I agree, there is no bias in the estimates, there is some bias in the representation of users. There are many of them and they are active, others are very few, and others are simply inactive.

Such comparisons give ground for reflection.
Good job, my friend!