Hello,
I have a general question regarding the assumptions of correlations--independence. Specifically, someone is using a Gamma correlation (nonparametric form of correlation) measure to investigate the relationship between two variables (A and B). The two variables of interest are independent (in terms of composition). The problem is that there is a third variable (C) which is composed in part of one of the two independent variables (B). C = B*X (X is a place holder for several other components of interest). She would like to partial out the effects of C (presumably through a partial gamma correlation statistic). Given that C is composed in part by B, would it be wise to attempt to partial C from the relationship of A and B?
Thank you very much,
Emanuel
__________________________
A. Emanuel Robinson
Graduate Student
School of Psychology
Georgia Institute of Technology
Phone: 404-894-7450
Email: [log in to unmask]
___________________________
"You know how you work 25 hours a day? You get up an hour earlier."
"How do you get good ideas?...You have a lot of ideas and you throw out the bad ones."
"That's just what everyone does discuss--things they know nothing about: the weather, social problems, psychology..."
|