Thursday, June 26, 2008

Statistics, storches and babies.

As I wrote statistics is difficult. And a statistical correlation does not mean anything unless well interpreted. Let me take a nice example. When I grew up in the Netherlands, the storch had become an extremely rare bird. There were not many left. The decline of the storch during a number of decades was very clear. The decline of the number of children born per capita was clear as well. I did not calculate, but looking at the graphs a strong statistic correlation could doubtlessly be proved.

But does that mean that there was any factual correlation, or causal relationship between those two declines - the number of births and the number of storches? Doubtlessly there is, would one say who believes in the storch theory of child birth. Others might look for alternative explanations. Is there an indirect relationship? An independent factor, influencing both (such as industrialization)? Or is it a coincidence?
Statistics is a useful tool. But one has to use it wisely.

No comments: