How to spot whodunnit: academics crack Agatha Christie's code

The article notes that M=27 books were analyzed to create the "whodunnit" formula, but N=83 books were published in her lifetime. So, does the formula work for N, or only M?

If we employ our little gray cells, it should be possible for us to determine just how accurate the formula is by running it against the novels that weren't analyzed.

If I had to put my cards on the table, I'd guess that the formula is overfitted.

http://www.badscience.net/2006/12/mediaslut-ideas-money-corp...