Dear Statalisters,
I am trying to plot a constrasts of predictive margins plot after a GMM regression in order to find out at which levels my interaction term is significant.
My model is specified the following:
where n stands for the endogenous variables and o for exogenous variables to the model. c.indep#c.fin is an interaction term of two continous variables and I now want to plot the marginal effect of fin after finding the very helpful paper of Mr. Richard Williams 'Interpreting Interaction Terms' (2015). . However, I keep receiving the error message: "factor variables may not contain noninteger values".
Therefore I would like to know if there is another, simple way to find out at which levels of indep the interaction term becomes significant then using the marginal effects graph.
I am trying to plot a constrasts of predictive margins plot after a GMM regression in order to find out at which levels my interaction term is significant.
My model is specified the following:
Code:
xtabond2 L(0/3).value indep n o industry* i.year c.indep#c.fin, /// gmmstyle(L(1/3).value indep n , lag(4 6)) /// ivstyle( o c.indep#c.fin industry*) /// ivstyle( i.year, eq(level)) /// twostep robust orthogonal quietly margins r.fin, at(indep= (0.1 (0.1) 2.5)) marginsplot, scheme(sj) ytitle(Predicted Company Value)
Therefore I would like to know if there is another, simple way to find out at which levels of indep the interaction term becomes significant then using the marginal effects graph.