Using "dtable" to display standard deviation below mean.

[Seungmin Lee]

Hi,

I want to create a table of descriptive statistics, where standard deviation (SD) is displayed below mean.

I tried "dtable" command to do so, but it only displays SD right next to mean as below.

. sysuse auto, clear
(1978 automobile data)

.
. dtable price mpg rep78

----------------------------------------
                          Summary       
----------------------------------------
N                                     74
Price              6,165.257 (2,949.496)
Mileage (mpg)             21.297 (5.786)
Repair record 1978         3.406 (0.990)
----------------------------------------

But I need to display standard deviation below mean, just like "esttab" command does as below.

. eststo temp: estpost tabstat price mpg rep78, statistics(mean sd) columns(statistics)

Summary statistics: mean sd
     for variables: price mpg rep78

             |   e(mean)      e(sd)
-------------+----------------------
       price |  6165.257   2949.496
         mpg |   21.2973   5.785503
       rep78 |  3.405797   .9899323

. esttab temp, main(mean) aux(sd)

----------------------------
                      (1)   
                            
----------------------------
price              6165.3   
                 (2949.5)   

mpg                 21.30   
                  (5.786)   

rep78               3.406   
                  (0.990)   
----------------------------
N                      74   
----------------------------
mean coefficients; sd in parentheses
* p<0.05, ** p<0.01, *** p<0.001

Is there a way to do this using "dtable" command? I tried "\n" as a delimiter in composite option, but it doesn't work.
I assume it can be done using "collect", but do not know how to do this.

Thank you.

[Jeff Pitblado (StataCorp)]

Thank you for the minimal working example.

You will need to use the collect suite of commands to change how the table is arranged. Here are the steps I took to accomplish this.

sysuse auto, clear

dtable price mpg rep78

* report the layout specification
collect layout

* look into which result levels are being shown
collect query autolevels result
collect levelsof result
collect query composite
collect query composite _dtable_stats
* stop using the composite result, we want results to be stacked into a
* single column
collect style autolevels result frequency mean sd, clear

* change how the cells are arranged;
* stack the results across rows for each variable;
* use -cmdset- for a column header
collect layout (var#result) (cmdset)

* nicer column header
collect label levels cmdset 1 "Summary"

collect preview

* hide result levels in row headers
collect style header result, level(hide)

collect preview

* make the sample size last
collect style autolevels var price mpg rep78 _N, clear

collect preview

Here is the resulting table.

------------------------------
                     Summary  
------------------------------
Price                6,165.257
                   (2,949.496)
Mileage (mpg)           21.297
                       (5.786)
Repair record 1978       3.406
                       (0.990)
N                           74
------------------------------

[Seungmin Lee]

Jeff Pitblado (StataCorp) Thank you so much!