Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I mean, Hunter & Schmidt says three really key things:

1. IQ tests work real good.

2. Structured interviews -- which are not IQ tests -- work about as well and strongly correlated.

3. Work sample tests -- which are _definitely not_ IQ tests -- work as well and in ways NOT correlated.

Meaning you can combine these two and get a reasonably effective hiring screen.



Yes, and even when applying those criteria the results still were far from great. They were at least better than random chance though.


How good can the results possibly be when you have ~1 day with someone to assess how they'll be over the next year or 2?


I agree. The person to whom I was replying seemed to be implying that interviews can provide good signal.


I mean, 70 percent is pretty good for a selection strategy. I wasn't aware H&S had published a followup and now I'm curious if it ever got a peer review, because their table contradicts some earlier work suggesting that structured works better than unstructured.

It's also pretty keen to recommend Integrity tests which I think the authors play a role in studying.

edit: I guess it's "operational validity" which maybe isn't "percent that did well later"?




Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: