I have another question. I have only 40 patients and 5 domains (or factors) made of 16 (questionnaire) items. In order to get correlation coefficients controlling for age, is multiple regression better or partial correlation better to investigate the relationship between the outcome and these domains or factors? I know I can get corr coeff in both ways, but I do know in multiple regression analysis, I may not have enough people. I am not sure how partial corr works in SPSS. Could you give me a piece of advice?