Problem with storing regression estimates while using "scfcombo" package for Survey of Consumer Finance data

[Caleb Hill]

All,

I am trying to neatly store and organize regression results in Stata 16.0 version. However, I am using Survey of Consumer Finance data and thus have been using the "scfcombo" package to run my regression commands.

Code:

*install scfcombo macro in Stata
ssc install scfcombo

*Regression Analysis (scfcombo)

scfcombo TOTAL_FAM_INC NON_AC_BUS [aw=WGT0], command(regress) reps(200) imps(5)
estimates store m1, title(TOTAL_FAM_INC)

So when I try to store the regression results using the "estimates" command, I receive an error saying: "last estimation results not found, nothing to store. " Is it possible to store "scfcombo" regression command results in Stata version 16.0?

Thanks in advance!

[Moritz Scheidenberger]

The most important results such as coefficient, standard error, pvalue, etc. are stored in r(table).
You can see see how things are stored / how the stored results look by running the following command right after having executed scfcombo:

 
 matrix list r(table)

Therefore, 'estimates store' does not work but you can write some code that reads in r(table).

[Moritz Scheidenberger]

Moreover, it turns out that using eststo is able to store the results with the correct standard errors:

eststo: scfcombo y x

Some features don't work correctly such as the number of observations but this can be fixed using local macros.
To ge the number of observations, run the regression without scfcombo and add N to your scfcombo results previously stored under the name "Probit1"

scalar Nobservations = e(N)
estadd scalar Nobs = Nobservations: probit1

.

Then, use the stats() option of esttab to include Nobs.