Earlier quoted context omitted.
Note the correlations between MBTI types and Big Five personality dimensions: https://en.wikipedia.org/wiki/Myers%E2%80%93Briggs_Type_Indi... These are not insignificant, particularly in the social sciences. 0.45 would be a large effect size in most psychology papers; 0.72 is huge. I think a large part of whether people think MBTI is bunk or a useful tool depends on their willingness to overlook small details to get…
One bit of psuedoscience correlates to another. Okay. One test measuring extroversion only manages 0.74 correlation with another. I'm not impressed.
They didn't invent traits and then came up with fake data to prove their model. It's the opposite: they used statistical methods to discover those traits.
It might be an imperfect model, but it's still scientific.