It cannot possibly be statistically valid for 2 reasons. 1) People report their own salaries and can be lying; 2) It is not a random sample but instead only uses the data from those that respond. That is begging for extreme bias from the most successful of the graduates.