[Balgair]

I think GP is talking about Type I errors; where you go hunting for correlations that are p<0.05 in a data set.

https://en.wikipedia.org/wiki/Type_I_and_type_II_errors