I sent the following letter yesterday to the editorial office of The Times in UK, following an awkward reading I had on their newspaper the same day. I wish Stephen J Gould were here to see me 🙂
To Whom it may concern,
I bring to your attention what I consider a disturbing reading from today’s newspaper, Thursday June 11, by Science Editor Tom Whipple with the title “Paedophiles more likely to be left-handed”. The article covers a publication in Archive of Sexual Behaviour, published by Springer and authored by Fiona Dyshniku et. al.
I am said to see a journalist who should be interested in disseminating the true advances of science to be captured by a poorly written, bias and statistically flowed publication that aims to link men physical characteristics to their likelihood to be a paedophile.
Research aimed to find a meaningful correlation between body features (preferably a disgraceful one) and human behaviour dates back to the 19th century, when craniometry and criminal anthropology were fooling societies to believe that brain dimension, distance between eyes and forehead area were all meaningful measurement to identify criminals ahead of the crime. The measurement of subjective ‘physical abnormalities’ was used to strengthen the position of white men in societies, to justify colonialism and delay women emancipation.
I am sorry to see that today, The Times is bringing this old idea back in fashion.
In an indelicate and shoddy game of juggling numbers, Dyshniku et. al. measures head circumference and curvature of fingers to link physical characteristics to atypical sexual behaviour. The ‘Minor Physical Abnormality’ index used in the study was invented by someone who thought having the third toe longer than the second is to consider abnormal. Adherent hear, furrowed tongue and a (poorly defined) ‘excessive gap between first and second toe’ were considered by the investigators – in all seriousness – a relevant measurement to address the existence of a pre-natal predisposition to paedophilia.
To begin with, I will inform journalist Tom Whipple that the paper did not find a correlation between left-handed and paedophilies (though they try hard to). After a number of seemingly unnecessary way to measure a person’s propensity to use the left hand, authors regret to inform us that previous founding on this matter were not supported by the new data [though, of course, they find a positive association that yet has no statistical meaning].
I will also make The Times’ readers aware that the study in question is flowed. When performing a ‘Minor Physical Abnormality’ tests (a checklist of ‘abnormalities’ presents on a person’s body) investigators knew about their patient’s record, and whether he was charged of child-abuse crimes. Researchers bias is a factor known to interfere the final result, because unconsciously researchers may see flaws in bodies of criminals where there are none, just to fit their expectation.
Another, intriguing surprise is that no difference in physical abnormalities between paedophilie and non-paedophilie exist, though it is what everyone will assume by reading The Times today. What authors found, was a correlation between a ‘paedophilie index’ (described below) and a new number: they divided the number of abnormalities present in the head (let’s say, 4) agains the abnormalities in rest of the body (let’s say 2). This produce what they call ‘Anomaly Distribution Index’ (=2) that indicates the distribution of your abnormalities across a person’s body. The more on the face, the higher the Index will be. This number is a patchy arithmetic’s work that tries to normalise for the fact that the test measure more the head than the feet. This number correlates (quite poorly r=0.19) with your likelihood to get arouse when listening to child-sexual messages (the ‘paedophilie index’). Distorsion of numbers and unnecessary subtractions and divisions is a way to transform values that are not significant. We have no idea what is the biological meaning of the ‘Anomaly Distribution Index’ is – or whether it has a meaning at all. This abstraction also seems to suggest that are the abnormality in the face, rather than in the periphery, to spot a paedophilie. I hope the no-sense of all this to be widely acknowledge.
A second ‘index’ in the paper that present a similar problem is the ‘paedophilia index’. Investigators used changes in penis volume following different sexual-messages to find the man sexual preferences. But the value used for the correlation is not simply a change in penis volume, but a confusing sum of different tests that would enhance the likelihood of anyone to be seen as paedophilie, by the sole nature of how the formula is written.
To conclude, the high number of statistical analysis used is suspicious: a Pearson, a point-biserial, a Phi and a Spearman tests were picked in not particular order to demonstrate a correlation between any desired physical trait and different classifications of paedophilia.
- A link between facial characteristics and sexual behaviour is not existing.
- A link between viral infection or drug use during pregnancy and sexual behaviour is absent.
- Being left-handed does not make people paedophile.
I beg Your journal to consider more carefully its scientific interest and ensure to disseminate literature that isn’t just published, but it is understood by its journalists and properly received by the rest of the scientific community.