Thursday, August 05, 2010

How to use statistics in arguments

I've dealt several times with how you can make statistics state whatever you want by using mathematical tools without telling your audience what you're doing, but this morning's Breakfast displayed three great ways to avoid even having to do that.

The argument was between Brake and the Association of British Drivers regarding speed cameras. Being early in the morning you want to avoid complex arguments yet still refer to the studies done that prove your point.

Technique number 1: Use a different metric to your opponent.

The representative from Brake kept referring to collision statistics, the rep from the ABD to casualty statistics. This can keep an argument running for ages with the trick being for one side to show there's a connection to the other's metric while the other tries to show there isn't.

Technique number 2: Slip in a qualifier and hope no-one notices.

The rep for Brake kept discussing collision statistics as a general point and then in one sentence added "in that area". So before that was she referring to a decline in collisions as a whole, or a decline in the area in which a speed camera has been placed?

Technique number 3: If all else fails appeal to common sense.

Again used by the Brake representative. No sorry you can't keep quoting studies and statistics and then fall back on "common sense" if we could just do that we wouldn't need said studies.

I would also like to applaud the rep from ABD for his attempt to explain that there was a decline in casualties prior to the introduction of cameras and that the decline has decreased since that point. This had the presenters scratching their heads for a second as they tried to parse "the decline has decreased" into something they thought their audience would understand.

On the matter at hand. From the studies I see that speed cameras can reduce both collisions and casualties, but that this reduction seems dependent on many other factors some of which if changed could have a greater impact than the simple erection of a camera. In other words it's not as simple as Speed Cameras = Good/Bad

0 comments: