I got a negative correlation (Pearson correlation) between two variables but a positive beta coefficient of one variable predicting the other in multiple hierarchical linear regression. Is it acceptable? Please justify it with research evidence and literature
1
-
1$\begingroup$ In addition to the worthy answer by @PeterFlom, you might find my answer to this question illuminating: stats.stackexchange.com/questions/89547/… $\endgroup$ – Alexis Mar 26 at 17:08
-
1$\begingroup$ Also: Welcome to CV! $\endgroup$ – Alexis Mar 26 at 17:08
5
$\begingroup$
$\endgroup$
There is no reason that this is not possible. It follows from the nature of multiple regression, which controls for variables, which correlation does not do. See any decent book on multiple regression.