This is actually quite flawed. The correlations between starting salary and continuing salary (at the same company) is close for large and meaningful data partitions. You can glean some of the indights (if you don't have firsthand experience, if you do...this is obvious) from something like this study. The TLDR is that adverse starting points are meaningful for explaining career earnings anomoly.

http://www.nber.org/digest/nov06/w12159.html