Statistical Malpractice
Not everything that counts, can be counted."
"Follow the Money!"
--Anonymous
"Statistics Don't Lie,
But Statisticians Do!"
http://pss.sagepub.com/content/22/11/1359.short
Commissioner/Politician/Master Propagandist Tom Farley
NYC Department of Health and Mental Hygiene
Since most people are not skilled in analyzing statistical data, it is very easy to scare the public
by making exaggerated claims of projected risks,
even when those claims are based on unproven theories and flawed data.
The Rabbi/Mohels, who have been unfairly attacked and demonized by NYC Mayor Bloomberg
have spent a great deal of time and effort in mobilizing reknowned experts
capable of carefully reviewing the City's data and exposing their unwarranted conclusions.
The City has not cooperated in this effort. They have stonewalled the Rabbis
at every stage of the investigation. It was necessary to file
"Freedom of Information Act" lawsuits to pry the underlying data
from the hands of reluctant City officials.
We leave it to you, the public, to judge why the City is so unwilling to reveal
the basic data on which they based their fear-mongering campaign
against the Rabbi/Mohels
Why has the City--up to this very day--been
unwilling to do DNA matching tests to either confirm or invalidate their claim
that the Rabbi?Mohels have been responsible for
an infinitesimally small number of herpes cases in New York City?
Here are analyses by experts of the intentional skewing of data
In New York City's fear-mongering campaign.
For those of you who are unfamiliar with, here are the relevant statistical concepts
as they apply to the BrisMILAH issue, here are some definitions and examples:
Selection Bias. If you don't do a random sampling on large numbers of cases, you can't draw any conclusions at all, because there may be any number of unidentified variables that have really cased the phenomenon you are studying. The BrisMILAH accusations are not randomly sampled. At best, they may be classed as a case series, which requires confirmation by DNA testing to have any credibility at all in a scientific study.
Base Rate Fallacy. If you do a comparison (update) of new information as compared to old information to gauge the effect of the new variable, you often make this mistake. If you incorrectly identified the prior probability, your conclusions will be incorrect, even if you performed the update process correctly.
Sally Clark and Husband
A Victim of Statistical Malpractice
A notable example of the horrible result of such statistical malpractice is the Sally Clark story, in which a British mother was put into jail based on mistaken statistical analysis (Base Rate Fallacy) applied mistakenly to her two children who died of SIDS. After serving three years in prison, she died from drinking too much alcohol, caused by depression from being wrongly accused of killing her two children.
A Famous study of medical professionals found that only 15% of doctors gave the correct answer when confronted with a Base Rate Fallacy problem. http://libra.msra.cn/Publication/1267301/interpretation-by-physicians-of-clinical-laboratory-results
