"The study could neither confirm nor deny a link between the leukemia incidents and proximity of nuclear power plants because such statistical studies, by their nature, cannot prove cause and effect." So many scientific papers end with these words. It is as if Sherlock Holmes were going out with a shattered magnifying glass and then assiduously reporting that he could not see anything.

The problem is not the intention, but the tool. A cluster is a mysterious grouping of disease, in place or time, possibly a fluke and possibly statistically "real"—that is, not due to chance alone. Statistics can determine the "real" clusters but not what the cause may be.

By definition, a statistical study sniffs at a cold trail.

This content is only available via PDF.
You do not currently have access to this content.