Friday, July 17, 2009

Statistics: A Perverted Art?

In my own statistical test of other statistical tests, I have concluded there is a 95% probability that any given conclusion is 800% more likely to be complete bullshit.....though there is only a 20% chance of that when it is cloudy outside! Then again, if my test is typical of most then it is bullshit anyway meaning the other tests might very well be valid.

I'm being facetious to illustrate how statistics are so commonly perverted as a means to explain anything or everything. Do you want to presuppose legalizing guns helps or hinders society? Or how about "proving" global warming is predominantly caused by man (or woman)? Perhaps you want to conclude the opposite? Just selectively choose your data and variables and I'm sure you can “prove” either.

I believe one of the dangers inherent in most of these statistical tests is that it unconsciously encourages many to identify with and support only conclusions already believed or at least wanted to be believed. In this way, it inflates their egos by giving them something to feel “right” about.

“This statistical conclusion performed by The Institute of Environmental Fascists (IEF) confirms my suspicion about us being the primary cause of global warming, therefore it must be true”, says the man. “Oh really, this test performed by Fossilized Carbon Collectors' Institute (FCCI) suggests otherwise,” says another man. One can see the inherent bias of these silly examples and though they are a bit exaggerated, I think the point is made. 

People want to believe things in order to feel superior and righteous about their preferred causes. Perhaps their belief in the “truth” also gives them a sense of control they would otherwise be without.

Let's also suggest the potential for a self-fulfilled prophesy to transpire only because it is foretold by a given conclusion. As an example, I can't help but think how many hypochondriacs have been created only as a result of having an allergy test which “dictates” that a given person should react to a certain substance. Might a reaction come about purely or at least partially in response to the belief that it should or would?

With diseases, social problems, mental problems, environmental problems, blah, blah, blah. The issue I have with statistical tests is not the tests themselves, but rather the underlying assumptions and variables used to reach the conclusion(s) which are almost always chosen only to validate something already believed or wanted to be believed, thereby creating meaningless circular arguments.

No comments:

Post a Comment