Centering variables, linear regression

[monsmatglad]

I am working with multiple regression with two independent variables, and interaction between them.
the expression is: y = b1x1 + b2x2 and b3x1x2
The question is: does one center both independent variables at the same time, when checking for the significance of the effect of the independent variables separately?
Or, should I center one of the IV, and then rerun regression centering the other IV?

Hope this was understandable.

Mons

 

[FactChecker]

If you are worried about "centering" the variables, you should probably include a constant term in your model. That will allow the regression algorithm to determine the best constant value.

A step-wise regression algorithm would determine which variables should be included based on the residual statistical significance. Every statistics package that I am familiar with includes such an algorithm.