How Not To Get Fooled (even by yourself)


To demonstrate that point, here's another set of clusters; again, there is a correlation between some items and clusters of disease. But I cheated - this is the same data as before - this time the stars represent areas where the occurrence of the disease is substantially less than the average - since the pattern of connection to the squares looks essentially the same, that's evidence that there is really no connection.

 Previous Slide Next Slide