I think I've confused myself into a corner. I have a pre- and post- measure, and the change is sometimes +ve and sometimes -ve. I want to see if there's a difference, regardless of sign change. What is the most appropriate way to deal with it? Paired t-test with low values in one column and high values in the other? Wilcoxon signed-rank test? The data is normally distributed, paired, and sphericity assumptions are not violated.
I appreciate any clarification!