zlacker

>>zdw+(OP)
As usual, the actual numbers don't back up the narrative. For example, significantly more men in the study left due to unfairness than women do: 40% vs. 31%. So either women are treated more fairly or these numbers don't mean anything, yet you would never know it from the text, which is all about the horrible things that happen to women.

Hmmm...

>>mpweih+D4
An important point here - every time I read "X% of class A had this happening to them" I automatically ask "what's the baseline rate? If you weren't in class A, would it happen to you more often or less often?" If I don't find this comparison - and usually I don't - then the number of X% is useless for the purpose of deciding whether class A is an outlier or not. Unfortunately, as I said, very small number of people quoting the statistics do this basic due diligence.

In this particular case, it is natural that people that believe they are treated unfairly leave, and conversely, if somebody left, one of the probably reasons is because they were unhappy, and one of common reasons of unhappiness is that they don't believe they were treated fairly. The question pertinent to the purpose of the article is do e.g. women have it worse? For that, one needs more than one number, so any claim that includes only one number is automatically disqualified from providing proper information - it provides no more than half of it.