Search This Blog

Tuesday, September 05, 2006

Why Is It Allowed?

I was thinking about Dr. Blumsohns problem with Procter and Gambel. They gave him some data from a clinical trial. He was to analyze the data and provide his results to the company. They would then use this as evidence that their drug works. The problem was that they didn't give him the codes so that he could complete his analysis. I wanted to think of an analogy that could describe the case. For the life of me I cannot understand why everyone in the world, including non-scientists, doesn't see what is wrong here. Why isn't this an open and shut case where P&G apologizes and fires the staff who tried to pull this trick on the world?

I thought about the O.J. Simpson murder trial. You label blood samples. Some come from the crime scene. Others come from suspects. Others come from people who were known to be at the crime scence such as the victims and the investigators. You then analyze the blood and put the matches together. You still don't have anything. What makes this process useful is knowing the codes. Let's say you've got samples 2 and 17 matching up. You reveal the codes and you find sample two came from the crime scene and sample 17 coming from a volentary blood draw from O.J. Simpson who is claiming to have not been at the scene.

The technical aspects of matching DNA samples to each other doesn't say anything. What matters here (what brings the cargo) is that a suspect now has a piece of evidence linking him to the scene of the crime. He says he wasn't there. His DNA says he was.

Back to Dr. Blumsohn. Is this similar to his case? He has the data but not the codes. He finds very little differences between any of the subject groups. Perhaps the drug did not do much better than the placebo. Perhaps the placebo worked a little better. I'm going to assume that DNA profiling is much more reliable than data analysis from a clinical trial. None the less, you need to know what group was what in order to really finish the job.

Dr. Blumsohn also accuses P&G of cherry picking from the data to make up a story different than the one he would have told. How would P&G know what data they needed to weed out? They would have had to know what the codes were. If group A was known as the drug group and it was similar to Group B, the placebo group, then these two groups would have to have some data thrown out. What is the desired outcome? That determines what data stays and what data goes. You can't pull this scam without the knowledge of the codes. And this is what Dr. Blumsohn was trying to get so that he could practice something a little different, known as science. Scams and science require in depth knowledge of the truth.

They wanted Dr. Blumsohn to remain blind. Why isn't this obvious? Why is it allowed? My guess is that there are mortgages to be paid, braces that are needed and so on. These transgressions are thought of as white lies. One data point here and there are not big deal. The little details matter however. It should not be allowed. P&G and Sheffield University should be ashamed. They have a new Cargo Cult ceremony. They've covered the eyes of those standing along the runway. They are making airplane noises. The lack of cargo can be easily explained away, because that is what they who make airplane noises are also skilled at. But the let one of the commoners stand along the runway with his eyes open. He saw what they do and he is trying to share with us what is happening. We still can't see. We heard the airplanes. We know the cargo will be here soon. People are getting ready for it. Isn't that proof enough?

No comments: