[madboston]

Also wouldn't the dataset being from different number of people affect this? During 2021 probably much more people with higher pay self reported their pay on the website and 2022 just didn't see that much contribution?

[comte7092]

The number shouldn’t really matter as long as the 2022 number was of sufficient size.

The primary concern is whether or not they were truly random/unbiased samples, which, is unlikely given that the data is self reported.