outreg2 (ssc): estimating multiple interactions and omitting the reference category automatically

[Johannes Muller]

Dear all

I am estimating models with several interaction terms -categorical*continous and categorical*categorical- and I am presenting the results using outreg2.
For example,

logit DV i.gender##c.age i.gender##c.age_sq i.gender##c.education i.gender##education_sq i.gender##i.married i.gender##i.year
outreg2 using results.xls, replace eform    eq(auto)  nodepvar stats(coef pval) paren(pval) alpha(0.01, 0.05, 0.1) symbol(*** , ** , *) dec(3) pdec(3)

In the output file, outreg2 automatically also reports the omitted reference category, such that for example:

2006o.year#1b.gender		0.000
		(.)
2006.year#2.gender		0.085
		(0.639)

Reworking this manually is a bit of a pain, so I was wondering whether there was any way of directly omitting these (bold) reference categories?

Thank you very much in advance.

Best

[Wouter Wakker]

I'm pretty sure outreg2 doesn't have this option. You may want to use esttab or estout (from Stata Journal) which have a 'nobaselevels' option. They generally require a bit more work to get what you what but you get a lot of flexibility in return.

[Johannes Muller]

Thank you, Wouter.

[Dan Daugaard]

Hi Johannes, would the options drop or keep in outreg2 do the trick? Eg outreg2 using doc1.doc, keep(var)
I saw this advice my William Lisowski and Charlie Joyez at
https://www.statalist.org/forums/for...-using-outreg2
hope it helps, Dan

[Johannes Muller]

Dear Dan
Thank you, you are right!
I am always using drop to not have to re-write both the equation and outreg2 line, so the keep option slipped my mind, but, of course, that should do the trick, thank you!
Best
J