Effect size and predective relevance
Posted: Sat Sep 15, 2018 9:57 pm
Dear all,
I have a question regarding effect size and predective relevance.
I have a model with 3 endogenous variables. Two of them have multiple predecessors and one has only one predecessor.
I managed to calculate the f2, Q2 and q2 for the two variables with multiple predecessors using the formulas:
f2=(R2 included-R2 excluded)/(1-R2 included)
Q2=1-(∑DSSED)/(∑DSSOD)
q2= (Q2 included-Q2 excluded)(1-Q2 included)
But I cannot calculate these indexes for the endogenous variable with one predecessor, since I cannot find any values of the R2 excluded and Q2 exluded in the generated report. There is only a blank space.
Is there any way I can calculate them or is it unnecessary to do it when there is only one predecessor?
Thank you very much for your help.
Boch
I have a question regarding effect size and predective relevance.
I have a model with 3 endogenous variables. Two of them have multiple predecessors and one has only one predecessor.
I managed to calculate the f2, Q2 and q2 for the two variables with multiple predecessors using the formulas:
f2=(R2 included-R2 excluded)/(1-R2 included)
Q2=1-(∑DSSED)/(∑DSSOD)
q2= (Q2 included-Q2 excluded)(1-Q2 included)
But I cannot calculate these indexes for the endogenous variable with one predecessor, since I cannot find any values of the R2 excluded and Q2 exluded in the generated report. There is only a blank space.
Is there any way I can calculate them or is it unnecessary to do it when there is only one predecessor?
Thank you very much for your help.
Boch