You can prove anything with statistics. I know that is not strictly behavioural economics, but it is true. Sometimes, however, attempts to prove 'anything' are a tad blatant.
As I sat through the cinema adverts before The Hobbit started last night, one advert caught my attention. It was for a men's perfume that claimed to be better than the rest (I can't remember the brand). In order to back up this claim it had a statistic along the lines of "Most men prefer our perfume!"
In small print at the bottom of the screen was displayed the following sentence:
88% of 32 men agree
Argghhh!
That is a mind-bogglingly small (and thus unreliable) sample! Leaving aside the fact that we do not know how they conducted the survey or how the selected the 32 men (for all we know they could have been employees...), a sample size of 32 is not large enough to be conclusive.
Conclusion: You can prove anything with statistics, especially if you are unscrupulous.
You can't prove anything with statistics, because not everything is true.
ReplyDelete