zlacker

It's not a matter of representativity. It's a matter of measuring a different thing.

The article measures shares of reasons between the people that left, but those are not correct by the share of people that left on each subgroup.