Whenever we offer an empirical claim as true, we are expected to support that claim with evidence — pieces of information which, when taken together, tend to point to the truth of our claim. Evidence which proves our claim is best, but not absolutely required. Naturally, when searching for evidence one of the first sources we look to is our own experience in the world, because it is typically that experience which we rely upon most heavily when trying to understand reality.
Nevertheless, this reliance upon personal experience can be too strong and become too unreasonable. The use of small pieces of anecdotal evidence as a basis for conclusions about wide-ranging phenomena and complex systems can be classified as a type of Rash Generalization, an informal fallacy which involves generalizing from a particular example to an the entire class of which that example is a member, without regard for any other possible factors.
Because the use of anecdotal evidence and testimonials from others is so common and often goes unrecognized for the problem that it is, it deserves to be given special attention here. It’s not a question about whether those personal experiences are true — even if we assume that the explanations are 100% accurate and true, problems necessarily remain. So what’s wrong with relying upon personal experiences?
People actually use anecdotal evidence all of the time; in casual situations, that may not be much of a problem. When considering what restaurant might be a place to go for dinner, using friends’ stories about their own experiences in different places makes a lot of sense. If we’re trying to figure out whom to hire to fix the plumbing in our house, we should take the time to ask neighbors about the best and worst people to call.
In serious discussions and with more debatable questions, however, the use of anecdotal evidence is insufficient. If we are trying to decide if accupuncture is a good and effective means to cure illnesses, it isn’t good enough to simply look to the experience of yourself and/or a few friends and upon that basis declare that, indeed, accupuncture is effective. If we are trying to decide if divorce is a problem in society, it isn’t good enough to look at the divorces in your own family and conclude, based upon what you see there, that there must be some sort of divorce epidemic requiring a solution.
Scientific inquiry and rational conclusions do not rely upon just a few personal observations about our own lives. Such observations are often the perfect *starting point* for inquiry and hypothesizing, but they are insufficient as the basis for solid *conclusions. Thus, one’s own experiences with accupuncture or divorce would be a justifiable basis for supposing that, perhaps, maybe there is something to accupuncture or maybe there is something to people’s worries about the breakdown in marriage. But that’s as far as anecdotal evidence alone can justifiably take us.
Merely because certain events have occurred in a certain fashion in our lives doesn’t mean that they always occur in that manner, or that they must necessarily occur in that manner. Our experiences are inherently limited and narrow in scope: we only see certain things and we only perceive them from a certain perspective. Even those among us with the most diverse and interesting of lives must acknowledge serious limitations.
Of course, we have throughout this examination been assuming that the stories which constitute the anecdotal evidence have been 100% accurate — but how often is that really the case? Eyewitness accounts, often the heart of anecdotal evidence, are notoriously unreliable, as is human memory. When people report their personal experiences they are also unable to control for things like self-deception, wishful thinking, confirmation bias, and subjective validation. In other words, reliance upon anecdotal evidence and personal testimonials is an error which compounds other errors!
There are all sorts of problems which can affect the quality of testimony and memory, which is why prosecutors in criminal cases prefer to have scientific evidence and use eyewitness accounts only as support. Where eyewitness accounts are the primary evidence, reasonable doubt is much easier to obtain — by herself, a witness testifying in a murder trial is no more inherently reliable than an eyewitness to an alien abduction or levitation.
All of this is why independent corroboration is so vital in serious scientific investigations — multiple accounts and multiple lines of converging evidence lend credibility to a claim, credibility that simply cannot exist when we rely upon anecdotal evidence alone. If such evidence is insufficient as a basis for many conclusions even when it unimpeachable, how can we possibly use it when we know just how unreliable it really is?
