I know we are supposed to convert our r-values to z-values (and standard error) in order to calculate confidence intervals. I am a bit confused why we must do this. The sampling distribution for Pearson's r is a t-distribution and we use a t-crit in confidence intervals when our N is small (for single data sets, not correlations), so then why can't we just use the r-values with the t-crit to create confidence intervals? Or can we?

I have tried to find the answer to this in my textbooks and online, but nothing that I have found seems to address it.

Thank you for your input on this!!

Kristin