Trouble horizontally orienting estpost tabstat table of summary statistics

[Jack Glaser]

Hi all—I am a Stata 17 user on several operating systems. I have run into some version of this problem a few times trying to create basic tables of summary statistics comparing values across different by() groups.

I have no problem creating a table comparing means and SDs for multiple variables across different by groups that looks very nice, but can only do so in a vertically oriented way:

clear all
sysuse auto 

*Table I want transposed

estpost tabstat price mpg weight, ///
  by(foreign) stats(mean sd) listwise nototal c(stats)

esttab, main(mean) aux(sd) nostar nogaps ///
  noobs nomtitle nonumber label unstack /// 
  note("Standard deviations in parentheses")

Which produces (sorry in advance that this isn't rendering in an attractive way) :

----------------------------------------------
Domestic Foreign
----------------------------------------------
Price 6072.4 6384.7
(3097.1) (2621.9)
Mileage (mpg) 19.83 24.77
(4.743) (6.611)
Weight (lbs.) 3317.1 2315.9
(695.4) (433.0)
----------------------------------------------
Standard deviations in parentheses


I would like a transposed version of this, with the rows being the by-groups and the columns being the variables, with the same structure of mean (sd) preserved. Yet when I try the direct modification (using c(variables) instead of c(stats) in the estpost command), I get no table at all:

est clear
*My attempt at using c(variables) does not work: 
estpost tabstat price mpg weight, ///
  by(foreign) stats(mean sd) listwise nototal c(variables)

esttab, main(mean) aux(sd) nostar nogaps ///
  noobs nomtitle nonumber label unstack /// 
  note("Standard deviations in parentheses")

This produces just 3 horizontal lines with no data at all.

Thank you in advance for the help!

[Andrew Musau]

estout is from SSC (FAQ Advice #12). You need to look at how the matrices holding the results are named.

clear all
sysuse auto

*Table I want transposed

estpost tabstat price mpg weight, ///
  by(foreign) stats(mean sd) listwise nototal c(stats)

esttab, main(mean) aux(sd) nostar nogaps ///
  noobs nomtitle nonumber label unstack ///
  note("Standard deviations in parentheses")

est clear
*My attempt at using c(variables) does not work:
estpost tabstat price mpg weight, ///
  by(foreign) stats(mean sd) listwise nototal c(variables)

esttab, cells("price mpg weight") nostar nogaps ///
  noobs nomtitle nonumber label

Res.:

. esttab, cells("price mpg weight") nostar nogaps ///
>   noobs nomtitle nonumber label

-----------------------------------------------------------
                            price          mpg       weight
-----------------------------------------------------------
Domestic                                                  
mean                     6072.423     19.82692     3317.115
sd                       3097.104     4.743297     695.3637
-----------------------------------------------------------
Foreign                                                    
mean                     6384.682     24.77273     2315.909
sd                       2621.915     6.611187     433.0035
-----------------------------------------------------------

To add parentheses around the SD estimates, there are a number of ways including manipulating the results' matrices or using some of estout's options. If it is my problem, I would simply deploy the estimation command mean which will transfer these results to e().