We do an all-resume-blind work-sample (WST) hiring process: a code challenge, a written (email) product design challenge, and a 2-hour interactive Slack design/implementation challenge.
A benefit of doing Slack “interviews” (this isn’t one but bear with me) is that you get a transcript of the session.
One thing we’ve learned: after-the-fact evaluations of sessions are very different than in-the-moment evaluations; maybe to the point where, if you’re the person administering the session, you shouldn’t be trusted to evaluate it at all.
A process feature I think I like: generate transcripts, pass them off to a review board. The “interviewer” is just a proctor.
It seems natural to assume that the after-the-fact evaluation is better (because it doesn't suffer from halo bias). How could we test that though (and how would we define "better")?
