Dec 12, 2023 20:00:00

How can 'wrong science and statistics' in courtrooms convict innocent people?

Modern criminal investigations place emphasis on physical evidence based on scientific analysis and the opinions of experts, but overconfidence in science can sometimes lead to serious mistakes. Scientific American featured a case in which seemingly correct scientific testimony led to a wrongful conviction.

Bad Science and Bad Statistics in the Courtroom Convict Innocent People | Scientific American

https://www.scientificamerican.com/article/bad-science-and-bad-statistics-in-the-courtroom-convict-innocent-people/

According to a report (PDF file) compiled by the University of California, Irvine and the University of Michigan in the United States, of the 233 exoneration cases in 2022, 44 cases were based on forensic evidence or expert testimony. It is said that this was a factor.

Many of these problems stem from overreliance on scientific evidence and lack of legal expertise, but are exacerbated by expert witnesses who overstate the evidence or underestimate uncertainty. You may.

A classic example of this is the 1999 trial in which an innocent mother was framed as a murderer through the testimony of British pediatrician Roy Meadow. Meadow, who appeared in court in a case involving a woman named Sally Clark who lost her unborn children in quick succession, said, ``The probability of a child dying from

Sudden Infant Death Syndrome (SIDS) in such a case is 8,543.'' 'In other words, the probability that two children will die by chance is 1 in 73 million, squared.'

According to Meadow's law , ``Unless proven otherwise, the sudden death of one infant is a tragedy, a second is suspicious, and a third is murder.'' Clark was treated as a murderer by the media and at trial. He was found guilty and sent to prison.

The claim that 'the probability of 1 in 8543 occurring twice is 1 in 73 million' may apply in cases where each trial is completely independent, such as a coin toss, but if it occurs unevenly, the situation becomes it's different. In particular, there is epidemiological evidence that SIDS can occur multiple times within a single family, and the idea that it occurs independently is not supported.

Multiple incidents of SIDS are certainly rare, but multiple incidents of maternal infanticide are also rare, so to determine which is more likely, we must compare relative

likelihoods , or plausibilities. not. In the Clark case, the Royal Statistical Society severely criticized Meadow's testimony, and a rebuttal article was published in a medical journal.

Statistical errors like this that are often seen in court are called `` prosecutor fallacies .''

After a long legal battle, Clark's conviction was overturned in 2003, and Meadow's testimony also acquitted other convicted mothers. Meadow has also been found guilty of professional misconduct by medical authorities and banned from practicing medicine. However, the exoneration did not save Clark, and he died of acute alcohol poisoning in 2007.

Prosecutorial error occurs in a variety of different cases. For example, even if a blood sample taken at a crime scene has a rare blood type with a 5% prevalence rate and a suspect with that blood type is found, it does not mean there is a 95% chance of being guilty. Because even in a small town of about 2,000 people, there are 100 people with that blood type, so the chance that the suspect is the culprit is only 1 in 100, or 1%.

Scientific American writes about the role of science in courtrooms: ``Science and statistics are vital in the pursuit of justice, but their uncertainties and shortcomings must be made as clear as their strengths.'' 'Evidence and statistics also require context so that they are not misleading.'