Hacker News new | comments | show | ask | jobs | submit login

So the thing is, "smart" is a pretty sloppy term. Furthermore, claiming that measuring this "smartness" in a small-sample, unrelated set of tasks (a relaxed, 30 question test), and that this "smartness" on the test would transfer to "ability to write good code well and work in a team" is utter crap.

Doing well on a test is usually an indicator that you do well on that kind of test. Even good psychometrics tend to fail with calibration problems - the vast majority of these tests were written and calibrated using psychology undergraduates at western universities. That's a significant sampling problem. [0]

So I'm personally not opposed to selecting for "smart" employees. But I do think that the belief that a short test actually selects for that kind of smartness in any meaningful way is sloppy thinking.

0: http://blogs.scientificamerican.com/primate-diaries/the-weir...




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

Search: