interaction plots

[Valerie Herz]

Dear All,
I know that there have been numerous posts about interaction plots but I still have not found an answer to my problem. I have been trying to plot my regressions for two months now with no success, so maybe you can help me out.
For my research I am running the following regression analysis:

xtgls builtlastyear c.c_ACG2##c.c_existingboats c.c_rprov##c.c_existingboats c.c_mastlength##c.c_existingboats c.c_disadv##c.c_exboatsnd c.c_standalone##c.c_existingboats GDPgrowth2 leisure olympic texist i.year, panel(hetero) //
builtlastyear is my dependent variable, existingboats my moderating variable, others independet variables and finally control variables.
I now want to display each of the interactions in an interaction margins plot.

As example one plot for xtgls builtlastyear c.c_ACG2##c.c_existingboats
And now comes my difficulty: I want to set each minimum value to 5% and each maxmimum value to 95%, so I want to see the distribution between 5 to 95%. How do I do that?
I greatly appreciate any help in this.
Many thanks,
Valerie

[Dave Airey]

It seems like you are having trouble understanding continuous by continuous interactions. Have you seen this posting?

https://www.stata.com/stata-news/news32-1/spotlight/

Or this older posting?

https://stats.idre.ucla.edu/stata/fa...tion-stata-12/

[Valerie Herz]

I am not sure that is my Problem.
Again builtlastyear is my dependent variable, existingboats my moderating variable, others independet variables
Anyone able to help?
This is what I did

clear all

set more off

use "W", clear
xtset id year
xtgls builtlastyear ACG2 existingboats c.c_ACG2##c.c_existingboats
predict m1b

matrix beta_coeff=e(b)
matrix Var_cov=e(V)

scalar beta1=beta_coeff[1,1]
scalar beta3=beta_coeff[1,3]

scalar varbeta1=Var_cov[1,1]
scalar varbeta3=Var_cov[3,3]

scalar covb1b3=Var_cov[1,3]

scalar list beta1 beta3 varbeta1 varbeta3 covb1b3
drop if existingboats < 93
drop if existingboats > 10000

gen marginal_effect_builtlastyear=beta1+beta3*existing boats if _n<100000

gen stderror_me_ACG2=sqrt(varbeta1+varbeta3*(existingb oats^2)+2*covb1b3*existingboats) if _n<100000

gen ninefiveband=2*stderror_me_ACG2

gen upper_me_ACG2=marginal_effect_ACG2+ninefiveband

gen lower_me_ACG2=marginal_effect_ACG2-ninefiveband

gen zeroyline=0

graph twoway hist existingboats, width(4) start(93) freq color(gs14) yaxis(2) //////
|| line marginal_effect_ACG2 existingboats, clpattern(solid) clwidth(medium) clcolor(black) yaxis(1) ///
|| line upper_me_ACG2 existingboats, clpattern(dash) clwidth(thin) clcolor(red) ///
|| line lower_me_ACG2 existingboats, clpattern(dash) clwidth(thin) clcolor(black) ///
|| line zeroyline existingboats, clwidth(thin) clcolor(black) clpattern(solid) ///
|| , ///
xlabel(15(3)66, nogrid labsize(2)) ///
ylabel(0 4 8 12 16 20, axis(1) nogrid labsize(2)) ///
ylabel(0 4 8 12 16 20, axis(2) nogrid labsize(2)) ///
yscale(noline alt) ///
yscale(noline alt axis(2)) ///
xscale(noline) ///
legend(off) ///
xtitle("adoption" , size(2.5) ) ///
ytitle("Observation Frequency" , axis(2) size(2.5)) ///
ytitle("Marginal Effect of participation on adoption" , axis(1) size(2.5)) ///
xsca(titlegap(2)) ///
ysca(titlegap(2)) ///
ysca(titlegap(2) axis(2)) ///
title("Marginal Effect Experience with 95% Confidence Bands") ///
subtitle("Frequency Distribution of participation -- right hand scale")
#delimit cr