Advice from my HR friend on this kind of campus fresh grad survey is to take a 15-20% discount on the value and that is the real market.

Reason such surveys are higher is 1) sample bias because only people who feel good about themselves will bother to take part and 2) no independent verification, anyone can put a higher number and minimal checks are done 3) ignore NS effect and higher public sector starting as a result of front loading

The schools themselves are not stupid and know that such raw data should be further adjusted to reflect underlying conditions, but they knowingly ignore doing that as reporting higher starting salaries does well for the school's branding.