Correlation vs. Causation
Published on 2 Jun 2006 at 12:00 pm |
No Comments |
Trackback
Filed under The Cargo Cults of Business.
I had two interesting discussions lately into which the distinction between correlation and causation entered. In the first case a six-year-old boy remarked that a radio commentator’s argument could be likened to (he said "was the same as”) saying that every time people put up umbrellas the streets flood, so umbrellas cause flooding. I laughed out loud at his remark but was taken aback when I realized that he was making a distinction I’d been unable to impress upon adults in a discussion a few weeks earlier.
In the earlier case I remarked, "Correlation doesn’t imply causality." This statement (which I thought a commonplace) was greeted with blank stares and then denial. How can grown people with advanced college degrees and long and productive careers behind them be ignorant of such a thing? What kind of follies has it led them to over their lifetimes? The mind boggles.
Here’s my guess about how people fall into this trap. The answer is three fold:
1. The fallacy is pervasive; it is an accepted part of pop culture. Listen to the radio, read the newspaper, watch a little TV, and like the 6-year-old you’ll find it there. People become accustomed to this type of argument and then repeat the form.
2. Correlation is an excellent way to form a hypothesis about causation. While it seems absurd to believe that umbrellas cause flooding, it is not unreasonable to use it to form a hypothesis which one could then test.
3. Perhaps the most insidious of all is the question of necessity without sufficiency. Since the lack of correlation is a perfectly good argument to refute causality the very human impulse toward symmetry tempts us to believe that the opposite is also true. If I assert that umbrellas cause flooding, you may successfully refute me with a visit to a busy beach on an August afternoon with a hot sun blazing and no signs of flooding. If honest, I would at least have to retreat into "usually" or "sometimes" — qualifiers, which by their nature suggest complex or multiple causes. Unfortunately while correlation doesn’t imply causation, neither does necessity imply sufficiency.
I should also say a word about a word… "Imply" I suppose the use of "imply" to mean something like "hint at" is legitimate, but I’m using it (and usually hear it) more as something like "requires by the nature of it."
Computer Recyclers |
---|
Pre-Owned Macintosh Computers, Parts, & Service |
Clearance iMac G4s & G5s 10% off on all PPC imacs in stock during February. |