There's obviously going to be a bias of some sort, but it would probably be worth looking at the distribution of the reported salaries to get a better sense of that. (The author may have already looked at the spread of salaries, I haven't read through the whole thing yet).

You could also try comparing with visadoor.com or another source of H1B data for the same companies to try to see where they fit in. There may be a systematic difference between H1B salaries and the larger employee base at these companies, but at least you're not limited by self-reporting.