Thursday, March 6, 2008

John Godfrey Saxe neatly summarizes scientific inquiry

I've always thought that the John Godfrey Saxe poem Six Blind Men & the Elephant was a great metaphor for scientific inquiry.

You are probably familiar with the poem, but to summarize it, six blind men go to visit an elephant and each explores a different part of the elephant with his hands and comes to a different conclusion about what the elephant looks like. The first man feels the elephant's side and concludes that the elephant is like a wall. The second feels a tusk and concludes that the elephant is like a spear. The third feels the trunk and declares than the elephant is like a snake. The fourth feels the elephant's knee and declares the creature to be tree-like. The fifth touches the elephant's ear and says it is like a fan. And the sixth feels the tail and says it is like a rope.

Of course, they are all partially right and yet also wrong. Ideally, the next step would be for all six men to discuss their findings and to thereby collectively come to a broader understanding of the elephant. That obviously doesn't happen in the poem and it doesn't always happen in science, either. I think it also doesn't help that media accounts of science tend to focus on single studies rather than on the body of knowledge in a given field (news report: Elephants are like spears!).

I guess what I am really trying to say here is that while every study is limited in its scope, those limitations don't necessarily make a study "wrong." One particular example I encounter a lot in reading blogs is that people frequently critique epidemiologic studies (studies looking at behavioral and disease trends in large populations) by pointing out that correlation and causation are not the same thing. That is a fair enough point, but that doesn't mean that correlation is not a useful piece of information. It's often a great jumping-off point for further studies that may establish causation.

So, basically, it's a good thing to be aware of any study's limitations. But a limitation doesn't necessarily mean that the findings are invalid.

No comments: