Good Interview Questions

[Jim Logajan]

There are no good interview questions because it has been shown that the typical interview technique sucks at predicting job performance relative to other measures.

Few people who are called on to do interviews have any training in doing them.

Consider the following table from "Validity and utility of alternative predictors of job performance." Hunter, J. E. and Hunter, R. F. 1984 Psychological Bull. 96:72-98

Predictor: Validity in Predicting Job Performance Ratings

Cognitive test score: 0.53
Biographical data: 0.37
Reference checks: 0.26
Education: 0.22
Interview: 0.14
College grades: 0.11
Interest: 0.10
Age: -0.01


There is something known as structured interviews that are almost as usefule as cognitive scores, but nothing anyone has posted indicates familiarity or use of them.

Interviews might provide some gut comfort, but unless you've studied and carefully designed your interview technique, you may as well stick to using interviews to select based on compatible personalities, and stick to other mechanisms for selecting for highest probability of good job performance.

 

[jsstevens]

Interviewing software engineers I like to ask "What's the worst bug you ever found and fixed and how did you go about it?" it generally relaxes a nervous candidate because now we're just swapping war stories. It also gives me insight into their approach to problem solving. And if they're just BSing me I'll figure it out pretty quick.

John
Sent from my iPad using Tapatalk HD

 

[gismo]

Back when I was at FedEx, I had three candidates that seemed equally qualified for an opening. One was an engineer, one a mathematician, and one a physicist. So I asked each one a single tie-breaker question. I brought them into my office individually and pointed to a red rubber ball sitting on the meeting table.

"Calculate the volume please." I asked.

The mathematician measured the diameter, divided it by two to obtain the radius, and then used the formula 4/3*pi*r^3.

The physicist submersed the ball in a graduated beaker of water. The displaced volume of water is the volume of the ball.

The traditional engineer turned to her reference text The Physical Properties of Balls and in the chapter entitled "Rubber", found the table labelled "Red". Searching for a row that the contained the appropriate model number (which is stamped on the ball), she read across to the column "volume", ignoring those dealing with "coefficient of thermal expansion" and "software rev. level".

If you'd included a candidate with business/marketing degree his answer would probably have been "Who cares? We can sell it no matter what the actual volume."