Along with the good advice others have mentioned, let me add one thing.
I find it useful (and actually enjoyed interviews where this was done with me)
to bring a practical problem to the interviewee. For example,
'If you had to solve problem X, how would you do it?'. Then, give them
a piece of paper to allow them to write (maybe a short flow chart, or
pseudo-code). Then you can interact with them as you see their though
process. This was done at an interview with me, and I had a good time
interacting with the fellow interviewing. Not only could he see how I think
in real-time, but I was also able to follow his thought/design process.
It allows you to ask other follow-up questions like 'Why are you using module
Z and not module A?', and 'But what if X occurs?', etc...
Cheers,
KM