Nancy,
Your co-worker is correct. Suppose you have three independent variables:
(1) Gender coded 0/1 (Male Versus Female)
(2) Race coded 0/1 (White Versus NonWhite)
(3) Age (Continuous Variable)
If you include all three variables [without any interaction terms] in your linear regression analysis, then you are fitting the following equation:
Salary = b0 + b1*Gender + b2*Race + b3*Age
Assuming you've met the assumptions to run a linear regression analysis and the predictors you've included are valid, then the equation above is what you should use to predict Salary. Do not remove Gender or Race from the equation when predicting salary. If, after fitting the model, you decided that Gender and Race should not be in the model, then you would need to rerun the analysis without those variables to obtain the estimated intercept and regression coefficient for Age.
Ryan
Nancy Rusinak wrote
I've run a multiple regression in SPSS, using various independent variables
to predict income. I have included gender and race in my model as
controls. Now, I need to predict salary and I am unclear on if I include
the partial slopes for gender and race in my prediction or exclude them from
the formula. I was under the impression that putting them in the model
controlled for them and then they were excluded. A co-worker says I need to
include them in my formula. Please advise. And thank you. N.
Locked
