An Information Theory

by on January 31, 2009

Quoting a single statistic to support an argument is rarely very impressive, regardless whether the numbers themselves are right or wrong. I would say that most  statistics are nothing without context. Context is the air that statistics breathe and the engine which powers them to make a point.  Yet far too many people simply pluck them off a tree and offer them up as withered, emasculated and pale.

Here’s an example: the famous statement, “Half the world has never made a phone call.” The effect of this adage was analysed by Clay Shirky in 2002, and it’s a prime example of a number rendered powerless by a lack of context.

Now compare the statement to this one – both made up for the sake of argument:

“Half the world had never made a phone call in 1990, but two thirds had in 2001.”

or this:

“Half the world had never made a phone call in 1990. Two thirds had in 2001, and in 2008 70% of Africa had mobile network coverage compared to 12% five years earlier.”

The value of that original statistic is now plainly flourishing with meaning. Note also how adding just two extra figures (in this case about rate of change) makes the first number fairly erupt into context. Subsequent numbers exhibit diminishing returns though.

So, here’s a theory: the value of a statistic is equal to the inverse square of the value of additional supporting metrics.  Hooray for cod science!

Leave a Reply

Your email address will not be published.