. Too many people in these comments are approaching this article with biases of their own (on both sides).
There is never going to be enough information and no matter how you slice the data you have it will always be arbitrary and biased. This whole debate is silly.
Right, which is why we need more information before we start throwing out notions that a particular group are systematically victimized. Hopefully studies like these will advance the conversation beyond "well on average $group1 makes more than $group2 ergo discrimination".
For example, imagine a company with 7 male employees and 3 female employees. That company has 9 engineers and one administrative assistant. All the engineers are underpaid. The male engineers are underpaid by $1000 and the female engineers are underpaid by $1200. The administrative assistant is a woman who is fairly compensated. That means the men are underpaid by $7000 total and the women are underpaid by $2400 total. Men make up 70% of the employee workforce and are roughly 75% of the underpaid total. More men are underpaid than women. The per capita underpayment of men is $200 more than the underpayment of the average woman. There are plenty of ways to frame this data that make it look like men are the biggest victims and yet woman engineers are all payed $200 less than their male counterparts.