Page 317 - Special Topic Session (STS) - Volume 2

STS496 Alphonse L.M.
misapplication of established methodology, the public at large normally learns
of the failure of statistics from media. Two examples will be presented to
illustrate the issues involved in cases of alleged misconduct of statisticians.
Miscarriage of justice: Netherlands. On 24 March 2003 Lucia de Berk, a
paediatric nurse, was convicted to life imprisonment accused of the murder of
four patients and attempted murder of three others. The case was brought to
the attention of the police by hospital authorities where she worked after it
was alleged that a large number of incidents (deaths and resuscitations) took
place during her shifts at the hospital. The probability of such occurrences by
chance was considered very low (1 in 342 million. Her conviction was officially
based on two non-natural causes certified by a medical expert, but in the
media and the proceedings repeated reference was made to the low
probability that the number of incidents could occur by chance. The
calculations had been made by a court witness/specialist, a professor of
psychology of law, who some 30 years before had obtained a Master’s degree
in mathematical statistics, but who had abandoned statistics and continued
advanced studies in geography, psychology, economics and law. On 18 June
2004 she was convicted by a Court of Appeal to life imprisonment for the
murder of 7 patients and the attempted murder of three others. The conviction
was based on two cases for which the prosecution claimed to have sufficient
proof that she had poisoned them. For the other cases no proof of her guilt
was apparent, but by implication (chain-link reasoning) she was also held
responsible because they also had occurred on her shifts. The case had
considerable media exposure nationally and internationally, but there were
considerable doubts about the soundness of the procedures and her
conviction. A Committee of Support was created led by a Dutch philosopher
of science and a medical practitioner who doubted the correctness of the
medical and statistical evidence and the way these were used. Prominent
Dutch and foreign statisticians also expressed serious reservations about the
statistical calculations presented by the prosecution’s statistical expert witness.
Following a widely supported public petition, at a retrial in 2009/2010 Lucia
de Berk was acquitted and in 2010 received an undisclosed amount of
compensation from the State.
Although the retrial allegedly was decided purely on legal arguments
observers had throughout stated that the process was incorrect because the
data were flawed, they had been collected post-hoc; the calculations of the
statistical prosecution witness/expert were incorrect; the Court committed a
prosecutor’s fallacy, and the same team of experts was used to establish
suspicion, measure the magnitude of the suspicion and acted as expert
witnesses in the court. Gill (2006) Derksen (2007), Meester (2007)
Incorrect census methodology, 2012 population census. Chile is one of the
Latin American countries with an excellent statistical organisation and

306 | I S I W S C 2 0 1 9