Several decades ago, even in socialist schools, it was the custom to teach children the phrase "
Examples of
post hoc reasoning are boundless and have been the source of much humor. Mark Twain often used post hoc humor such as in his story about the woman who died after neglecting to adopt the vices of smoking and drinking. Mark Twain's audiences in post Civil War America were very literate in the logic of common sense learned in a free enterprise society. If he were with us today, Mark Twain might find it prudent to have a helper planted in his audiences to give cues about the appropriate times to laugh.Americans today are still partially immunized against
post hoc arguments by this legacy from our intellectual past. Some mutant forms of post hoc logic do afflict us, however, such as the notion that a man cannot stand on his own two feet in society without tax subsidies and preferences because his ancestors over a century ago were enslaved or the fallacy that, since many Americans lead productive and decent lives after attending socialist schools, the socialist schools are to be credited with their accomplishments.(Selective observation is usually a part of such arguments. Socialist schools which encourage intellectual dishonesty and moral degeneration are not credited with the subsequent moral decay of adults.) Today, far more people fall victim to a close relative of the
post hoc virus than to the original virus itself. This new affliction is the fallacy that when two things are statistically correlated they are, therefore, cause and effect. Paraphrasing the original fallacy, people have come to believe "statistically together with this, therefore on account of this.'' (Fear of scholars who read Access to Energy prevents me from attempting the Latin.) Examples of the logical error that correlation proves causality appear almost daily in newspapers, magazines, and other information sources. They also appear regularly in scientific publications. "Scientists'' who should know better are even using this logic in situations where one of the two correlated phenomena is only postulated or even known not to exist.In the article "Fried Frogs' Eggs'' (
Access to Energy 21-9 p 4), we reported the claim by scientists that world-wide amphibian populations are decreasing because of increased UV light levels caused by depletion of the ozone layer. UV light is capable of disrupting living systems as illustrated by its use in water purification and the sterilization of medical instruments or its known role in the induction of skin cancer. It was no surprise, therefore, that these people found that frogs eggs had a higher hatch rate when protected from UV light.With this mundane observation in hand, however, and the goal of snaring large amounts of your tax dollars in government grants, they immediately pointed out a correlation between a reported recent decrease in world amphibian populations and a claimed recent increase
in world UV light levels - in their scientific report and in their (oh, so important for grant funding) statements to the press. Fame and fortune await those who link their work to the ozone mania.
The problem is that there were no measurements showing an increase in UV light. The best available data showed a slight decrease. The only increase in UV was in the public perception formed by false press reports. Correlation was claimed to prove causality in this case even when the correlation was to a
perceived reality, but not to measured reality. For achieving this new low in logic, these scientists were rewarded with large tax-financed grants.The abandonment of logic in the pursuit of tax money is the most important force driving the correlation-causality arguments that fill our media. Electric shavers and brain tumors; breast cancer and power plants; logging and fish populations - each is a correlation in pursuit of government expenditures, either direct monetary grants or indirect exercise of power through regulation and litigation.
There is only one way to immunize ourselves and our countrymen against this sort of lie. Each person must learn the truth that:
Correlation does not prove causality!
Examples can help. A common one is the strong correlation between churches and murders. There is a very strong positive correlation between the number of churches and the number of murders -because both are correlated with high population, not because churches cause murders or vice versa.
Since correlations are usually bolstered by statistical calculations giving the statistical reliability of the correlation, many people believe that "statistics lie'' or that it is possible to "prove a lie with statistics.'' This is nonsense. Statistics is a very valuable branch of applied mathematics and is a great help in establishing scientific truth.
The correlation between murders and churches is statistically reliable, since this correlation is definitely true. The error is in the logical conclusion that, if two things are correlated, they are cause and effect.
AIDS is definitely correlated with the HIV virus. Unproved, so far, is the hypothesis that HIV alone causes AIDS. One or more other AIDS-causing factors may correlate with both.
High cholesterol is correlated with heart disease. Unproved is the hypothesis that high cholesterol causes heart disease. (Recent experiments showing actual lowered death rates with use of cholesterol-lowering drugs do not indisputably prove this either, but they are a significant step in that direction.) Logic is an ordinary part of common sense. To this end, comedians may be much more effective than scientists in correcting errors in logic because they use memorable humor to ridicule errors of common sense that are carried to illogical conclusions.
Global temperatures have increased during the last century. Fortunately, their correlation with increased CO
2 fails because the temperature increase preceded the CO2 increase - unless one looks at the whole century and not at the yearly values - or at the ground data, not the satellite data. Correlations are often creatively manufactured.Correlation does not prove causality! This truth provides protection against much of the "scientific'' propaganda we endure.
|
|
Vol. 23, No. 9
Newsletter: Access to Energy Newsletter Archive Volume: Issues Issue/No.: Vol. 23, No. 9 Date: May 01, 1996 03:17 PM Title: Correlation and Causality
Copyright © 2004 - Access to Energy Newsletter Archive
|