On the Scrumdevelopment list, there was a thread about interviewing candidates. I didn't see the thread until too late to be useful. But I have comments and thought I would share them with the greater community.
A poster said,
We didn't try to pair with the whole team during the interview – just one person with perhaps 2 others observing. Our general interview was 45-60 minutes of question discussion of topics with 2 – 3 people & candidate. Then about 45 – 60 min. of pair coding exercise.
Issue 1: Auditions. Pairing, as long as you pair at work, is a great audition. If you do not pair at work, it's a terrible audition. Why? Because it sets the candidate's expectation that you do pair at work. Then, if the candidate gets to work, and you do not pair, you have disappointed the candidate on the first day. Why do that??
I assume these folks did pair at work. For those of you who do not pair at work, develop your own audition. I have many audition examples on this blog. Search for auditions and use them as idea generators for your position.
Issue #2: Observers. I did not get to ask the original poster any questions, so I don't know the context. But for those of you who think that 45-60 minutes of questions with 2-3 people is a good thing—nope, I disagree. If I'm the candidate, I can control the room, because I'm great at interviewing. That's all, just great at interviewing. And, that wastes the time of the observers.
Remember, the best interviews are conversations. So have a conversation—a one-on-one conversation. Use an interview matrix, with each interview focused on just two areas, letting the conversation flow with behavior-description questions.
Now, you have 45 minutes of experience with the candidate yourself. And your two other colleagues also have 45 minutes of experience, so you have three times 45 minutes of experience with the candidate, not just 45 minutes of experience with the candidate. You can compare notes about the candidate and see if the candidate said something different to each of you, or said the same thing three times. You don't get that kind of data when you have the same interview.
Issue #3. Learning from the Interview. No matter what I like about your interviewing or not, you need to learn from your interviews. Have you set up a way to learn from your interviews?