[jheriko]

anecdotal evidence is evidence, its just not of a good quality.

a lot of things rely on it for their justification. the best example i think i can give in the context of wanky hipster startups would be "agile methodologies"... despite the completely different area of concern they are also attributed properties which are inherently difficult to measure and perform experimentation with. i've seen very little evidence except for anecdotal evidence to back the claims surrounding them (although certainly, they are based on quite reasonable conclusions to reach with no evidence at all - by using logic and "common sense" to derive from better known quantities)

[avindroth]

Anecdotal evidence is evidence. But just one data point.

The bias most commit is thinking that a data point is more valuable because you know more about it.

Just because you know the story behind it doesn't mean it's more important.

[jheriko]

i'd say its often too wooly to even call a single data point, but in other cases its obviously very many.

if 1000s of people are telling the same story, thats 1000s of potential data points... they lose value from being ill defined rather than being small in number.

[flukus]

> if 1000s of people are telling the same story, thats 1000s of potential data points... they lose value from being ill defined rather than being small in number. reply

They don't exist in isolation though. If thousands are telling the same story and there is no independent corroboration then it's more likely that one person told a story and thousands repeated it.

[jheriko]

that is a good insight. thanks.