Bad Stats | The Crypt

There are lies, damned lies and statistics. Statistics offer the promise of an impartial reduction of a problem into a scientifically irrefutable answer, and are thus very persuasive. But more often than not statistics are used merely to deceive; even an outright lie hidden in numbers can be convincing – it is well known that 77% of statistics are made up on the spot! By manipulating the manner in which data is collected, processed and presented almost any conclusion can be supported by seemingly legitimate research.

A case in point is the “education” surrounding the US-led war on drugs. Here’s an extract from the National Youth Anti-Drug Media Campaign’s information on marijuana:

Longitudinal research on marijuana use among young people below college age indicates those who used have lower achievement than the non-users, more acceptance of deviant behavior, more delinquent behavior and aggression, greater rebelliousness, poorer relationships with parents, and more associations with delinquent and drug-using friends.

Did you notice the use of Bad Stats in there?

nnnnn

Let me rephrase that quote: Longitudinal research among young people below college age indicates those who have lower achievement, more acceptance of deviant behaviour, more delinquent behaviour and aggression, greater rebelliousness, poorer relationships with parents, and more associations with delinquent and drug-using friends; are more likely to use marijuana.

The research shows a correlation but does not show causality. The phrasing of the statement of correlation implies causality. If you rephrase the statement you can reverse the implied cause and effect.

Bad Stats show up in other research discussed on the page. Heavy marijuana use correlates with attention and learning impairment, but the research cannot validly infer causality due to the research technique.

Bad Stats ultimately hurt everyone. In the short term they are effective at persuading most people (often to their detriment), but in the long term they hurt the credibility of those who use them. It’s a lose-lose situation. Friends don’t let friends do Bad Stats.