« Martin Schoeller for TIME | Main | Open Mike: Nefarious Blogger Confesses »

Saturday, 12 May 2012


Anecdotes Rule.....Data Drool.

Always a good one although I know it as the reverse: the singular of data is not anecdote.

My favorite response to this is: It therefore follows, BY STRICT LOGIC, that the SINGULAR of anecdote IS data!!!

That blog post also speaks volumes about the problems of attribution in general.

As the famous inventor Thomas Edison said, "The problem with quotes on the Internet is that you can't track down who actually said them."

"Consider that when I post a photograph here on TOP, just to find out who it really should be credited to can occasionally be a half hour's worth of research."

But that posted picture is worth a thousand words

Reminds me of my favourite quote: "since it is often hard to distinguish common sense from equally common nonsense, professional advice is useful". Leslie Lamport.

"Consider that two wrongs never make a right, but that three do."

- Tony Hendra

MM: "As the famous inventor Thomas Edison said, 'The problem with quotes on the Internet is that you can't track down who actually said them.'"
I'm sure it was Tesla, Edison always gets the credit.

Anyway, if everything in the universe is data , then surely anecdotes are data.

But wait , have I told you the one about the percentage of universes where there are no anecdotes?

Dear Andrew,

Ummm, shouldn't that be “datum?"

(he said pedantically)

A line I frequently feel compelled to trot out when faced with semi-ignorant Internet chatter is, “Sorry, but experiment trumps theory.”


“Real data trumps idle musings.”

pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 

I teach science and I deal with those tricky questions (god, etc) by explaining the levels of quality in evidence and hopefully getting students to make up their own minds by questioning their sources. In declining order of value:

1. Experimental evidence. Reproduceable results. Something you can prove by demonstration.
2. Correlation. Two things appear related, but one may not be causing the other to happen. Requires experimental investigation. (eg smoking and cancer - get those beagles puffing).
3. Logic and reason. Sounds good, but it proves nothing. Philosophy sits here.
4. Witness testimony. Anecdote. Something someone tells you. The weakest kind of evidence.

I round it off with convicted rapists locked up by witness testimony set free by DNA evidence (a procedure developed and proven valid by experiment). Logic suggests juries should place more emphasis on forensic evidence than witness testimony; lawyers get round it by making the courtroom a drama.

Don't get me started on "it's just a theory".

To muddy things further..... I seem to remember that while a grad student at Columbia U., in 1957, taking coursework in experimental design and statistical analysis, that the phrase (or a very similar variant) was used. Can't prove it at this late date, but it would be a logical thing to say in a course discussing data structure, quality, validity, and related matters. This also may be a quip that has multiple independent sources, since it addresses a fundamental issue in data analysis.

I remember my maths professor once commenting "you copy from one source and it's called plaigiarism, copy from two and it's called research..."

It so happens that I just read this very quote a few a few hours ago in Guy Kawasaki's book "Enchantement" ... and in there it is attributed to Ben Goldacre.

A great quote, by the way !

Nah - it was Descartes!

The plural of anecdote is Google Search.


The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007