Correlation does not imply Causality. If two measurements A and B are correlated, it does NOT imply one causes the other. Both could be caused by some other phenomenon C. If only the CMMI experts knew this simple theory, the lives of millions of Software Engineers would have been a lot simpler.
The general flow of the book is like this - A problem is identified. Data is collected on factors that could possibly influence/cause the problem. Then the authors analyse the data to identify the cause of the problem - not merely a correlated measurement.
Switching the cause and effect could lead to funny conclusions in some situations. For example, the volume of rain-coats, umbrellas sold in a city is HIGHLY correlated to the amount of rainfall it receives. It is easy to see the cause and effect here. In a far more complicated situation, it is easier to get the cause and effect reversed and not know about it at all. For examples - read the book :-)
Another interesting idea presented is the role of genes and parental behavior in a child's success. Even before a child is born, many factors that influence his/her success are already shaping up in the form of the parents' education and more so, the genes! According to the authors' analysis, other environmental factors do play a role, but rarely do they reverse the role played by the genes.
In all, an interesting read.
No comments:
Post a Comment