For this study, researchers conducted technical interviews of 48 computer science undergraduates and graduate students. So they weren't studying real interview

For this study, researchers conducted technical interviews of 48 computer science undergraduates and graduate students. So they weren't studying real interviews. They were studying fake interviews as performed by a bunch of PhD students at NC State. And it turns out those fake interviews weren't good at assessing software skills. This just doesn't seem like a meaningful study. Interviewing is a skill; you can't just assume that a random PhD student is just as good an interviewer as an experienced, professional software engineer. Heck, many PhD students won't be able to pass a typical tech interview themselves.