Amanda Knox and Statistical Nullification

In today’s New York Times, Leila Schneps and Coralie Colmez correctly warn that

… math can become a weapon that impedes justice and destroys innocent lives.

They discuss Lucia de Berk, and Sally Clark, two unfortunate people who were convicted of crimes based on bogus statistical arguments. Statistician Richard Gill helped get de Berk’s conviction overturned.

So why do Schneps and Colmez argue that statistics should be used to help re-try Amanda Knox in a what will be another Italian travesty of justice?

They criticize the judge for bad statistical reasoning. Referring to a new test of DNA on a knife, they say:

His reasoning? If the scientific community recognizes that a test on so small a sample cannot establish identity beyond a reasonable doubt, then neither could a second test test on an even smaller sample.

They go on to argue that basic statistics tells us that further testing would “tell us something about the likely accuracy of the first result. Getting the same result after a third test would give yet more credence to the original finding.”

Indeed. It would confirm that she had touched a knife in her own apartment. I guess you would find my DNA all over the cutlery in my house. I hope no one uses that as evidence in a murder trial.

The Amanda Knox trial was a joke; it never would have made it to court in the U.S.

All of which raises a question. Who would testify as a statistical expert in a court case without taking into account the context? Even if I thought Schneps and Colmez were correct in their statistical reasoning, I would not testify on the narrow question of whether a second test would strengthen that particular piece of evidence if I thought the trial was a scam.

Some time ago, I was asked to be an expert witness on a statistical issue regarding random drug testing of high school students in Pennsylvania. The attorney working for the tyrants (the school district) tried to convince me that he had a good statistical case for doing such testing. I told him that, despite the attractive fee and the correctness of his statistical arguments, I would not testify in a case that helped the government harass its own citizens.

What Schneps and Colmez failed to mention was context. I don’t care how sound the statistical argument is. If it is being used for unethical and unjust means, we should refuse to testify. Call it Statistical Nullification.