1. 15
    1. 4

      One threat to validity not mentioned is male vs. female speech patterns; if gender was still identifiable, bias could still creep in.

      Pretty interesting data nonetheless.

    2. 2

      Fascinating research and great write-up but I think there might be a flaw in the logic at the end: if a big chunk of the women who do poorly drop out after the first couple of interviews but very few men do then the equal performance of the ones who stay isn’t really a signal that women perform equally well because we’re then just looking at the ones who did well in the beginning, which is a skewed sample.