using the @expand-function in the rhs of an equation can give rise to an implausible labelling of the estimated paramters, see for example the programme below.
With the programme provided below, the estimated parameters for the dummy-interaction terms from equation
ls y c x x*@expand(level,color,@dropfirst)
are labeled as
X*(X*(LEVEL="high"),COLOR="red")
X*(X*(LEVEL="low"),COLOR="blue")
X*(X*(LEVEL="low"),COLOR="red")
However, this should rather be like
X*(LEVEL="high",COLOR="red")
X*(LEVEL="low",COLOR="blue")
X*(LEVEL="low",COLOR="red")
The estimated parameters are correct, its just the labeling which may give rise to confusion.
Best,
mamo
Code: Select all
wfcreate u 100
alpha level = @recode(nrnd>=0, "high", "low")
alpha color= @recode(nrnd>=0, "blue", "red")
series x=nrnd+4
series y=10+1.2*x-0.6*(level="high" and color="red")*x+0.6*(level="low" and color="blue")*x+0.6*(level="low" and color="red")*x+nrnd*0.8
equation eq.ls y c x x*@expand(level,color,@dropfirst)
freeze(mode=overwrite, regresults) eq.results
show regresults