dtable - order of statistics

[Suzanna Vidmar]

I have three questions. My first question is about the order of the summary statistics that appear in the cells of my continuous variables. Here's my code:

*Reading in the low birthweight dataset available on the Stata Press website.
use https://www.stata-press.com/data/r18/lbw,clear

*Defining and attaching value labels.
label define noyes 0 No 1 Yes
label values low ht noyes

*Clearing any collections that may be in Stata's memory.
collect clear

*Creating descriptive table.
dtable age lwt i.race i.ht, by(low, nototals) ///
sample(, statistic(frequency) place(seplabels)) ///
sformat("n=%s" frequency) ///
define(rangei = min max, delimiter("-")) sformat("[%s]" rangei) ///
define(iqi = q1 q3, delimiter(",")) sformat("(%s)" iqi) ///
continuous(age, statistic(mean sd rangei)) ///
continuous(lwt, statistic(median iqi rangei)) ///
nformat(%2.0f fvpercent mean median sd iqi rangei) nformat(%3.1f sd) ///
title(Table 1: Mother's demographic and clinical characteristics) ///
titlestyles(font(Calibri, size(14) bold))

And here's the resulting table:

Table 1: Mother's demographic and clinical characteristics
-----------------------------------------------------------------------------
                                              Birthweight<2500g              
                                          No                     Yes         
                                         n=130                  n=59         
-----------------------------------------------------------------------------
Age of mother                         24 (5.6) [14-45]       22 (4.5) [14-34]
Weight at last menstrual period [85-250] 124 (113,147) [80-200] 120 (103,130)
Race                                                                         
  White                                       73 (56%)               23 (39%)
  Black                                       15 (12%)               11 (19%)
  Other                                       42 (32%)               25 (42%)
Has history of hypertension                                                  
  No                                         125 (96%)               52 (88%)
  Yes                                           5 (4%)                7 (12%)
-----------------------------------------------------------------------------

I've deliberately enclosed the ranges in brackets [] rather than parentheses () and highlighted them so they can more easily be spotted. For "age of mother", the range is the last statistic displayed in each cell while for "weight at last menstrual period" it's the first statistic displayed. What determines the order? In my continuous options, rangei appears at the end of the statistic suboption, so the order isn't determined by the order each statistic is listed here.
I've come up with a work around, but would still like to know how to control the order in which each summary statistic is displayed.


In my work around, I've placed the range in a separate row. To do this I've generated a copy of each continuous variable. My second question: Is there a more efficient way to do this?

*Creating a copy of the continuous variables included in the table. This creates
*a separte row to display the range.
foreach var of varlist age lwt {
    gen `var'2=`var'
    label var `var'2 "(Range)"
}

*Creating descriptive table.
dtable age age2 lwt lwt2 i.race i.ht, by(low, nototals) ///
sample(, statistic(frequency) place(seplabels)) ///
sformat("n=%s" frequency) ///
define(rangei = min max, delimiter("-")) sformat("(%s)" rangei) ///
define(iqi = q1 q3, delimiter(",")) sformat("(%s)" iqi) ///
continuous(age, statistic(mean sd)) ///
continuous(lwt, statistic(median iqi)) ///
continuous(age2 lwt2, statistic(rangei)) ///
nformat(%2.0f fvpercent mean median sd iqi rangei) nformat(%3.1f sd) ///
title(Table 1: Mother's demographic and clinical characteristics) ///
titlestyles(font(Calibri, size(14) bold))

Table 1: Mother's demographic and clinical characteristics
-----------------------------------------------------------
                                     Birthweight<2500g     
                                      No           Yes     
                                    n=130          n=59    
-----------------------------------------------------------
Age of mother                        24 (5.6)      22 (4.5)
(Range)                               (14-45)       (14-34)
Weight at last menstrual period 124 (113,147) 120 (103,130)
(Range)                              (85-250)      (80-200)
Race                                                       
  White                              73 (56%)      23 (39%)
  Black                              15 (12%)      11 (19%)
  Other                              42 (32%)      25 (42%)
Has history of hypertension                                
  No                                125 (96%)      52 (88%)
  Yes                                  5 (4%)       7 (12%)
-----------------------------------------------------------

My third question is about adding a blank row between each variable. I know I can do this for the factor/categorical variables using this code:

collect style row stack, spacer

Is it possible to add a blank row after each continuous variable using collect? I've been able to do it via putdocx so it's OK if not.

Kind regards,
Suzanna

[Jeff Pitblado (StataCorp)]

1. dtable builds the statistics in the order given. To see this use the following commands to see the autolevel for dimension result and the definition of the composite result created by dtable.

collect query autolevels result
collect query composite _dtable_stats

2. Yes, you can redefine dtable's composite result to put the statistics in the order you want, when dtable fails to follow your order/logic. Here is how to redefine the above composite result to get your intended result order.

collect composite define ///
        _dtable_stats = frequency ///
                mean sd ///
                median iqi ///
                rangei ///
                fvfrequency ///
                fvpercent ///
                , replace
collect preview

Here is the resulting table

. collect preview

Table 1: Mother's demographic and clinical characteristics
-----------------------------------------------------------------------------
                                              Birthweight<2500g              
                                          No                     Yes         
                                         n=130                  n=59         
-----------------------------------------------------------------------------
Age of mother                         24 (5.6) [14-45]       22 (4.5) [14-34]
Weight at last menstrual period 124 (113,147) [85-250] 120 (103,130) [80-200]
Race                                                                         
  White                                       73 (56%)               23 (39%)
  Black                                       15 (12%)               11 (19%)
  Other                                       42 (32%)               25 (42%)
Has history of hypertension                                                  
  No                                         125 (96%)               52 (88%)
  Yes                                           5 (4%)                7 (12%)
-----------------------------------------------------------------------------

3. spacer does not put blanks between continuous variables, it follows the logic of Stata's coefficient table.

[Suzanna Vidmar]

Thanks again Jeff for your clear answers to all three of my questions.
Best wishes,
Suzanna