Omitted category of absorbed variable in areg

[Marta Lachowska]

Dear Statalist users,

When estimating "areg y x, absorb(cvar)" is there a way to determine what the omitted category cvar is?

For example,

use https://www.stata-press.com/data/r18/auto2, clear
areg mpg weight gear_ratio, absorb(rep78)

I'd like to know which category of i.rep78 is dropped. Is it the first, the last, the most common? I cannot tell from the documentation.

Thank you in advance,
Marta

[FernandoRios]

Hi Marta
The answer is Neither category is dropped.
One way to think about it is that all fixed effects are estimated imposing the restriction that the sum of all fixed effects is zero.

you can also see how this is done:

sysuse auto, clear
drop if rep78==.
foreach i in mpg weight gear_ratio {
bysort rep78:egen mn`i'=mean(`i')
sum `i'
gen dm`i'=r(mean)+`i'-mn`i'
}
reg dmmpg dmweight dmgear_ratio
areg mpg weight gear_ratio, abs(rep78)

HTH

Standard errors is a bit different because you need to account for the not estimated fixed effects. But point differences here are identical.

[Marta Lachowska]

Thank you, Fernando!