There may be a difference in what I am talking about and what you are talking about if you consider legal definitions of evidence. I am talking about evidence one uses to draw conclusions. The admissibility of evidence in a court is a completely different matter.
Personal experience is a more precise term for what I am referring to. 'Anecdotal evidence' is the scientific term for a personal experience one uses as evidence.
Right, yeah I was thinking legally. Problem is that I have seen all observations and personal experience classed as anecdotal, and as such then statements like "my father is called Gunnar" become ancedotal and considered less valuable. Now as a historian, I would note that the most valuable evidence availsble to us is primary witness reports, written often, and without direct observation history would go nowhere as a discipline. Ditto Darwin. If he had ignored his personal observations in the Galapagos, and not then formulated theories based upon his own observations, no theory of evolution. And so on and so forth. Alll of modern science was developed with observations as a key criteria. So anecdotal clearly means something far more specific than "observation based" - though increasingly I see it misued that way - Sagan did it for instance - it means hearsay or unrelaible second hand testimony, the stuff of rumour. If Soapy saw the locjh Ness monster eating petunias in his living room tonight,and reported his experience, that would not be anecdoatal evidence. If I said Soapy had had that experience and i knew cos Larsen mentioned it to me while we were buying laptops, that would.
So anecdotal evidence must be defined very strictly - and as the term is taken from law, I have used that definition. Here is a modern American definition of the type I regard as nonsensical
Anecdotal Evidence: Anecdotal Evidence is information you obtain from a subjective report, an observation, or some kind of example that may or may not be reliable. In addition, anecdotal evidence is not scientifically valid or representative of a larger group or of conditions in another location.
That is so wide as to cover almost any empirical observation. So an experiment ceases to be anecdotal when? When you type it up? If a naturaluist informs me they saw a
Turdus merula in their garden this morning does that mean I should dismiss it out of hand as a mere anecdote? If so the forum birdwatching is a pretty pointless enterprise, because seeing a
Turdus merula is not an incredible claim where I come from I can assure you! WHat about written anecdotes? When Darwin discusses his brother experience at the seance at their home, and Huxley and his brothers initial enthusiasm in his letters, do I assume this is mere anecdotal evidence and ignore it? It is an anecdote, that is clear, but nope, it's not anecdotal evidence.
Anecdotal evidence properly defined in law is hearsay as far as I can see, and in it's literal sense is just a "story told about something usually based on personal experience." I ses no reason to not accept it, one simply requires more caution perhaps, but to a historian that would probably seem a bit odd. Its the most relaible evidence available in many cases - a letter by Darwin carries more weight than a newspaper article about him?
One can also believe such personal anecdotes by proxy such as the grandmother's anecdote that Roboramma described in the above post. The 911 truthers that believe the anecdotes of witnesses have more significance than the testimony of structural engineers would also be considering anecdotal evidence (or in that example, eye witness testimony).
However we have a relationship one presumes between eyewitness testimony and the structural engineers reports. While much eyewitness testimony is doubtless flawed, as observations are, I would assume the structural engineers reports simply explain what eyewitnesses experienced on the day? If there was a large disagreement between the eyewitness reports and the official theory, as in some early models of the Hindenburg disaster, we might well want to revise till we find agreement, or a clue to the oddities of the eyewitness reports?
Take the classic case, the assasination of JFK. Eyewitness reports are confusing and contradictory, and the physical evidence is clearly vital, but we might expect some degree of agreement in a correct theory?
Some people have a very hard time believing that the conclusions one draws from either their own or someone else's personal experience is subject to all sorts of errors. Take the flu shot example. Lots of people get sick after flu shots. We give flu shots during the time of the year when commonly acquired infections are the most frequent. Kids return to school. The few children passing things to an occasional neighbor, cousin, or sibling over the summer are suddenly in confined classrooms with dozens of kids. Then those kids go home and infect their parents who by October are passing the infection on all over the community.
Surely one might also argue that in some people a hypersensitive immune reaction gives minor flu like symptoms (but no flu)? I'm assuming flu is a dead vaccine, and sure the correlation of dates is not necessarily significant - for exactly the reasons you say. It would be simple to conduct a trial where the vaccine was administered at a different time of year, or patients were isolated from possible disease vectors?
The conclusion the flu shot was the cause of the illness is common but mistaken conclusion. How do we know? Using the scientific method rather than basing assumptions on uncontrolled experiences, we take a large enough group of people, blindly randomize them into two groups (or 3 if you want a no-treatment arm of the study) and give half a flu shot and the other half a shot of saline. Then you collect the data on who got sick and then compare the group who got the vaccine with the group(s) that didn't.
Lo and behold the rate of illness is the same in both groups as was found in the study I cited in a previous post.
Nice study, but the possibility of a hidden variable remains. Saline solution is just as likely to provoke an immune response as the actual vaccine - it's not the virus making people ill, but there bodies response to the process of vaccination. If you include the no treatment arm however you have a perfectly sound study, and I would fully accept the results.
Scientific method leads you to a valid conclusion. Anecdotal or personal experience in this case leads you to a false conclusion. Your personal experience was unable to control for other variables affecting the outcome while the scientific method was. But some people are unable to recognize that fact even when presented with clear incontrovertible evidence. That personal experience was so convincing they cannot overcome its influence.
Yes, but we are now back to my original point - neither of the thinsg cited in the article were actually anecdotal. In both cases what happened was simply a false correlation, as in the flu vaccine example. And this happens just as much in real science as in pseudo-science.
Imagine I perform a study of the population of sticklebacks in a river, and measure the variable pollution froma sauce factory. As pollution increases, the stickleback population drops drastically. Applying careful statistics i find the correlation is significant.
If however I fail to note the increase in the population of stickleback-in-Worcester-sauce-loving herons, I may well have made a false corellation. Sure my brain is programmed to accept this - I have noted a correlation between say snogging my girlfriends sister and being dumped, putting my hand in a fore and it burning, and getting drunk at office parties and humping off the roof and physical injury. In almost all cases correlation does indicate causality - in a few it soes not.
Many years ago I became ill three times after visiting a burger joint near my house. I had eaten there for ears on and off with no ill effects, but now i seemd to be getting sick every time. Why? Well I thought maybe the new management had let hygeine slip. It semed a bit unlikely as the staff were mainly the same though. I thought it through, and kept going. And I was ill again, a fourth time. I switched to chicken burgers, and was fine. It took weeks of patient experimentation and a couple more queasy nights before I discovered it was nothing to do with beefburgers (the false correlation), or hygeine. It was that some items were served with a blue cheese sauce, and for reasons I have still not sicovered eating blue cheese sauce makes me really ill. I tested this by buying aj jar, and waiting till my girlfriend used some in a meal - and siure enough I was ill, even though I did not know it was in there.
So my point here is the problem is nothing to do with anecdotal evidence: it's actually
false correlation, where the apparent correlation is actually explicable by another variable. The problem is in our lives and in science the overwhelming majority of correlations are significant, which renders us prone to pay attention. Correlation does not require Causality: but in most cases it implies it. When it doesn't, then we have a problem, and make mistakes.
And Soapy's post said all this much more elegantly I felt, but hey I like whittering.
cj x
)
