However you must prove both a lack of selection-bias and a suitable sample-size before that correlation is meaningful. :)
However you must prove both a lack of selection-bias and a suitable sample-size before that correlation is meaningful. :)