It's not possible to compute the false negative rate because once you turn down a candidate you get no more information. I suppose if people apply multiple times and are accepted later and then go on to do well on the job, you can count them as false negative in the first interview. The sample size for that would be small though. And I'm sure the true believers would say that these candidates improved their programming skills between the two interviews, not their interviewing skills.

I am quite interested in the false positive rate which can be computed by correlating performance reviews with interview performance.