The case of Lucia de Berk

The posting of a paper by Richard Gill, Piet Groeneboom, and Peter de Jong on arXiv today reminded me of a conference of Richard Gill in Ottawa two years ago where he vehemently defended the Dutch nurse Lucia de Berk. (She has been exonerated from all murder accusation this year, after spending several years in jail.) The current paper gives a very simple explanation of the lack of strong (statistical) evidence against this nurse, which makes the earlier conviction based solely on statistical arguments the more puzzling. (As in earlier cases, the fact that the statistical arguments were delivered by a non-statistician is also very surprising, This shows that judges should both get some basic training in Statistics, rather than considering forbidding statistical argument in court, which I think also is the position of the French courts, and that they should involve statisticians as experts.)