The Unreliability Of Statistics

Statistics about people can be thrown into arguments as if they actually have some authority, but a rational examination of statistics shows that they have little to no epistemic value.  For some people this may be new information.  But the authority ascribed to statistics is nonexistent.  Before I continue into explaining three important limitations of statistics about people, I will provide some definitions.  In statistics a population is the set of things/people researchers want to learn about, and a sample is the specific subset of that group that is targeted in observation or analysis.


Self-Reporting

If I were included in a study poll as part of a sample of the religious beliefs of a certain population, I can know that I have answered honestly, but not that the strangers in the poll have.  Self-reporting allows for the possibilities of mistakenly answering questions when one is personally confused (perhaps even unknowingly) or answering in a way intended to fit in with social expectations.  For instance, just because a majority of people in a study, even a nation-wide study, identify as Christians does not mean that they are.  They might not even know what it means to be a Christian; they might know nothing about actual Christianity at all.  The poll results have no ability to be legitimately verified.  Or consider another example.  Just because a member of a certain demographic gives an answer doesn't mean he or she really means it.  Perhaps that person just answered as he or she did in order to fit into stereotypes of the demographic to feel validated by arbitrary societal standards.


Unknown Causes

When it comes to studies about the characteristics of various people, even if a statistical number is correct ("67% of men have a tendency to overspend"), the statistic itself cannot legitimately be used to say that the cause or causes of this phenomenon are known.  In the hypothetical example I just gave about men and overspending, even if the results were true when extrapolated, I am sure that someone would claim that this study "proves" that men have a mental characteristic that makes them likely to spend money gratuitously.  But even if the results were correct this conclusion does not follow at all.  It is not that the men overspend just because they are men--either individual personality or external pressure via societal conditioning will inescapably be at the root of this, though the statistics cannot reveal this part, as they can reveal only the alleged results.  It is logic, isolated from statistics, that reveals this.  But fallacious minds might easily use confirmation bias to argue that statistics confirm some societal myth.

Another example might be a study from a particularly racist era of the 1900s concluding that African Americans are less intelligent that whites--when intelligence has nothing to do with being black or white, but, once again, with individual personality characteristics and societal forces.  In an time when blacks were not educated or taken seriously as intellectual beings, someone could have used a study like this to argue that they are less intelligent by nature of being black, when that not only does not follow at all, but it also is proven false in its entirety by reason.


Limited Results

Then, of course, there is the fact that even if all of the answers from each participant are honest and accurate, the answers at most only apply to those in the poll.  To say, to use another contrived example, that "50% of Americans believe in Platonism" because half the members of an American poll group said they believe in Platonism is very fallacious.  This commits the fallacy of composition and treats what is true of some people as necessarily true of others, when that is not the case.  And yet how many articles claim that a certain percentage of Americans or whites or women or some other population are a certain way just because of the alleged answers (which could be very misleading as I addressed above) provided by a very limited population sample?


Statistics are not a reliable basis for believing almost anything about other people.  The limited extent of the results, possible inaccuracy of the results, and inability to confirm the cause of the results (individual personality, social conditioning) altogether disqualify statistics from being a source of certain knowledge.  And yet some people prioritize statistical information!  This only testifies to the ignorance of some people, not some nonexistent reliability of statistics.