Proofiness - Charles Seife [79]
Even though it was a bogus argument that helped put a likely murderer back on the streets, Dershowitz was doing his job. Defense attorneys are supposed to use every legal resource at their disposal to protect their clients. They are the last line of defense when the government tries to rob someone of his freedom, and the price of a robust legal defense system is, on occasion, letting some criminals go free. Defense attorneys sometimes use proofiness to get their clients off, but it’s not nearly so grave a sin as using proofiness to put someone behind bars.
There’s a saying: any prosecutor can convict a guilty man, but it takes a great prosecutor to convict an innocent man. Proofiness is a wonderful tool for creating great prosecutors. By using an expert witness who spouts mathematical nonsense under oath, a prosecutor can make even a blameless person seem guilty.
One notorious case comes from the United Kingdom. Sally Clark, a British lawyer, was convicted in 1998 of murdering her two infant children. She was convicted in part because of the spectacular testimony of a famous pediatrician, Sir Roy Meadow, who used an incorrect probability to “prove” that Clark’s children weren’t victims of sudden infant death syndrome (SIDS). With SIDS out of the way, Clark had to be the killer.
Meadow made his name in the 1970s by identifying a syndrome that he named “Munchausen Syndrome by Proxy,” a mental disorder where a parent (usually the mother) gains attention by faking—or causing—a child’s illness. His career soon became a crusade to protect infants from their mentally ill parents. He was particularly concerned that many infants with the poorly defined catchall diagnosis of SIDS had in fact been murdered. He was so skeptical of the condition that at one point he suggested abandoning the term SIDS altogether.
In 1996, Clark’s newborn son died, and the coroner treated it as a case of SIDS. But when her second newborn son died two years later, the coroner changed his mind. Within a month, Clark was accused of serial murder: smothering her first child and shaking the second to death.
Meadow was a key witness for the prosecution, and helped seal the case with the stunning statistic that the likelihood of having two SIDS cases in the same family was about 1 in 73 million—or as Sir Roy put it, it was more likely that a gambler would win four successive 80-to-1 bets at the racetrack than for Ms. Clark’s family to suffer two cases of SIDS. A two-SIDS family, he argued, would only occur in Britain about once in a century. The jury found Clark guilty, and the judge sentenced her to life in prison.
The problem was that Meadow’s testimony was worthless; it was a piece of risk mismanagement. The 1 in 73 million figure came from multiplying the likelihood of SIDS in an older, nonsmoking, wealthy woman’s household (1 in 8,543) by itself. Unfortunately, this figure was vastly inaccurate for a number of reasons. SIDS seems to have a genetic component (as well as an environmental one), thus a family that experiences one SIDS death is at increased risk—the odds are rather greater than 1 in 8,543—for a second, so the overall odds are significantly greater than 1 in 73 million. Even more serious is that the figure is misleading in a similar way to the O. J. Simpson abuse statistic. In isolation, the likelihood of two SIDS deaths looks very small, but the proper statistical procedure needs to compare that probability to the probability of a person’s being a serial murderer, which is even smaller. Just as it’s better than even odds that a previously abused woman who winds up murdered has been murdered by her abuser, it’s better than even