Good Afternoon John!
You should be able to click on "Save" in the regression dialogue box and request that SPSS save all predicted values, so that you have that value saved for every case in your sample.
In terms of what to include, there is no "right" or"wrong" answer to that question, but it is certainly true that including only "significant" predictors in your model will make your estimate more precise. NOTE: those won't necessarily make your predictions more "accurate" overall, just precise, in that there will be less chance for "random" error that might be associated with the non-significant predictors in your model.
I hope that is helpful!
I'm trying to construct a predictive/forecast equation for sales using the output of a multiple regression using SPSS. The casewise diagnostic table shows me a few cases which have predictions that are considerably off the observed value. However i am unable to calculate these predicted values using the coeffieicnets and observed values for the outlier cases.
Is there something im overlooking?
Should only the significant values be included in this equation?