drop i.* controls in -outreg2-

[Jen Zhen]

Dear all,

I do regressions of the form
- reg y x i.control-
to control for the 5 categories of "control"
as well as in another context
- reg y i.x#i.control-
to control for the interactions of the 5 different categories of "control" and the 4 different categories of x.

Afterwards I use -outreg2-
Now the coefficients on the different categories of "control", which Stata labels in some pattern whose logic I do not fully follow, are not of interest to me.
However, when I add to -outreg2- the option "drop(i.control)" or "drop(i.*)" that has no effect.

What do I need to specify instead to drop these coefficients?

Thank you so much and all best,
JZ

[Carlo Lazzaro]

Jen:
did you try:

drop control

?
As an aside, I realize that, in your second code, you omitted the "main" effect of -x- and -control-.
As it is quite unusual to omit those effects, you might want to rephrase your second code as follows:

reg y i.x##i.control

[Jen Zhen]

Thank you. Somehow -drop(control)- does not do the job, no clue why?

Thank you for the aside though, that's a fair point.

Best regards,
JZ

[Carlo Lazzaro]

Jen:
try the suggested without brackets.

[Jorrit Gosens]

Use the i.var in your outreg command

sysuse auto, clear
reg price weight i.foreign
outreg2 using myfile, drop(i.foreign) replace

....which is what you appear to say in post #1, I see. It drops the variables when I do above code, though. What else is wrong? What code are you using exactly?

[Enrique Hernandez]

I understand from your question that you are using the interaction terms just as controls. If your goal is to keep only certain specific variables from your model in the table produced by outreg2 it might be better to use the option keep instead of drop.

webuse nhanes2.dta
reg  bpsystol i.sex##i.race  age height weight
outreg2 using table1, excel replace  keep(age height weight)

However, it would help if you explained what you are trying to estimate and the exact results you want to display in your table. Show us your code as well.