I think it is easy to nitpick studies, especially when they are trying to measure something that is this complicated. While the soundness of the paper should be grounded in the methods and the data, it is far too easy for an individual to discount it's value based on a perceived mistake in method. The more complex the methods, the more polarizing the results, the more this is the case. For this particular issue, I think the methods by definition need to be fairly complex.
I am comfortable with my assertion that I will need to see some data before I take the position that women are not paid less than men in a discriminatory fashion. In my opinion, that women are discriminated against is a historical truth. I am not trying to be insulting, but I do find it hard to believe that people disagree on this.