Is anecdotal evidence dangerous?

[fairybasslet]

I have to agree with Clayjar. I find scientific articles so hard to understand since I am not a scientist. But one piece of anecdotal evidence that I always offer, and then I get blasted by the people who demand scientific proof, is...nah...I won't go there.

 

[nereas]

Lynne makes a good point about when anecdotal evidence is all that is available we are forced to make a decision knowing the risks.

Where anecdotal evidence is nothing more than a testimonial, I tend to review it carefully. However, in some cases a testimonial works for me. Such as, does this harness fit well, is this camera easy to use or, I did not like the way that mask fit me or that (gear) feel apart after only a few dives.

When it come to real safety issues, I simply don’t trust anecdotal evidence.

Your views in your 2 posts supra are the traditional conservative technically orthodox, TeamCasa Dave. And the traditional support normally given for the conservative technically orthodox is that anyone's anecdotal experiences are going to be statistically insignificant compared with the industry-wide scientific common database.

However ...

Having said that, and having agreed with you, lets not forget that for technical diving, anecdotal experience is a major foundation for expanding your own technical diving to new depths and for longer durations.

In other words, you are not supposed to just jump in the water and do a dive to 100 meters for 20 mins your first time out. You are supposed to work up to it. And working up to it is anecdotal.

The science is simply not that good or precise either, probably due to the major variations among all individuals who are trying to be scuba divers.

 

[Rockhound76]

Anecdotes, exclusively 1st person anecdotes, are fine when you recognize them as such.

Interesting stories are just that: Interesting Stories.

As a scientist, I find the subjective misinterpretation of objective data much more dangerous. Unfortunately, I see those misinterpretations passed along as fact by the media every day.

The person on the street thinks, "I read it in 'Time" or "National Geographic' so it must be so".

That is much more dangerous than any "story", since most people recognize storytelling at face value, yet these same people (us) will blindly accept the diatribes and speculations of "experts" who quote facts, subjectively organized or interpreted, they claim support their mistaken conclusions.

I think Twain said(paraphrasing, anecdotally and inaccurately), "There are three kinds of lies in the world. Lies, Damn Lies and Statistics."

 

[Teamcasa]

Anecdotes, exclusively 1st person anecdotes, are fine when you recognize them as such.

Interesting stories are just that: Interesting Stories.

As a scientist, I find the subjective misinterpretation of objective data much more dangerous. Unfortunately, I see those misinterpretations passed along as fact by the media every day.

The person on the street thinks, "I read it in 'Time" or "National Geographic' so it must be so".

I agree!

Side note: A few years ago, my brother-in-law actually told me, "I read about (insert product name) effectiveness in (insert name of) a magazine, it must be true, they could not print it unless it was true!” He now sells nutraceuticals.

 

[The Chairman]

Too often science begins with an anecdote (or two) and every now and then, it ends there. Then we have people offering up anecdotal evidence as being scientific.

It's best to weigh ALL types of evidence: anecdotal and otherwise.

 

[Damselfish]

I agree with the statement that an anecdote is one simply one data point. A single data point may or may not be good/useful whether it is part pf a scientific study or not. Some studies are pretty much based on collections of anecdotal data, or at least start from ideas generated by such data.

You need to exercise common sense and healthy skepticism whatever the source of information. Anecdotal evidence is not automatically dangerous, believing whatever you hear can be. Used to be the world was flat and Nitrox was voodoo gas, too.

 

[gcbryan]

I trust anecdotal evidence when it's my own! Otherwise, I'm not so trustful. As they said during the cold war...trust then verify. I feel the same regarding scientific evidence however.

 

[nereas]

Too often science begins with an anecdote (or two) and every now and then, it ends there. Then we have people offering up anecdotal evidence as being scientific.

It's best to weigh ALL types of evidence: anecdotal and otherwise.

And this is the view presented by the modern philosopher (and genius) Robert M. Pirsig in his 2 books touching on modern philosophy vs science and technology.

Robert Pirsig - encyclopedia article about Robert Pirsig.

To give credit where credit is due, remember that it was the ancient Greeks who invented philosophy. They had lots of free time on their hands, after defeating the Persian navy at Salamis and the Persian army at Plataea.

 

[Thalassamania]

I agree with the statement that an anecdote is one simply one data point. A single data point may or may not be good/useful whether it is part pf a scientific study or not. Some studies are pretty much based on collections of anecdotal data, or at least start from ideas generated by such data.

You need to exercise common sense and healthy skepticism whatever the source of information. Anecdotal evidence is not automatically dangerous, believing whatever you hear can be. Used to be the world was flat and Nitrox was voodoo gas, too.

A collection and analysis of "anecdotal" data turns it into science. The real question is when does anecdotal data cross over into expert opinion?

 

[ClayJar]

The more points you have, the higher your confidence that they are representative of the larger population. With only one point, you have a value, but you do not have adequate data from which to project the spread of the data points or the mean of the population.

Last time I went to a pizza buffet, I had 2.1 pizzas (and two small cinnamon rolls). It is likely readily apparent that 2.1 pizzas is much greater than the mean number of pizzas eaten by those eating at the restaurant that day. If you sample even 10 diners, you will have a much more reliable mean, and you may likely find that my consumption of 2.1 pizzas is an outlier. It's all statistics.

Of course, as you collect more data, you can model the causes, not just analyze the effects. With downloaded profile data (and not just for dives with incidents), they can start correlating and then modeling, and from there, knowledge can be built.