Multiple Regression/p-values
Message posted by Jeffrey (via 155.247.29.147) on October 24, 2001 at 1:31 PM (ET)

I'm looking at a regression analysis where my response variable is hospital patient satisfaction (satis) and my independent variables are age, severity of illness and anxiety level. After using S-Plus to run a base regression analysis, I got the following coefficient regarding my p-values in my summary:

age -> .0007
severity -> .4274
anxiety -> .4902

I am using a 10% level for significance.

Additionally, after running an anova on this same data using S-Plus, I got the following p-values (Pr(F))

age -> .0000112
severity -> .0660264
anxiety -> .4902110

My dilemna is that the p-value with respect to severity is both significant at the 10% level (in the anova) and not significant in the 10% (in the regression summary). Why, or how could this variable be significant in one sense and not significant in the other?