I have 2 continuous variables that are measured at 3 time points in a within-person repeated measures fashion (T1, T2, T3). The variables follow a non-normal distribution (N = 42).
I would like to obtain the correlation between the two variables at each time point. Naïvely, I have calculated Spearman correlation coefficients at each of the time points separately and found that the correlation is significant at T1 but not at T2 and T3. The research question is whether the correlations between the measurements differ between the time points.
Is this the correct approach? To me it seems a bit silly, also considering papers I found that suggest correcting for within-person effects/dependence in those correlations (e.g., with the rmcorr R-package). Any help is greatly appreciated!