The data would be more meaningful if it separated out direct hires (Company X directly hires the employee) verses proxy hires (Company X contracts Company Y for

The data would be more meaningful if it separated out direct hires (Company X directly hires the employee) verses proxy hires (Company X contracts Company Y for employees) as generally the later relationship is a lot more exploitive than the former in terms of salaries as its much easier to hide the actually end of day take home salary. Not sure how you would do that.