…doesn’t mean you understand statistics.
I am not even going to pretend that I understand all there is to know about statistics. I know enough however, in the course of my job to understand the prinicple of causality and correlation.
Causality (but not causation) denotes a necessary relationship between one event (called cause) and another event (called effect) which is the direct consequence (result) of the first.
In probability theory and statistics, correlation, (often measured as a correlation coefficient), indicates the strength and direction of a linear relationship between two random variables.
So one is does A have any effect on B and the other is how strong is that relationship.
The reason I am going off on this is in the Calgary Herald there was this editorial which included the following statement in it’s conclusion.
Enthusiasts of adolescent sex, their progressive parents, and educators would be wise to recognize that abstinence and later monogamy have never failed to prevent HPV infection.
Stanislaw Iwanicki, MD,
This is seriously flawed logic. Promiscuity may increase potential exposure to the disease, but there are some 130 types of HPV and about 30-40 which are transmitted sexually. One study found that, during 2003–2004, at any given time, 26.8% of women aged 14 to 59 were infected with at least one type of HPV. (from Wikipedia)
The good doctor would have us believe that Christians, Catholics in particular if you read the article, do not get HPV or cervical cancer. We all know that isn’t true. So his statement is false. Abstinence and later monogamy have never failed to prevent HPV infection, but they have never succeeded in it either. They may have nothing to do with each other. While monogamy may decrease exposure to the disease it is not a guarantee.
Another example: The majority of people in car accidents are married. Therefore marriage makes you a bad driver. True or false?
Obviously the answer is false. It could be that because it is cheaper to live as a couple, more married people own a car. Or it could be that the median age of accident victims is 47 years old and by that time the majority of the population is married. It doesn’t matter because one has nothing to do with the other.
If you are interested in statistics I would recommend watching TED talks. Here is one by Peter Donnelly that is especially good.
At 11:15 in there is an interesting example of an AIDS test and statistics, and then a mind blowing example of how a poor woman was accused of killing her baby because a doctor got the statistics wrong.