Residuals as Independent Variables

[gkavcak]

Hi,

I am working on a regression model to estimate sales prices by using 5 independent variables.

Before I start to formulize my regression model, I was advised to use "clean" residuals in my main model.

I suppose using residuals in the main model is a method to avoid adding to many additional variables (i.e. lagging variables etc.) in the main model.

By the way, for your information, "clean" means time series without trend, seasonality; normally distributed and stationary variables.

Briefly, I firstly cleaned all my independent variables’ residuals. Then I tried to create my regression model by using these clean residuals.

However, even I do know that there are strong relations between the dependent and independent variables; the F score of the regression model is insignificant, correlation ratios became low and some T scores (unexpectedly) insignificant.

My question is firstly, is the method I use (using clean residuals) is an excepted one in common? Moreover, if it is used, then why the correlations became weaker and the main model's F and independent variables' T scores are insignificant?

Thanks in advance for your help.