You know what sucks? Spending weeks stitching together a complicated data set for a validation study, including mismatched, incomplete, and longitudinal data from half a dozen sources, then get results like this (variable names omitted for confidentiality, correlation on top with p value below it):
So I got nothin’. But all may be lost. I’m investigating ways to …massage the data to see if there are relationships there that aren’t being picked up by my clumsy, ham-fisted analyses. It sounds like a fairy tale, but it’s true!