It's literally the entire point of the article.
1. There is no interviewer bias: because women performed exactly the same even with voices that sound like men's voices. Which has the serious flaw I pointed out. I.e "we made women sound like men but they still under-performed in interviews compared to men" so therefor interviewers must not be biased. Ugh. That's terrible science. Because if interviewers can still detect gender after voice modulation then the voice modulation accomplished nothing.
2. Since the interviewers could no longer be biased, the gender gap in performance must come from something else. Oops, the experiment never proved that the interviewers couldn't detect gender. The writer goes on to talk about how bad the problem of attrition is, but I see nowhere that the writer claims performance differences disappear after controlling for attrition. And there is no proof that the attrition wasn't caused by....... women leaving after interviews they felt were sexist.
This is just piss poor science. Which explains why it's an only article and not a peer-reviewed study. And why it got no attention after it was published in 2016.
>>Once you factor out interview data from both men and women who quit after one or two bad interviews, the disparity goes away entirely.
And it definitely does not prove that the voice masking worked to hide gender.
I have personally heard interviewers say:
- a candidate is "too sexy"
- they want "hot, young Java talent"
- you were hired because "you looked good in a suit"