I need some advice. I am finding strange bivariate correlations among my vairables of interest... there are strong patterns of findings that appear meaningful, but they are in the complete opposite direction than would be expected. I've isolated the problem to a few variables from an observational coding system that is similar to what we've used in past studies. I traced the coded data back to the raw data, and it is fine-- not reverse scored or anything like that. There is also no evidence of skew or kurtosis. Any ideas?! I know this is a long shot, but I am stuck here. Thanks!