V is for Victory!

by endlesspsych

By Keir Liddle

By now most of you will have heard that the BCA has dropped its libel suit against science author Simon Singh (more details can be found here). Hilariously the case, which hinged on Simon’s use of the word “bogus” to describe Chiropractic treatments, has been dropped during Chiropractic Awareness Week, where it has been discovered that:

New consumer research by the British Chiropractic Association shows that, in just two years, the incidence of poor posture appears to have risen at an alarming rate. Over half (56%) of the UK population currently believe they have a bad posture, compared to 38% in 2007, an increase of 16%.

A claim that many skeptics may be chomping at the bit to test, and see if it is bogus. However, they would do well to remember that, although the battle is won and the case against Simon has been dropped, the English libel laws are still crying out to be reformed. All of us at The Twenty-First Floor encourage you to sign up to the libel reform campaign if you have not already – if anything deserves to be an election issue, then this (not forgetting sci-vote, of course) does!

As pleasing as the announcement of Simon’s victory has been, it is not the only victory for rationalism this week as Dutch Nurse Lucia De Berk yesterday won her appeal against seven counts of murder and three of attempted murder. While Simon Singh’s campaign has been the focus of the skeptical and scientific communities for many months now, Lucia’s case has perhaps attracted less attention – although Ben Goldacre of badscience has been championing her cause (see here and here) and there has been an active campaign to acquit her.

Lucia was essentially convicted because of the innumeracy of the Dutch legal system (I don’t suppose that ours or anyone else’s is any better), and a massive misunderstanding of statistics and probability. Henk Elffers’ former statistician (now a professor of psychology of law) was called in to determine whether seven deaths in a row during Lucia’s shifts could have occurred by chance, and came up with the figure 1 in 6,000,000,000. He was then further requested by the court to analyse data arising from other hospitals at which De Berk had worked and came up with the figure 1 in 342,000,000. This, he stated, was the chance that these individuals had died by chance during Lucia De Berk’s shifts. However, Henk had made an elementary statistical error: he extended his analysis to two wards of another hospital where Lucia had worked; but not the wards of two other hospitals where she was also charged with murder, and not in an earlier period during which she worked at JKZ, but when nothing happened.

In short, the selection of events which Ellfers explored biased the results in favour of finding Lucia to be guilty. Looking at the data, and correcting for Ellfers’ combination blunder, it has been calculated that the chance of something like this happening by chance was in fact 1 in 44, not 1 in 342,000,000. As campaigners pointed out, that would mean that 1 in every 44 Dutch nurses could find themselves convicted of murder. Or, as Dr Goldacre, put it:

Let’s say you worked in 20 hospitals, each with a harmless incident pattern: say p=0.5. If you multiply those harmless p-values, you end up with a final p-value of 0.5 to the power of 20, which is p < 0.000001, which is extremely, very, highly statistically significant. With this mathematical error, if you change hospital a lot, you automatically become a suspect. Have you worked in 20 hospitals? For god’s sake don’t tell the Dutch police if you have.

I recommend anyone who is interested in any other cases where the legal systems’ innumeracy has led to dubious convictions to read Gigerenzer’s “Reckoning with Risk“.

While we do have double cause for celebration this week, with two victories for rationalism, it is important to remember that Lucia was convicted of these murders and sentenced to life imprisonment in 2003 – it has taken until yesterday for the truth to win out and for her to be set free. It is also important to remember that while we Singh while we are winning, we must not let the libel reform campaign stutter or falter – lest more scientists, science writers and skeptics fall foul of quacks and pseudoscientists who are better funded than they perhaps ought to be.