Where is alpha stored after using zinb

[Zhongying Gan]

Hello,

The following is the example from the Stata manual for a zinb regression.

I would like to include alpha and its confidence interval in the table generated by outreg2. When using outreg2, only the coefficient of lnalpha and its confidence interval are reported, not alpha.
I would like to manually put alpha into the outreg2 table by using adds() option, but I cannot find where alpha is stored. I've tried `_b[alpha]', but it doesn't work.
I would really appreciate it if someone could let me know how I could make it work. Thank you.

[Leonardo Guizzetti]

I can't speak to using -outest- but alpha is obtained easily enough by exponentiating -lnalpha- (which literally means the natural log of alpha) as back-transformation.

One way to get these estimates is to use -lincom- which will conveniently exponentiate the results for you. Save the results in scalars and then you can do whatever you wish with them.

lincom _b[/lnalpha], eform
scalar alpha = r(estimate)
scalar alpha_lb = r(lb)
scalar alpha_ub = r(ub)

scalar list alpha alpha_lb alpha_ub

Result

. lincom _b[/lnalpha], eform

 ( 1)  [/]lnalpha = 0

------------------------------------------------------------------------------
       count |     exp(b)   Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   1.667029   .3028685     2.81   0.005     1.167604    2.380076
------------------------------------------------------------------------------

. scalar alpha = r(estimate)
. scalar alpha_lb = r(lb)
. scalar alpha_ub = r(ub)

. scalar list alpha alpha_lb alpha_ub
     alpha =  1.6670288
  alpha_lb =  1.1676037
  alpha_ub =  2.3800757

[Jared Greathouse]

Just for a little more explanation, if it makes sense, you can always (usually) acces stuff like this from matricies.

use https://www.stata-press.com/data/r17/fish, clear

cls


qui zinb count persons, inflate(child camper)

mat A = e(b)

mat l A

[Zhongying Gan]

Thank you Leonardo. This works perfectly.

I can't speak to using -outest- but alpha is obtained easily enough by exponentiating -lnalpha- (which literally means the natural log of alpha) as back-transformation.

One way to get these estimates is to use -lincom- which will conveniently exponentiate the results for you. Save the results in scalars and then you can do whatever you wish with them.

lincom _b[/lnalpha], eform
scalar alpha = r(estimate)
scalar alpha_lb = r(lb)
scalar alpha_ub = r(ub)

scalar list alpha alpha_lb alpha_ub

Result

. lincom _b[/lnalpha], eform

( 1) [/]lnalpha = 0

------------------------------------------------------------------------------
count | exp(b) Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
(1) | 1.667029 .3028685 2.81 0.005 1.167604 2.380076
------------------------------------------------------------------------------

. scalar alpha = r(estimate)
. scalar alpha_lb = r(lb)
. scalar alpha_ub = r(ub)

. scalar list alpha alpha_lb alpha_ub
alpha = 1.6670288
alpha_lb = 1.1676037
alpha_ub = 2.3800757