The Inherent Invalidity Of Statistical Inferences

To infer in the sense of believing irrationally is to succumb to the idiocy of thinking that non-necessary truths about one object, person, or other thing must also be true of another.  An example is the fallacies of thinking that because the last two times someone saw a dog, the dog attacked them, that all dogs or even all dogs of that breed will be savage.  Another example is thinking that just because one has not become sick after years of a careful diet, that one will continue to not be sick as long as the same diet is adhered to.  There are examples of a more grand kind--there is no way to prove that the sun will rise tomorrow through memories of it rising each morning in the past, or to prove that that gravity will not suddenly act differently on matter than it does now.  Someone who believes these things are knowable has made assumptions through inferences.  A rationalist might realize that it seems like a given thing is true in light of other perception-based, probabilistic evidences that all fall short of logical proof, which are often memories of past experiences, but he or she would not believe that these evidence-based probabilities are true.

Inferences, as one can find out from a few moments of rationalistic thought, are the very essence of almost all popular beliefs about statistics.  Statistics about various things are not necessary truths like logical axioms.  They are only correspond to logical possibilities for the many variables and events that do not contradict logical axioms, even if the actual nature of accurate statistics is not understood by many people.  For instance, if four out of a set of four men choose a career in the military, this has nothing to do with the false idea that men are naturally violent or callous, and there is also no way to know if it was individual personality or cultural pressures that led them to make this choice.  This is yet another limitation of statistics obvious to anyone open to not making assumptions: not only is it impossible to prove hearsay statistics or to validly extrapolate from them to another population or set, but even an accurate statistic, such as that expressed by "42% people like steak," would not demonstrate why the statistic is accurate.  Almost all statistics are just unverifiable hearsay or exaggerated, assumed ideas in the first place, but even accurate statistics would not prove causal connections between things, only correlations at best.

There are many logically possible reasons why whatever percentage of people actually like steak--not that I or any being with my epistemological limitations could possibly know that percentage--might enjoy this kind of food, so it is folly to pretend like it could only be one factor that is responsible.  Moreover, one generation of people might like it for one reason, to be replaced by another generation of which different statistics would be true.  Even accurate statistics about most miscellaneous things people like to cite statistics for could suddenly change because there is nothing logically necessary about the statistic remaining constant.  The percentage of animals in a population with a specific gene could change, the ratio of people within a demographic who say they like a specific genre of music could change, the percentage of American buyers who purchase a specific product could change, and there is no logical necessity in these fluctuations, as each of these changes would be a logical possibility that cannot be proven by prior events or percentages.

It is asinine to believe that 68% of American taxpayers have a certain political stance or that 90% of a certain organism will/must exhibit some happenstance behavioral characteristic that does not define the nature of the species.  Fools confuse random events, correlations, or arbitrary trends that do not by logical necessity prove anything more for proof of random ideas.  Ask five people if they are rational, and if three say yes, it does not mean that 60% of the give people truly are rational (they would have to be rationalists for that to be the case), nor does it mean that 60% of all people are rational even if this was true.  Just because four out of a specific group of 10 people said they would sacrifice free time for the sake of a career does not mean 40% of people care about careers more than free time.  When people extrapolate statistics that cannot be proven to begin with, this is exactly the kind of stupidity they indulge in!  They are compounding assumptions by extrapolating something that might or might not be true another thing that is irrelevant or unverifiable.