03-26-2015 11:43 AM
I am using "proc corr' and "partial" statement to calculation partial correlations among 50 variables, controlling for variable X (my observation N=1300). "Partial" statement automatically uses "nomiss" option, which uses list-wise deletion for missing observations (meaning a complete case analysis and I only have N=30). However, I want to do pair-wise deletion to calculate correlations. I appreciate any suggestion to use list-wise deletion in the partial correlation analysis.