Misleading title. It does not say whether or not the salaries were normalized based on the type of job inside each occupation (hours, position, etc.). Almost surely they have not been.
There is no way to read that graph, or the data behind it, that indicates things should stay the way they are.
Of course, but it's a different problem with a different solution. Isn't that important to know?
You can say there's no way to read it as indicating things should stay the way they are. I believe azth's point is that there's no way to read it as saying anything coherent at all.
(EDIT: Typo.)
If pay is heavily influenced by seniority, and a profession used to be sexist in its hiring practices but no longer is, then it follows that there'll be a significant gender pay gap until the older - mostly male - members of a profession die out.
Of course, there are many ways to read it negatively too. It's so devoid of information that it's just meaningless.
The graph alone doesn't tell us a whole lot, other than women are almost always compensated less than men. There seems to be quite a bit of variation across job functions, which might be unexpected. Understanding why that variation exists might be useful in addressing the perceived problem, and avoiding corporate policies that accidentally make things worse.
For example, if the bulk of pay gap is years of experience, it would be foolhardy to increase starting salary for women; you'd be giving equally qualified men a competitive advantage in hiring they don't need. Similarly, the situation could be understated -- if I'm reading the methodology correctly, CPS data doesn't include independent contractors, since they buy health insurance separately, I expect a number of men forgo insurance entirely and pocket the difference, a gamble few women of child bearing age dare take. Corporate policy here could try to prohibit independent contractors, but that would substantially reverse the trend. Better, IMO, would be single payer health care, or at least an end to the tax subsidy we give employer paid health care.
More dimensions will give us a more nuanced view of reality, and a better grasp on the degree to which the gap is systemic vs personal.
http://www.bls.gov/cps/cpsaat39.htm
I somewhat trust the Bureau of Labor Statistics.
Just as an example of how the numbers might not mean what they appear to: "lawyers" is near the middle of the chart, at around the 82% mark, while "legal occupations" is way at the bottom, around 57%. Does "legal occupations" include lawyers, or is it everything else in the law field aside from lawyers? If lawyers are included, then there's an obvious explanation for the large gap in legal occupations as a whole (namely, more male lawyers, and lawyers being higher-paid than assistants etc., dragging up the median compensation of males in the legal field); if they're not included, then there's a huge and unexplained gender discrepancy in pay among legal assistants etc.
These numbers have a very different meaning if they don't account for seniority, hours worked, or level of responsibility than if they do.
I don't undestand your point. Those are exactly the mechanisms most likely responsible for the pay gap. No employer would ever pay double wage for an y chromosome, but it's not unlikely that preconceived gender roles influence decisions about who gets promoted, who stays home with the kids and stuff like that.
By way of analogy, one could look at rainfall averages over the whole of the US and conclude that some place is/is not in a drought. You could do this by ignoring the obvious confound that different places get different amounts of rainfall. For some places, it would be true. For others, not.
Most people unhappy about the 'pay gap' are indeed implying that employers are paying double (or 21% more) wage for a y chromosome.
If the wage gap rather has to do with decisions employees make themselves (such as who stays home with the kids and thus works fewer hours / has less experience / doesn't get promoted) - then it's not really the problem of employers to contend with.