jellie writes "An advisory judicial committee, the Dutch Posthumus II Committee, will be reviewing the case of Lucia de Berk, a.k.a the "Dutch 'Killer' Nurse". In 2003, she was sentenced to life imprisonment for the murders of seven patients and the attempted murder of three more, based on the probability of "1 in 342 million" that all those deaths would coincide with a nurse's shifts. However, as detailed in a page by a Dutch mathematician Richard D. Gill, many of been questioning the statistics used in the case. From the article: "Curious that a mass murderer could kill so many people and simultaneously take care that the total number of deaths on the ward is actually lower than in a similar period before she worked at this hospital: this data is not incorporated in the analysis or even made available!" and "[The expert for the prosecution] apparently does not know the meaning of p-value. He multiplies three independent p-values... and appears to present the product as a p-value." Statistics are often used in courts to convince the judge or jury, but what happens when unreliable or inaccurate methods have been used in generating those numbers?
Other commentary can be found on Bad Science and on Mark Buchanan's blog (NYT TimesSelect req'd)."