One of my jobs at work is to interview folks. And so, as my regular readers know, I am a major fan of predictive analysis of job skills.
Indeed, I thought we had pretty good evidence suggesting that the order of predictive strength went:
Self-efficacy > Delayed Gratification > Conscientiousness > IQ > most other stuff, with experience being also a major factor.
Imagine my surprise when I read instead this list (HT: Kling) reporting on a meta-analysis of job success predictivity.
IQ > Work Sample > Integrity(tests) > Conscientiousness > Structured Interview > Unstructured interview >>>> reference checks > years of experience >>> years of education > interests > graphology > age.
Hilarious....handwriting analysis is nearly as impactful as years of education.
I suppose then it's good that I've been pushing work-samples and more structured interviews as a prereq for IT employment at my company, and making that an integral part of the interviewing process I run.
It's also good that I am a damn fine estimator of IQ, from simple conversation.
However, it seems to demolish my position that Practice/Experience >>> other stuff. I reserve here the right to suggest that skills in many jobs are sufficiently narrow that the measures of years of experience SUCK. On the other hand, this fits well with the widely studied feature of Progammers, for whom there is a 10x variation between efficiency of a top 10% programmer and a bottom 10% programmer, with the same amount of experience. I had thought this wouldn't be as true of other occupations.