I thought this was one of the best descriptions of the many
statistical and scientific activities brought to bear to solve an
important medical mystery that I've read in a long time.  The
orchestration--and evaluation--of survey data, case-control
(retrospective) studies, prospective studies, mechanistic models,
laboratory studies, meta-analysis (the study of a set of studies to
look for evidence too subtle to see in a single study), etc., is very
nicely described.

Plus, it is on a topic that is of real interest:

    http://www.nytimes.com/2011/04/17/magazine/mag-17cellphones-t.html